Applied Statistics II Multivariable and Multivariate Techniques Test Bank [3 ed.]

Table of contents :
Chapter 1: The New Statistics
Test Bank
Multiple Choice
True/False
Essay
c6339e30-5d3d-4110-a3a9-9c16794262c4.pdf
Chapter 2: Advanced Data Screening, Outliers, and Missing Values
Test Bank
Multiple Choice
True/False
Essay
a7889e6b-4e37-4783-aded-3d4dd3694a5e.pdf
Chapter 3: Statistical Control: What Can Happen When You Add a Third Variable?
Test Bank
Multiple Choice
True/False
Essay
54289471-ce42-4bbf-a483-8b0e50e5f320.pdf
Chapter 4: Regression Analysis and Statistical Control
Test Bank
Multiple Choice
True/False
Essay
3a5e54f2-e064-4c5f-802d-9f6b1e7b647c.pdf
Chapter 5: Multiple Regression with Multiple Predictors
Test Bank
Multiple Choice
True/False
Essay
f3b4bad9-3be1-4345-b88c-5279bf8fa297.pdf
Chapter 6: Dummy Predictor Variables in Multiple Regression
Test Bank
Multiple Choice
True/False
Essay
89718b52-9b27-4f95-8be5-2321f3509b6f.pdf
Chapter 7: Moderation: Interaction in Multiple Regression
Test Bank
Multiple Choice
True/False
Essay
c2af5962-5656-4f94-8dcb-4270dc54466a.pdf
Chapter 8: Analysis of Covariance
Test Bank
Multiple Choice
True/False
Essay
1b1913d5-c1f5-4d18-880d-9ea2edc4d373.pdf
Test Bank
Multiple Choice
True/False
Essay
9c9dea3c-6095-47a0-ac16-42e7c068b494.pdf
Chapter 10: Discriminant Analysis
Test Bank
Multiple Choice
True/False
Essay
c118250d-f63b-4ec7-a88c-9a27bf533bcb.pdf
Chapter 11: Multivariate Analysis of Variance
Test Bank
Multiple Choice
True/False
Essay
58d49f2b-076f-43dd-b15a-8c1ed246c957.pdf
Chapter 12: Exploratory Factor Analysis
Test Bank
Multiple Choice
True/False
Essay
ed3402ec-b63a-465c-89d2-06d249d3b035.pdf
Chapter 13: Reliability, Validity, and Multiple-item Scales
Test Bank
Multiple Choice
True/False
Essay
8ad7dfc6-9d57-4cba-9d49-a74004bcd3fa.pdf
Chapter 14: More about Repeated Measures
Test Bank
Multiple Choice
True/False
Essay
a85fa00e-278e-442a-8f2d-168a534d3ea0.pdf
Chapter 15: Structural Equation Modeling with Amos: A Brief Introduction
Test Bank
Multiple Choice
True/False
Essay
4b7f4144-ba1a-41b5-9174-99df7cee1e5e.pdf
Chapter 16: Binary Logistic Regression
Test Bank
Multiple Choice
True/False
Essay
90c2b53a-9099-4b9e-9cbc-7b18303c8e1a.pdf
Chapter 17: Additional Statistical Techniques
Test Bank
Multiple Choice
True/False
Essay

Citation preview

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 1: The New Statistics Test Bank Multiple Choice 1. A variable that is statistically controlled in an analysis is a(n) ______. a. an independent variable b. a dependent variable c. a covariate d. a manipulated variable Ans: C Cognitive Domain: Knowledge Answer Location: 1.1: Required Background Difficulty Level: Easy 2. In past years, many investigators have failed to report ______. a. the effect size b. the sample sizes c. descriptive statistics d. obtained p values Ans: A Cognitive Domain: Knowledge Answer Location: 1.2: What is the “New Statistics”? Difficulty Level: Easy 3. A primary concern of investigators has been, and often still is, obtaining a p value of ______. a. .10 b. .05 c. .01 d. .000 Ans: B Cognitive Domain: Knowledge Answer Location: 1.2: What Is the “New Statistics”? Difficulty Level: Easy 4. A major change advocated by New Statistics is ______. a. reduced emphasis on effect size b. focus on the results of single studies c. not reporting confidence intervals d. understanding the limitations of significance tests Ans: D

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Cognitive Domain: Knowledge Answer Location: 1.2: What is the “New Statistics”? Difficulty Level: Easy 5. A primary change proposed by advocates of New Statistics involves summarizing effect sizes from a number of studies using ______. a. Bayesian statistics b. meta-analysis c. multivariate analysis d. descriptive statistics Ans: B Cognitive Domain: Knowledge Answer Location: 1.2: What Is the “New Statistics”? Difficulty Level: Easy 6. An emphasis of New Statistics is ______. a. using increasingly precise p values b. developing new methods of statistical analysis c. reducing emphasis on confidence interval interpretation d. adopting a more critical perspective when considering significance Ans: D Cognitive Domain: Knowledge Answer Location: 1.2: What Is the “New Statistics”? Difficulty Level: Easy 7. Which of the following types of information is provided to investigators by NHST? a. whether the hypothesis of the study has been proven b. the probability that the research hypothesis is true c. the probability of obtaining results about the H0 based on one sample d. the exact probability of obtaining the same results on replication Ans: C Cognitive Domain: Analysis Answer Location: 1.3: Common Misinterpretations of p Values Difficulty Level: Hard 8. When reading an article that reports the results of multiple statistical tests, readers should remember that multiple tests ______. a. provide inaccurate information about the risks of Type I errors b. provide inaccurate information about the risks of Type II errors c. enhance understanding of cause and effect relationships d. provide reliable information about the size and importance of effects Ans: A Cognitive Domain: Application Answer Location: 1.3: Common Misinterpretations of p Values Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 9. An advocate of New Statistics would object to a research report that included a statement like the ______. a. results of our study are consistent with a recent meta-analysis b. significant results should be interpreted cautiously due to the large sample size c. results of our study approached statistical significance d. results of our study provide support for our initial hypothesis Ans: C Cognitive Domain: Application Answer Location: 1.3: Common Misinterpretations of p Values Difficulty Level: Medium 10. If a statistical analysis yields p = .03, a research report can include a statement like ______. a. the results of our study are likely to be replicated b. the evidence is consistent with the hypothesis of our study c. there is a 97% chance that the null hypothesis of our study is false d. the results of our study cannot be explained by chance Ans: B Cognitive Domain: Application Answer Location: 1.3: Common Misinterpretations of p Values Difficulty Level: Medium 11. An advocate of New Statistics would support a p value reported as ______. a. p = .0000 b. p < .01 c. p < .05 d. p = .025 Ans: D Cognitive Domain: Knowledge Answer Location: 1.4: Problems with NHST Logic Difficulty Level: Easy 12. A problem with NHST logic is that it ______. a. eliminates the need to think in terms of double negatives b. encourages investigators to describe study results in terms of uncertainty c. cannot tell investigators what they want to know d. is used consistently and correctly in the majority of studies Ans: C Cognitive Domain: Comprehension Answer Location: 1.4: Problems with NHST Logic Difficulty Level: Medium 13. An important assumption underlying the use of an independent samples t test or one-way ANOVA is that ______. a. the grouping variables are normally distributed in the population

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 b. the variances of the groups on the dependent variable are equal c. all participants respondents responded to both the pretest and the posttest d. all scores in each group lie within one standard deviation of the mean Ans: B Cognitive Domain: Comprehension Answer Location: 1.5.1: Violations of Assumptions Difficulty Level: Medium 14. An important assumption underlying the use of Pearson r is that ______. a. the relationship between the two variables is linear b. the variances of the X and Y scores are equal c. all participants respondents responded to both the pretest and the posttest d. all scores in each variable lie within one standard deviation of the variable mean Ans: A Cognitive Domain: Comprehension Answer Location: 1.5.1: Violations of Assumptions Difficulty Level: Medium 15. A frequent violation of the rules for using NHST is ______. a. performing only one significance test b. selecting random samples from the population of interest c. selecting the statistical test after data collection d. selecting the criterion p value before analyzing the data Ans: C Cognitive Domain: Analysis Answer Location: 1.5.2: Violations of Rules for Use of NHST Difficulty Level: Hard 16. If scores on an attitude scale can range from 0 to 60 with most scores lying between 20 and 45, scores of 7 and 57 could be ______. a. likelihood errors b. probability errors c. extraneous variables d. outliers Ans: D Cognitive Domain: Analysis Answer Location: 1.5.2: Violations of Rules for Use of NHST Difficulty Level: Hard 17. Another term describing the process of testing the reproducibility of research results by repeating studies in different settings with different participants is ______. a. replication. b. generalization c. normalization d. refutation Ans: A

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Cognitive Domain: Knowledge Answer Location: 1.6: The Replication Crisis Difficulty Level: Easy 18. A proponent of New Statistics would be most likely to encourage ______. a. multiple statistical tests b. disregarding studies with nonsignificant results c. replacing α = .05 with α = .005 d. using p < .10 as the criterion for future studies Ans: C Cognitive Domain: Application Answer Location: 1.7.2: Replace α = .05 with α = .005 Difficulty Level: Medium 19. A proponent of New Statistics would be most likely to encourage ______. a. multiple statistical tests b. reporting effect size c. disregarding studies with nonsignificant results d. using p < .10 as the criterion for future studies Ans: B Cognitive Domain: Application Answer Location: 1.7.3: Less Emphasis on NHST Difficulty Level: Medium 20. The purpose of confidence intervals is to ______. a. combine the results of multiple statistical tests b. reduce the amount of measurement error in study variables c. overcome the negative aspects of nonsignificant results d. estimate a population value using sample data Ans: D Cognitive Domain: Application Answer Location: 1.8: Review of Confidence Intervals Difficulty Level: Medium 21. The confidence limits for a 95% confidence interval are determined using the ______. a. Z score for the top 95% of the distribution b. Z score for the bottom 5% of the distribution c. Z scores for the lower 2.5% and the upper 2.5% of the distribution d. Z scores for the lower 0.5% and the upper 0.5% of the distribution Ans: C Cognitive Domain: Application Answer Location: 1.8.1: Review: Setting Up Cls Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 22. The formula for determining the lower limit for a 95% confidence interval around the mean in a standard normal distribution is ______. a. lower limit = M + (+1.96) * SEMean b. lower limit = M + (-1.96) * SEMean c. lower limit = M + (+2.58) * SEMean d. lower limit = M + (-2.58) * SEMean Ans: B Cognitive Domain: Application Answer Location: 1.8.1: Review: Setting Up Cls Difficulty Level: Medium 23. The formula for determining the upper limit for a 95% confidence interval around the mean in a standard normal distribution is ______. a. upper limit = M + (+1.96) * SEMean b. upper limit = M + (–1.96) * SEMean c. upper limit = M + (+2.58) * SEMean d. upper limit = M + (–2.58) * SEMean Ans: A Cognitive Domain: Application Answer Location: 1.8.1: Review: Setting Up Cls Difficulty Level: Medium 24. The formula for calculating the standard error of a statistic involves the standard deviation of the group of participants in the study and the ______. a. range of scores of participants b. significance level used in the analysis c. mean of the sample of participants d. number of participants Ans: D Cognitive Domain: Knowledge Answer Location: 1.8.1: Review: Setting Up Cls Difficulty Level: Easy 25. The upper and lower ends of a confidence interval are referred to as the confidence ______. a. limits b. brackets c. estimates d. points Ans: A Cognitive Domain: Knowledge Answer Location: 1.8.1: Review: Setting Up Cls Difficulty Level: Easy 26. When interpreting confidence intervals reported in journal articles, readers should remember that as the ______.

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 a. standard deviation increases, increases, the width of a confidence interval increases b. level of confidence increases, the width of the confidence interval decreases c. sample size increases, the width of the confidence interval increases d. sample size increases, the size of the standard deviation increases Ans: A Cognitive Domain: Comprehension Answer Location: 1.8.5: Why Report CIs Instead of, or in Addition To, Significance Tests Difficulty Level: Medium 27. An advantage of reporting confidence intervals is that they ______. a. are available in most widely used statistical packages b. reinforce yes/no thinking c. may be more stable than p values d. are consistent across studies Ans: C Cognitive Domain: Comprehension Answer Location: 1.8.5: Why Report CIs Instead of, or in Addition To, Significance Tests Difficulty Level: Medium 28. If an investigator reports that the effect size for an analysis was estimated using Cramer’s V, the statistical analysis was ______. a. an independent samples t test b. a one-way ANOVA c. Pearson correlation d. Χ2 Ans: D Cognitive Domain: Application Answer Location: 1.9: Effect Size Difficulty Level: Medium 29. The results of which of the following statistical analyses directly provide effect size estimates? a. an independent samples t test b. a one-way ANOVA c. Pearson correlation d. Χ2 Ans: C Cognitive Domain: Application Answer Location: 1.9: Effect Size Difficulty Level: Medium 30. If an investigator reports that the effect size for an analysis was estimated using Cohen’s d, the statistical analysis was ______. a. an independent samples t test b. a one-way ANOVA c. Pearson correlation

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 d. Χ2 Ans: A Cognitive Domain: Application Answer Location: 1.9: Effect Size Difficulty Level: Medium 31. If an investigator reports that the effect size for an analysis was estimated using η or η2, the statistical analysis was ______. a. an independent samples t test b. a one-way ANOVA c. Pearson correlation d. Χ2 Ans: B Cognitive Domain: Application Answer Location: 1.9: Effect Size Difficulty Level: Medium 32. A characteristic of effect sizes is that they ______. a. depend on sample size b. have a potentially infinite range of values c. emphasize the value of yes/no thinking d. may be presented standardized units Ans: D Cognitive Domain: Knowledge Answer Location: 1.9.1: Generalizations About Effect Sizes Difficulty Level: Easy 33. Judgments about the practical importance of research results should be based on the ______. a. reported p value b. clarity of the study hypothesis c. effect size d. sample size Ans: C Cognitive Domain: Comprehension Answer Location: 1.9.1: Generalizations About Effect Sizes Difficulty Level: Medium 34. Test statistics are a function of the effect size in a study in combination with the ______. a. sample size b. difference between the groups means c. sum of the two group means d. pooled standard deviation Ans: A Cognitive Domain: Knowledge

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Answer Location: 1.9.2: Test Statistics Depend on Effect Size Combined with Sample Size Difficulty Level: Easy 35. If a practitioner in a certain area believes that even the relatively small change in client behavior noted in a study will lead other practitioners to change their approach to clients, the practitioner is referring to the study’s ______. a. statistical significance b. empirical significance c. design significance d. practical significance Ans: D Cognitive Domain: Comprehension Answer Location: 1.9.4: Use of Effect Size to Evaluate Practical or Clinical Importance (or Significance) Difficulty Level: Medium 36. A research team has received reports that the average test scores increased by 1015% on average in several classes for students who followed the study recommendation and began to take notes by hand instead of using computers. This report suggests that the results had ______. a. ethical significance b. practical significance c. internal significance d. basic significance Ans: B Cognitive Domain: Comprehension Answer Location: 1.9.4: Use of Effect Size to Evaluate Practical or Clinical Importance (or Significance) Difficulty Level: Medium 37. An investigator who wants to maximize the probability that the general public will understand the practical significance of research results should report the ______. a. effect size b. group means c. group standard deviations d. exact p value Ans: B Cognitive Domain: Comprehension Answer Location: 1.9.4: Use of Effect Size to Evaluate Practical or Clinical Importance (or Significance) Difficulty Level: Medium 38. The first step in planning a meta-analysis is to ______. a. conduct a thorough search for past studies that address the topic of interest b. review appropriate articles and then establish rules for including or excluding articles

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. create a data file that includes information considered necessary for the analysis d. clearly define the question of interest for the analysis Ans: D Cognitive Domain: Knowledge Answer Location: 1.10.1: Information Needed for Meta-Analysis Difficulty Level: Easy 39. An important goal of meta-analysis is to ______. a. evaluate variance in effect sizes across studies b. eliminate consideration of potential moderator variables c. focus only on more recent studies d. eliminate studies that do not support the rationale for the analysis Ans: A Cognitive Domain: Knowledge Answer Location: 1.10.2: Goals of Meta-Analysis Difficulty Level: Easy 40. A frequently used graphic summary for presenting data from meta-analyses is a ______. a. box plot b. frequency distribution c. forest plot d. histogram Ans: C Cognitive Domain: Knowledge Answer Location: 1.10.3: Graphic Summaries of Meta-Analysis Difficulty Level: Easy 41. A study that includes too few participants to detect a difference that actually is present is said to be ______. a. unimportant b. unreliable c. biased d. underpowered Ans: D Cognitive Domain: Knowledge Answer Location: 1.11.1: Recommendations for Research Design and Data Analysis Difficulty Level: Easy 42. The practice of focusing on significant results when a study has been completed is called ______. a. generalizability error in publication b. confirmation bias in publication c. selection bias in publication d. falsifiability error in publication Ans: B

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Cognitive Domain: Knowledge Answer Location: 1.11.1: Recommendations for Research Design and Data Analysis Difficulty Level: Easy True/False 1. Even when applied correctly, the results of significance tests may be misinterpreted. Ans: T Cognitive Domain: Knowledge Answer Location: 1.2: What Is the “New Statistics?” Difficulty Level: Easy 2. Advocates of New Statistics encourage discarding studies that report incorrect results. Ans: F Cognitive Domain: Knowledge Answer Location: 1.2: What Is the “New Statistics?” Difficulty Level: Easy 3. New Statistics involves replacing current methods of statistical analysis. Ans: F Cognitive Domain: Knowledge Answer Location: 1.2: What Is the “New Statistics?” Difficulty Level: Easy 4. An investigator who obtains a p > .05 can accept the null hypothesis. Ans: F Cognitive Domain: Knowledge Answer Location: 1. 3: Common Misinterpretations of p Values Difficulty Level: Easy 5. A nonzero difference can be determined to be statistically significant given a sufficiently large sample. Ans: T Cognitive Domain: Knowledge Answer Location: 1. 4: Problems with NHST Logic Difficulty Level: Easy 6. The changes recommended in New Statistics have been developed in the last decade. Ans: F Cognitive Domain: Knowledge Answer Location: 1.5: Common Misuses of NHST Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 7. Computer-based statistical programs tend to underestimate the actual risk of Type I errors. Ans: T Cognitive Domain: Knowledge Answer Location: 1.7.3: Less Emphasis on NHST Difficulty Level: Easy 8. The true population mean lies within the 95% confidence interval 95% of the time. Ans: F Cognitive Domain: Comprehension Answer Location: 1.8.2: Interpretation of Confidence Intervals Difficulty Level: Medium 9. If the confidence intervals for two group means do not overlap, the difference between the means is not statistically significant. Ans: F Cognitive Domain: Comprehension Answer Location: 1.8.4: Understanding Error Bar Graphs Difficulty Level: Medium 10. Effect sizes may be calculated from information typically reported in journal articles. Ans: T Cognitive Domain: Knowledge Answer Location: 1.9.1: Generalizations About Effect Sizes Difficulty Level: Easy 11. An important aspect of determining whether the results of a replication are consistent with the original study is to determine whether the confidence intervals overlap. Ans: T Cognitive Domain: Knowledge Answer Location: 1.11.1: Recommendations for Research Design and Data Analysis Difficulty Level: Easy 12 A potential problem with emphasizing replication is that the publication process will be longer. Ans: T Cognitive Domain: Knowledge Answer Location: 1.11.2: Recommendations for Authors Difficulty Level: Easy Essay

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 1. Explain why small p values may not provide an accurate indication of the strength or importance of an observed effect. Ans: Small p values may be obtained for small effect using large samples. Cognitive Domain: Comprehension Answer Location: 1.3: Common Misinterpretations of p Values Difficulty Level: Medium 2. Even if a statistical analysis yields p = .0000, why is it inappropriate to say that the alternate hypothesis has been proved? Ans: The results may not be replicated in future studies using different samples in different environments, or the results may have been Type I errors originally. Cognitive Domain: Comprehension Answer Location: 1.4: Problems with NHST Logic Difficulty Level: Medium 3. What are two reasons that failure to replicate the results of an earlier study does not mean that the results of the original study were incorrect? Ans: The replication may be flawed or the results may depend on the context in which the studies were conducted. Cognitive Domain: Knowledge Answer Location: 1.6: The Replication Crisis Difficulty Level: Easy 4. How might investigators justify their reluctance to using α = .005 instead of α = .05 as their criterion for statistical significance to a potential funding agency? Ans: Possible answer: Small effect sizes would require very large samples to have adequate power, which might make the study too costly for studies to be funded. Cognitive Domain: Application Answer Location: 1.7.2: Replace α = .05 with α = .005 Difficulty Level: Medium 5. What are the three areas of emphasis in New Statistics? Ans: The areas are (1) confidence intervals, (2) effect size information, and (3) metaanalysis. Cognitive Domain: Knowledge Answer Location: 1.7.3: Less Emphasis on NHST Difficulty Level: Easy 6. List three advantages of reporting confidence intervals instead of, or in addition to, p values. Ans: The advantages are (1) moving away from yes/no thinking, (2) clarifying the lack of precision in estimates information, and (3) confidence intervals are more stable across studies. Cognitive Domain: Knowledge Answer Location: 1.8.5: Why Report CIs Instead of, or in Addition To, Significance Tests Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 7. List two reasons that investigators may not report effect sizes. Ans: The reasons are: (1) statistical packages like SPSS may not provide effect size information and (2) confidence intervals may be embarrassingly large, resulting in very small effect sizes. Cognitive Domain: Knowledge Answer Location: 1.9.3: Using Effect Size to Evaluate Theoretical Significance Difficulty Level: Easy 8. When writing an article to report the results of a study reporting a statistically significant difference between two methods of cognitive therapy in a news release to the general public, what are two important considerations? Ans: The criteria are (1) will people notice the difference between the methods, and (2) will people care about an effect of this size. Cognitive Domain: Application Answer Location: 1.9.4: Use of Effect Size to Evaluate Practical or Clinical Importance (or Significance) Difficulty Level: Medium 9. Provide at least four reasons that effect sizes should be included in research reports. Ans: Possible reasons: (1) labeling the strength of relationships, (2) comparing effect sizes across studies, (3) in combination with other statistics, explain clinical or practical significance, (4) perform power analyses, (5) determine sample sizes, (6) use in metaanalyses Cognitive Domain: Analysis Answer Location: 1.9.5: Uses for Effect Sizes Difficulty Level: Hard 10. What are the three goals of meta-analysis? Ans: The goals are (1) estimate the mean effect size of the studies, (2) evaluate the variance of effect sizes across studies, and (3) evaluate whether certain moderator variables are related to differences in effect sizes. Cognitive Domain: Knowledge Answer Location: 1.10.2: Goals of Meta-Analysis Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 2: Advanced Data Screening, Outliers, and Missing Values Test Bank Multiple Choice 1. Prior to beginning to analyze data collected during a study, researchers should ______. a. report confidentiality violations b. plan secondary analyses c. conduct extensive data screening d. notify involved parties about anticipated results Ans: C Cognitive Domain: Knowledge Answer Location: 2.1: Introduction Difficulty Level: Easy 2. Empty cells in an SPSS worksheet indicate ______. a. data entry errors b. missing values c. standard errors d. outliers Ans: B Cognitive Domain: Knowledge Answer Location: 2.2.2: Codes for Missing Values Difficulty Level: Easy 3. In a dataset including participant age in years and cognitive ability scores that can range from 50–145, an entry of 999 for several participants indicates a(n) ______. a. a missing value b. an outlier c. a data entry error d. a replication error Ans: A Cognitive Domain: Knowledge Answer Location: 2.2.2: Codes for Missing Values Difficulty Level: Easy 4. For which of the following items would an iresearcher who has administered an inventory designed to measure work-related stress need to use reverse scoring ? a. Work interferes with my family life. b. My responsibilities are much different than I had anticipated.

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. I do not get the recognition I believe I deserve. d. Colleagues at work contribute their fair share. Ans: D Cognitive Domain: Analysis Answer Location: 2.2.4: Use Different Variable Names to Keep Track of Modifications Difficulty Level: Hard 5. Researchers should screen datasets carefully so that the ______. a. probability of rejecting the null hypothesis is increased b. results of their analyses will not be biased c. overlap of confidence intervals will be minimized d. results will have maximum practical significance Ans: B Cognitive Domain: Comprehension Answer Location: 2.2.5: Save SPSS Syntax Difficulty Level: Medium 6. Over- or under-estimation of a statistic is ______. a. error variance b. plagiarism c. bias d. miscoding Ans: C Cognitive Domain: Knowledge Answer Location: 2.3: Sources of Bias Difficulty Level: Easy 7. When participants in a study can influence the behavior of other participants through such actions as cooperation or competition, the resulting bias may ______. a. make the values of t or F too large b. reduce the risk of Type I errors c. increase the size of the group standard deviations d. increase the practical significance of statistically significant results Ans: A Cognitive Domain: Comprehension Answer Location: 2.3: Sources of Bias Difficulty Level: Medium 8. When an analysis based on the general linear model is planned, an important assumption is that ______. a. residuals for all variables are dependent on each other b. are skewed to one tail of the distribution, but not both tails c. have equal variances for all values of the predictor variables d. have a mean of 1.00 for all values of the predictor variables Ans: C Cognitive Domain: Comprehension

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Answer Location: 2.3: Sources of Bias Difficulty Level: Medium 9. In order to reduce or remove skewness from a distribution, a researcher can consider ______. a. adjusting the expected p values b. adding a minimum of five points to each score c. ignoring residuals d. removing outliers Ans: D Cognitive Domain: Comprehension Answer Location: 2.3: Sources of Bias Difficulty Level: Medium 10. A researcher who eliminates participants in the comparison group whose scores on an important variable overlap scores in the experimental group in order to obtain a p value < .05 is practicing ______. a. HARKing b. power analysis c. p-hacking d. replication Ans: C Cognitive Domain: Comprehension Answer Location: 2.4: Screening Sample Data Difficulty Level: Medium 11. Obtained p value estimates are not seriously biased when______. a. SD ≤ 10 for each group b. 1-tailed tests are applied c. outliers are retained d. n ≥ 30 per group Ans: D Cognitive Domain: Comprehension Answer Location: 2.4.3: Data Screening for Comparison of Group Means Difficulty Level: Medium 12. The most appropriate transformation for a distribution in which the smallest score is 10 and the largest score is 1,000 is to ______. a. square each score b. use a log transformation c. use an arcsine transformation d. convert Fisher r to Z Ans: B Cognitive Domain: Application Answer Location: 2.5: Possible Remedy for Skewness: Nonlinear Data Transformations Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 13. If a dataset consisting of proportions is seriously skewed and requires transforming, the appropriate transformation is ______. a. square each score b. use a log transformation c. use an arcsine transformation d. Fisher r to Z Ans: C Cognitive Domain: Application Answer Location: 2.5: Possible Remedy for Skewness: Nonlinear Data Transformations Difficulty Level: Medium 14. If a dataset consisting of correlations is seriously skewed and requires transforming, the appropriate transformation is ______. a. square each score b. use a log transformation c. use an arcsine transformation d. Fisher r to Z Ans: D Cognitive Domain: Application Answer Location: 2.5: Possible Remedy for Skewness: Nonlinear Data Transformations Difficulty Level: Medium 15. In order to quantify the distance that an outlier lies from the cloud that contains most of the data in a bivariate distribution, a researcher can calculate the ______. a. Mahalanobis distance b. standard deviation c. interquartile range d. coefficient of nondetermination Ans: A Cognitive Domain: Knowledge Answer Location: 2.6.2: Bivariate and Multivariate Outliers Difficulty Level: Easy 16. The test used to determine the statistical significance for Mahalanobis distances is ______. a. t b. F c. R2 d. Χ2 Ans: D Cognitive Domain: Knowledge Answer Location: 2.6.2: Bivariate and Multivariate Outliers Difficulty Level: Easy 17. A researcher who has trimmed a dataset has ______.

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 a. Winsorized the dataset b. removed all extreme values from the dataset c. dealt with bivariate outliers d. recorded the square root of all outliers Ans: B Cognitive Domain: Knowledge Answer Location: 2.7.2: Handling Univariate Outliers Difficulty Level: Easy 18. If a researcher replaces a score that is one standard higher than the next highest score with the next highest score, the researcher is ______. a. truncating the dataset b. trimming the dataset c. Winsorizing the dataset d. hacking the dataset Ans: C Cognitive Domain: Knowledge Answer Location: 2.7.2: Handling Univariate Outliers Difficulty Level: Easy 19. The most direct way to assess whether a linear relationship exists between two variables is to ______. a. examine a bivariate scatterplot b. calculate the Mahalanobis distance scores c. regress X on Y d. regress X2 on Y Ans: A Cognitive Domain: Knowledge Answer Location: 2.8: Testing Linearity Assumptions Difficulty Level: Easy 20. When determining prior to conducting a regression analysis, the statistic of interest is ______. a. F b. Χ2 c. R2 d. t Ans: C Cognitive Domain: Comprehension Answer Location: 2.8: Testing Linearity Assumptions Difficulty Level: Medium 21. If adding Χ2 as predictor in addition to X results in a significant increase in R2, the trend is ______. a. linear b. quadratic

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. cubic d. polynomial Ans: B Cognitive Domain: Comprehension Answer Location: 2.8: Testing Linearity Assumptions Difficulty Level: Medium 22. If adding Χ2 and Χ3 as predictors in addition to X results in a significant increase in R2, the trend is ______. a. linear b. quadratic c. cubic d. polynomial Ans: C Cognitive Domain: Comprehension Answer Location: 2.8: Testing Linearity Assumptions Difficulty Level: Medium 23. If a researcher constructing a correlation matrix based on six variables eliminates any participant who does not have a score for all six variables, the deletion method is said to be ______. a. Winsorizing b. truncatation c. listwise d. pairwise Ans: C Cognitive Domain: Knowledge Answer Location: 2.10.1: Why Missing Values Create Problems Difficulty Level: Easy 24. If a researcher constructing a correlation matrix based on six variables includes all participants who have scores for both variables in any given pair of variables, the deletion method is said to be ______. a. Winsorizing b. truncatation c. listwise d. pairwise Ans: D Cognitive Domain: Knowledge Answer Location: 2.10.1: Why Missing Values Create Problems Difficulty Level: Easy 25. Using listwise deletion to handle missing data may not result in serious problems if the amount of missing data is ______. a. < 5% b. < 7%

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. < 9% d. 1.00 b. β > 10 c. R = 0 d. R2 < 1.00 Ans: C Cognitive Domain: Knowledge Answer Location: 5.8: Significance Test for an Overall Regression Model Difficulty Level: Easy 19. The statistic used to test the significance of individual predictors in a standard regression model is ______. a. t b. R2 c. β d. F Ans: A Cognitive Domain: Knowledge Answer Location: 5.9: Significance Tests for Individual Predictors in Multiple Regression Difficulty Level: Easy 20. Readers of research reports based on sequential regression analyses should expect to see b, β, multiple R, and F values from ______. a. the first step b. each step c. the steps with highest values of each d. the final step Ans: D Cognitive Domain: Knowledge Answer Location: 5.9: Significance Tests for Individual Predictors in Multiple Regression Difficulty Level: Easy 21. The statistic used to test the significance of each predictor variable standard regression is ______. a. t

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 b. R2inc c. β d. Finc Ans: A Cognitive Domain: Knowledge Answer Location: 5.9: Significance Tests for Individual Predictors in Multiple Regression Difficulty Level: Easy 22. In sequential and statistical regression, the contribution of each predictor is assessed by examining ______. a. t b. R2inc c. β d. Finc Ans: B Cognitive Domain: Knowledge Answer Location: 5.9: Significance Tests for Individual Predictors in Multiple Regression Difficulty Level: Easy 23. In sequential and statistical regression, the significance of the change in R2 can be assessed using ______. a. t b. R2inc c. β d. Finc Ans: D Cognitive Domain: Knowledge Answer Location: 5.9: Significance Tests for Individual Predictors in Multiple Regression Difficulty Level: Easy 24. For standard regression, the most common effect size index for each predictor is ______. a. t2 b. sr2i c. β d. Finc Ans: B Cognitive Domain: Knowledge Answer Location: 5.10.2: Effect Size for Individual Predictor Variables (sr2) Difficulty Level: Easy 25. For sequential or statistical regression, the most common effect size index for each predictor may be named ______. a. t2 or R2inc b. sr2i or Finc c. β or sr2i

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 d. sr2i or R2inc Ans: D Cognitive Domain: Knowledge Answer Location: 5.10.2: Effect Size for Individual Predictor Variables (sr2) Difficulty Level: Easy 26. Reporting effect sizes when regression analyses involve very large Ns is important because ______. a. the effects of nonnormal distribution shapes are taken into account b. the confidence intervals around slope coefficients may be unrealistically large c. effects that lack practical importance may be statistically significant d. detection of weak effects sizes becomes more difficult with larger Ns Ans: C Cognitive Domain: Comprehension Answer Location: 5.12: Statistical Power Difficulty Level: Medium 27. The inclusion of another predictor variable to a regression equation would be most justifiable if the tolerance was ______. a. .25 b. .45 c. .65 d. .85 Ans: D Cognitive Domain: Knowledge Answer Location: 5.13: Nature of the Relationship Between Each X Predictor and Y (Controlling for Other Predictors) Difficulty Level: Knowledge 28. A researcher examining a graph plotting residual scores versus Y’ scores to detect multivariate outliers would look for standardized residual z scores ______. a. > + or –1, but < + or –2 b. > + or –2 c. > + or –2, but < + or –3 d. > + or –3 Ans: D Cognitive Domain: Knowledge Answer Location: 5.14: Assessment of Multivariate Outliers in Regression Difficulty Level: Easy True/False 1. Simultaneous regression is preferred to hierarchical regression. Ans: T

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Cognitive Domain: Knowledge Answer Location: 5.1: Research Questions Difficulty Level: Easy 2. Predictor variables in statistical regression are treated equally. Ans: F Cognitive Domain: Knowledge Answer Location: 5.1: Research Questions Difficulty Level: Easy 3. Regression analysis is used in experimental situations. Ans: F Cognitive Domain: Knowledge Answer Location: 5.4: Issues in Planning a Study Difficulty Level: Easy 4. Correctly specified are readily detectable. Ans: F Cognitive Domain: Comprehension Answer Location: 5.4: Issues in Planning a Study Difficulty Level: Medium 5. The increment in R2 (R2inc) is equal to sr2inc. Ans: T Cognitive Domain: Knowledge Answer Location: 5.6.2: Sequential of Hierarchical (User-Determined) Method of Entry Difficulty Level: Easy 6. Use of the statistical regression is not recommended under any circumstances. Ans: T Cognitive Domain: Knowledge Answer Location: 5.7: Variance Partitioning in Standard Regression Versus Hierarchical and Statistical Regression Difficulty Level: Easy 7. If the test for the significance of an overall regression model yields a significant F, Type I errors are ruled out. Ans: F Cognitive Domain: Knowledge Answer Location: 5.8: Significance Test for an Overall Regression Model Difficulty Level: Easy 8. The same method is used to determine the significance of standard and sequential regression models. Ans: T Cognitive Domain: Knowledge

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Answer Location: 5.8: Significance Test for an Overall Regression Model Difficulty Level: Easy 9. When reporting the results of a standard regression, predictors may be listed in any order. Ans: T Cognitive Domain: Knowledge Answer Location: 5.9: Significance Tests for Individual Predictors in Multiple Regression Difficulty Level: Easy 10. As variables are added to a sequential regression equation, F can be expected to increase as well. Ans: F Cognitive Domain: Comprehension Answer Location: 5.11: Changes in F and R as Additional Predictors Are Added to a Model in Sequential or Statistical Regression Difficulty Level: Medium 11. When determining the desired N for a multiple regression analysis, the sample size should be adequate to detect strong effect sizes. Ans: F Cognitive Domain: Knowledge Answer Location: 5.12: Statistical Power Difficulty Level: Easy 12. A tolerance of 0 indicates that a variable includes no additional information not already present in the other predictors in the equation. Ans: T Cognitive Domain: Knowledge Answer Location: 5.13: Nature of the Relationship Between Each X Predictor and Y (Controlling for Other Predictors) Difficulty Level: Easy Essay 1. What are two possible goals of regression analysis with more than two predictor variables? Ans: (1) Determine whether entire set of predictor variables predicts Y’ and (2) identify predictors that are strongest predictors of Y’. Cognitive Domain: Knowledge Answer Location: 5.1: Research Questions Difficulty Level: Easy 2. What are three components of a correctly specified regression model?

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Ans: (1) All relevant causal variables believed to predict outcome variable, (2) measures to adjust for potential sources of bias, and (3) any necessary moderator variables have been included in the model. Cognitive Domain: Knowledge Answer Location: 5.4: Issues in Planning a Study Difficulty Level: Easy 3. What are three ways in which the predictive usefulness of any individual predictor is dependent on the context of a study? Ans: (1) It may be unique to a particular sample of participants, (2) it may be limited to types of participants enrolled in study, and (3) it may vary as function of other predictor variables included in, or excluded from, analysis. Cognitive Domain: Knowledge Answer Location: 5.4: Issues in Planning a Study Difficulty Level: Easy 4. Explain the difference between forward regression and backward regression. Ans: Forward regression begins with no predictors in the equation, and predictors are added one at a time based on which predictor yields the largest increase in R2. Backward regression begins with all predictors in the equation, and predictors are removed from the equation one at a time based on which predictor yields the smallest reduction in R2. Cognitive Domain: Comprehension Answer Location: 5.6.3: Statistical (Data-Driven) Order of Entry Difficulty Level: Medium 5. What are two possible justifications for entering a particular predictor variable in an early step in a regression analysis? Ans: (1) Strong theoretical justifications and (2) in situations in which variables have been measured at different times Cognitive Domain: Comprehension Answer Location: 5.7: Variance Partitioning in Standard Regression Versus Hierarchical and Statistical Regression Difficulty Level: Medium 6. How does partitioning of variance differ between standard regression and sequential or statistical regression? Ans: In standard regression, no predictor gets credit for variance that can be explained by other predictors. In sequential or statistical regression, contribution of each predictor assessed controlling only for predictors entered in earlier steps. Cognitive Domain: Comprehension Answer Location: 5.10: Effect Size Difficulty Level: Medium 7. What are two options available to a researcher when determines that two highly correlated variables are probably measuring the same construct?

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Ans: (1) Combine them by averaging the actual or z scores, or (2) dropping one of the variables from the analysis. Cognitive Domain: Knowledge Answer Location: 5.13: Nature of the Relationship Between Each X Predictor and Y (Controlling for Other Predictors). Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 6: Dummy Predictor Variables in Multiple Regression Test Bank Multiple Choice 1. Which of the following variables would be entered in a regression equation as a dummy variable? a. hours spent studying each week b. number of conversations initiated daily c. in a campus organization or not d. number of texts received while in class Ans: C Cognitive Domain: Comprehension Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Medium 2. If the GPA of students having an on-campus job is always higher than the GPA for having an off-campus job, these variables are said to be______. a. multicollinear b. monotonic c. orthogonal d. standardized Ans: B Cognitive Domain: Comprehension Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Medium 3. If a researcher included a multiple-group categorical predictor variable including three groups in a regression equation, the concern would be the ______. a. anticipated reduction in R2 across categories from the category labeled 1 to the category labeled 3 b. influence of outliers on the category scores c. linear relationship between the category scores and the quantitative variable scores d. effect of the category scores on the measures of multicollinearity Ans: C Cognitive Domain: Comprehension Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Medium 4. The number of dummy variables necessary when a categorical variable includes k groups or categories is ______. a. k

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 b. k + 1 c. k – 1 d. k2 Ans: C Cognitive Domain: Knowledge Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Easy 5. Regression analyses that include dummy variables provides information about group means that is similar to the information provided by ______. a. biserial correlation b. Χ2 c. Pearson correlation d. ANOVA Ans: D Cognitive Domain: Knowledge Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Easy 6. When dummy variables are used as predictors in multiple regression, the b raw score slope coefficients provide information about ______. a. differences in the means of each group b. changes in Y’ for each change in group membership c. standardized changes in Y’ for each change in X d. the interaction between changes in group membership and quantitative predictors Ans: A Cognitive Domain: Knowledge Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Easy 7. A researcher who finds that a group in a dummy variable that was intended to be a predictor in a regression may choose to exclude the group from the planned analysis if the group has fewer than ______ members. a. 10 b. 20 c. 25 d. 30 Ans: B Cognitive Domain: Knowledge Answer Location: 6.4: Issues in Planning a Study Difficulty Level: Easy 8. If a researcher wants to predict college GPA using off-campus job = 0 and oncampus job = 1, the best predictor of GPA for off-campus job holders is ______. a. b0 b. b1

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. b0+b1 d. b0-b1 Ans: A Cognitive Domain: Application Answer Location: 6.7.1.1: Two-Group Example With a Dummy-Coded Dummy Variable Difficulty Level: Medium 9. If a researcher wants to predict college GPA using off-campus job = 0 and oncampus job = 1 as predictors, the best predictor of GPA for on-campus job holders is ______. a. b0 b. b1 c. b0+b1 d. b0-b1 Ans: C Cognitive Domain: Application Answer Location: 6.7.1.1: Two-Group Example With a Dummy-Coded Dummy Variable Difficulty Level: Medium 10. If a researcher wants to predict college GPA using off-campus job = 0 and oncampus job = 1 as predictors, the difference in GPA for off-campus and on-campus job holders is ______. a. b0 b. b1 c. b0+b1 d. b0-b1 Ans: B Cognitive Domain: Application Answer Location: 6.7.1.1: Two-Group Example With a Dummy-Coded Dummy Variable Difficulty Level: Medium 11. If a researcher conducts a study using job status to predict college GPA codes students who work off-campus=0, on-campus job=1, and students who hold no job, students who hold no job would be coded by default as ______. a. 0, 0 b. 1, 0 c. 0, 1 d. 1, 1 Ans: A Cognitive Domain: Application Answer Location: 6.7.1.2: Multiple-Group Example With a Dummy-Coded Dummy Variable Difficulty Level: Medium 12. If a researcher who wants to predict college GPA based on whether a student attended a public high school or a private high school codes public high school

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 attendance as –1 and private high school attendance as +1, the researcher is using ______. a. dummy coding b. orthogonal coding c. effect coding d. linear contrasts Ans: C Cognitive Domain: Application Answer Location: 6.7.2.1: Two-Group Example With an Effect-Coded Dummy Variable Difficulty Level: Medium 13. If a researcher who wants to predict college GPA based on whether a student attended a public high school or a private high school codes public high school attendance as –1 and private high school attendance as +1, the best predicted value of GPA for students who attended public schools is ______. a. b0 b. b1 c. b0+b1 d. b0-b1 Ans: D Cognitive Domain: Application Answer Location: 6.7.2.1: Two-Group Example With an Effect-Coded Dummy Variable Difficulty Level: Medium 14. If a researcher who wants to predict college GPA based on whether a student attended a public high school or a private high school codes public high school attendance as –1 and private high school attendance as +1, the best predicted value of GPA for students who attended private schools is ______. a. b0 b. b1 c. b0 + b1 d. b0 – b1 Ans: C Cognitive Domain: Application Answer Location: 6.7.2.1: Two-Group Example With an Effect-Coded Dummy Variable Difficulty Level: Medium 15. If a researcher who wants to predict college GPA based on whether a student attended a public high school or a private high school codes public high school attendance as –1 and private high school attendance as +1, the grand mean for GPA for both groups combined is ______. a. b0 b. b1 c. b0 + b1 d. b0 – b1 Ans: A

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Cognitive Domain: Knowledge Answer Location: 6.7.2.1: Two-Group Example With an Effect-Coded Dummy Variable Difficulty Level: Easy 16. Using effect coding, a researcher conducting a study using job status to predict college GPA might code students who work off-campus as 0, students who work oncampus job as 1, and students who hold no job as ______. a. 0, 0 b. 1, 0 c. 0, 1 d. –1, –1 Ans: D Cognitive Domain: Application Answer Location: 6.7.1.2: Multiple-Group Example With an Effect-Coded Dummy Variable Difficulty Level: Medium 17. A researcher who wanted to use parenting style to predict the number of misbehaviors teachers observed in classrooms coded authoritative parenting as +1, authoritarian parenting as –1, and permissive parenting as 0, would be using a(n) ______. a. elimination contrast b. orthogonal contrast c. linear contrast d. monotonic contrast Ans: B Cognitive Domain: Application Answer Location: 6.7.3: Orthogonal Coding of Dummy Predictor Variables Difficulty Level: Medium 18. When dummy variables are included in a regression equation, the usefulness of predictor variables is assessed by examining the increase in ______. a. r b. β c. R2 d. Y’ Ans: C Cognitive Domain: Knowledge Answer Location: 6.9: Effect Size and Statistical Power Difficulty Level: Easy True/False 1. Dummy variables are limited to variables involving two groups. Ans: F

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Cognitive Domain: Knowledge Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Easy 2. A dummy variable characterizing the type of therapy as 1 = self-group, 2 = group therapy, and 3 = individual therapy would be recommended for predicting a clinical outcome. Ans: F Cognitive Domain: Comprehension Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Medium 3. Cases included in dummy variables can belong to more than one group. Ans: F Cognitive Domain: Knowledge Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Easy 4. Any change that can be observed across two groups can only be linear. Ans: T Cognitive Domain: Knowledge Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Easy 5. When dummy variables are included as predictors in regression equations, the usual assumptions for regression change. Ans: F Cognitive Domain: Knowledge Answer Location: 6.3: Screening for Violations of Assumptions Difficulty Level: Easy 6. Inclusion of dummy predictor variables does not affect multiple regression computations. Ans: T Cognitive Domain: Knowledge Answer Location: 6.5: Parameter Estimates and Significance Tests for Regressions With Dummy Predictor Variables Difficulty Level: Easy 7. One-way ANOVA is the most familiar approach to data analysis in research situations that involve categorical variables with more than two levels and a quantitative outcome variable. Ans: T Cognitive Domain: Knowledge Answer Location: 6.6.2: College Differences in Mean Salary Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 8. When effect coding of dummy variables is used, b0 reflects the weighted grand mean. Ans: F Cognitive Domain: Knowledge Answer Location: 6.7.1.2: Multiple-Group Example With an Effect-Coded Dummy Variable Difficulty Level: Easy 9. When only dummy variables representing three categories are included in a regression equation, the analysis is essentially equivalent to one-way ANOVA. Ans: T Cognitive Domain: Knowledge Answer Location: 6.9: Effect Size and Statistical Power Difficulty Level: Easy Essay 1. A researcher wants to use job satisfaction categorized as low, average, and high as a predictor in a multiple regression equation to predict life satisfaction. Explain why entering a single variable with job satisfaction categorized as 1 = low, 2 = average, and 3 = high might not be effective an effective predictor of life satisfaction. Ans: Scores on an outcome variable such as life satisfaction may not increase linearly with the scores on the job satisfaction variable. Life satisfaction scores may increase significantly as job satisfaction increases from low to medium, but may not change as job satisfaction increases from medium to high. Cognitive Domain: Application Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Medium 2. Why are only two dummy variables needed when there are three groups of interest? Ans: If two variables are coded 1 = yes and 0 = no, individuals who are not members of either group would be = 0,0 on first two variables. Cognitive Domain: Comprehension Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Medium 3. What are two reasons for including dummy variables as predictors in regression analysis? Ans: (1) Provides simple demonstration of fundamental equivalence between ANOVA and multiple regression, and (2) permits researchers to include group membership variables with other predictors in multiple regression. Cognitive Domain: Knowledge Answer Location: 6.1: What Dummy Variables Are and When They Are Used Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 4. Why is it important that groups represented in dummy variables include at least 10– 20 scores? Ans: Estimates of means for smaller groups may have wide confidence intervals. Cognitive Domain: Knowledge Answer Location: 6.3: Screening for Violations of Assumptions Difficulty Level: Easy 5. A researcher wants to use job satisfaction categorized as low, average, and high as a predictor in a multiple regression equation to predict life satisfaction. If there are only eight participants in the low job satisfaction group, what are two options available to the researcher? Ans: (1) Combine the low and medium satisfaction groups, or (2) exclude the low satisfaction group from the analysis. Cognitive Domain: Knowledge Answer Location: 6.4: Issues in Planning a Study Difficulty Level: Easy 6. How might the interpretation of b0 and b1 coefficients differ for dummy predictor variables compared to these coefficients associated with continuous predictor variables? Ans: B0 may correspond to one of the group means or the grand mean of Y. b1 for dummy predictor variables may correspond to contrasts between group means or to differences between group means and the grand mean. Cognitive Domain: Knowledge Answer Location: 6.5: Parameter Estimates and Significance Tests for Regressions With Dummy Predictor Variables Difficulty Level: Easy 7. What is the difference between dummy coding and effect coding? Ans: Dummy coding: members of the last group receive score of 0 on all dummy variables. Effect coding: members of last group receive score of –1 on all dummy variables. Cognitive Domain: Knowledge Answer Location: 6.7: Three Methods of Coding for Dummy Variables Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 7: Moderation: Interaction in Multiple Regression Test Bank Multiple Choice 1. An educational psychologist who finds that the effect of test anxiety on test scores differs according to whether the test is administered in a large or small classroom has detected a(n) ______. a. transformation b. interaction c. estimation error d. biased effect Ans: B Cognitive Domain: Comprehension Answer Location: 7.1: Terminology Difficulty Level: Medium 2. When used in the context of multiple regression, moderation refers to ______. a. a curvilinear relationship between Y and one or more predictors b. the presence of multicollinearity among the predictors c. mediation reflecting a causal relationship between two or more predictors d. the presence of an interaction between two or more predictors Ans: D Cognitive Domain: Knowledge Answer Location: 7.1: Terminology Difficulty Level: Easy 3. A researcher using self-esteem to predict coping ability after a natural disaster who finds that the slope relating self-esteem to coping ability is different for males and females would say that sex is a ______. a. controlled variable b. mediating variable c. moderator variable d. conceptual variable Ans: C Cognitive Domain: Comprehension Answer Location: 7.1: Terminology Difficulty Level: Medium 4. An interaction between two predictors in a multiple regression equation is written as ______. a. X1 + X2

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 b. X1 – X2 c. X1 * X2 d. X1 / X2 Ans: C Cognitive Domain: Knowledge Answer Location: 7.1: Terminology Difficulty Level: Easy 5. An interaction between work-group functioning (wgf) and task-role clarity (trc) in a multiple regression equation would be written as ______. a. wgf + trc b. wgf - trc c. wgf * trc d. wgf / trc Ans: C Cognitive Domain: Comprehension Answer Location: 7.1: Terminology Difficulty Level: Medium 6. Interaction terms in multiple regression should be included when ______. a. the researcher suspects from theory that an interaction exists b. two more continuous predictors are used in order to account for more variance in Y c. a curvilinear relationship exists between a predictor and Y d. the interaction involves two or more categorical variables Ans: a Cognitive Domain: Knowledge Answer Location: 7.1: Terminology Difficulty Level: Easy 7. A researcher using self-esteem to predict coping ability after a natural disaster finds that the regression lines relating self-esteem to coping ability for men and women cross, the relationship would be labeled ______. a. controlled b. disordinal c. uncentered d. hypothetical Ans: B Cognitive Domain: Comprehension Answer Location: 7.2: Interaction Between Two Categorical Predictors: Factorial ANOVA Difficulty Level: Medium 8. The correct formula for a regression equation that includes an interaction between one quantitative and one categorical predictor variable is ______. a. Y’ = b0 + b1X1 + b2X2 +b3(X1*X2) b. Y’ = b0 + b1X1 + b2X2 +b3(X1 – X2

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. Y’ = b0 + b1X1 + b2X2 +b3(X1 * X2 d. Y’ = b0 + b1X1 + b2X2 +b3(X1 / X2 Ans: C Cognitive Domain: Knowledge Answer Location: 7.6: Regression to Assess Statistical Significance of Interaction Between One Categorical and One Quantitative Predictor Difficulty Level: Easy 9. A researcher has hypothesized that job satisfaction (Y) is directly associated with work-related stress, and past research has shown that women have higher stress scores than men. If a multiple regression supported both stress and gender as significant predictors of satisfaction, but the interaction between stress and gender was not statistically significant, which of the following statements is true? a. The slope of the regression line for women would be much larger than that for men. b. The slope of the regression line for women would be much smaller than that for men. c. The slope of the regression line for women would be the same as that for men. d. The intercept for women and men would be the same. Ans: C Cognitive Domain: Comprehension Answer Location: 7.6: Regression to Assess Statistical Significance of Interaction Between One Categorical and One Quantitative Predictor Difficulty Level: Medium 10. A researcher has hypothesized that job satisfaction (Y) is directly associated with work-related stress, and past research has shown that women have higher stress scores than men. If a multiple regression supported both stress and gender as significant predictors of satisfaction, and the interaction between stress and gender was statistically significant, which of the following statements is true? a. The slopes of the regression lines for women and men would be different, but the intercepts would be the same. b. The slopes of the regression lines for women and men would be the same, and the intercepts would be the same. c. The slopes of the regression lines for women and men would be the same, but the intercepts would be different. d. The slopes of the regression lines for women and men would be different, and the intercepts would be different. Ans: D Cognitive Domain: Comprehension Answer Location: 7.6: Regression to Assess Statistical Significance of Interaction Between One Categorical and One Quantitative Predictor Difficulty Level: Medium 11. A researcher has used stress and gender to predict job satisfaction (Y). If the researcher examines the multiple R based on the prediction of job satisfaction by stress for women and men separately, the researcher is analyzing ______. a. compound effects

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 b. collinear effects c. nonadditivity effects d. simple main effects Ans: D Cognitive Domain: Comprehension Answer Location: 7.9: Follow-Up: Analysis of Simple Main Effects Difficulty Level: Medium 12. Centering scores when analyzing interactions in multiple regression equations ______. a. reduces the correlation of the interaction term with the two predictors from which it was computed b. requires that the centered terms be used instead of the original predictor values from which they were derived c. is necessary when one or both of the predictors is quantitative d. is primarily intended to reduce the collinearity between the two predictors Ans: a Cognitive Domain: Comprehension Answer Location: 7.10: Interaction Between Two Quantitative Predictors Difficulty Level: Medium 13. Centering X1 and X2 scores in regression equations is accomplished by ______. a. adding the mean of X1 to X1 and the mean of X2 to X2 b. subtracting the mean of X1 from X1 and the mean of X2 from X2 c. adding the mean of X1 to X2 and the mean of X2 to X1 d. subtracting the mean of X1 from X2 and the mean of X2 from X1 Ans: B Cognitive Domain: Comprehension Answer Location: 7.10: Interaction Between Two Quantitative Predictors Difficulty Level: Medium True/False 1. Interactions not predicted before a regression analysis should not be added after the initial analysis. Ans: F Cognitive Domain: Knowledge Answer Location: 7.1: Terminology Difficulty Level: Easy 2. Moderation and mediation are synonymous terms. Ans: F Cognitive Domain: Knowledge Answer Location: 7.1: Terminology Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 3. Regression results from nonexperimental data can be interpreted in causal terms. Ans: F Cognitive Domain: Knowledge Answer Location: 7.1: Terminology Difficulty Level: Easy 4. An interaction effect in regression requires that the interacting variables be correlated. Ans: F Cognitive Domain: Knowledge Answer Location: 7.1: Terminology Difficulty Level: Easy 5. When the regression lines for the categories of a categorical predictor variable are parallel, an interaction is present. Ans: F Cognitive Domain: Knowledge Answer Location: 7.5: Scatterplot for Preliminary Assessment of Possible Interaction Between Categorical and Quantitative Predictor Difficulty Level: Easy 6. Correlations between predictors in regression may not indicate the presence of moderation. Ans: T Cognitive Domain: Knowledge Answer Location: 7.1: Terminology Difficulty Level: Easy 7. The assumptions for moderation analysis differ from the assumptions of other regression analyses. Ans: F Cognitive Domain: Knowledge Answer Location: 7.1: Terminology Difficulty Level: Easy 8. If one predictor variable is quantitative and the other is categorical, the quantitative variable is considered to be the moderator. Ans: F Cognitive Domain: Knowledge Answer Location: 7.1: Terminology Difficulty Level: Easy Essay

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 1. How would one explain the interaction ratings of supervisory behavior and personal locus of control (internal vs. external) in a regression equation predicting overall job satisfaction? Ans: The slope to predict overall job satisfaction from ratings of supervisory behavior differs across scores of locus of control. Cognitive Domain: Application Answer Location: 7.1 Terminology Difficulty Level: Medium 2. What is an additional step that should be taken during preliminary screening when both a quantitative and a categorical variable are included in a regression equation? Ans: Create a scatterplot for the quantitative variable and Y for each category of the categorical variable. Cognitive Domain: Knowledge Answer Location: 7.4: Preliminary Data Screening: One Categorical and One Quantitative Predictor Difficulty Level: Easy 3. Why is centering important when planning a regression analysis that will include an interaction between two quantitative predictors and how is it accomplished? Ans: Centering is important because it reduces the multicollinearity between the two predictors and the interaction term itself. It is accomplished by subtracting the mean of each variables from each score on the variable. Cognitive Domain: Application Answer Location: 7.10: Interaction Between Two Quantitative Predictors Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 8: Analysis of Covariance Test Bank Multiple Choice 1. The primary goal of ANCOVA is detect differences between or among experimental groups while ______. a. holding the effects of specific treatments constant b. assessing differences in a specific characteristic after an experiment has been conducted c. controlling for individual differences among participants d. assessing differences in certain participant characteristics Ans: C Cognitive Domain: Comprehension Answer Location: 8.1: Research Situations for Analysis of Covariance Difficulty Level: Medium 2. A problem encountered when using quasi-experimental designs is ______. a. the absence of correlations between certain variables b. limits on the number of variables that can be manipulated c. use of pre- and posttests d. the inability to assign participants to treatment conditions Ans: D Cognitive Domain: Comprehension Answer Location: 8.1: Research Situations for Analysis of Covariance Difficulty Level: Medium 3. When inclusion of a covariate results in a substantially smaller SSresidual term, the covariate can be described as a(n) ______. a. true variance suppressor b. noise suppressor c. effect size suppressor d. nonequivalence suppressor Ans: B Cognitive Domain: Knowledge Answer Location: 8.1: Research Situations for Analysis of Covariance Difficulty Level: Easy 4. Means for experimental groups from which the effects of a covariate have been statistically removed are said to be ______. a. nonequivalent means b. controlled means

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. restricted means d. adjusted means Ans: D Cognitive Domain: Knowledge Answer Location: 8.2: Empirical Example Difficulty Level: Easy 5. In order to ensure that covariates are not influenced by the experimental treatment, they should be ______. a. unrelated to the dependent variable b. unrelated to the independent variable c. measured before the treatment d. measured after the treatment Ans: C Cognitive Domain: Knowledge Answer Location: 8.3: Screening for Violations of Assumptions Difficulty Level: Easy 6. A valid covariate in ANCOVA would be ______. a. attitude toward teachers measured before a new educational program is implemented b. reading ability measured after a program to improve scores in social science courses c. trait anxiety measured following treatment to address a specific phobia d. manual dexterity scores obtained before a study of the effects of meditation Ans: A Cognitive Domain: Analysis Answer Location: 8.3: Screening for Violations of Assumptions Difficulty Level: Medium 7. Correlation between a covariate and the treatment variable should be ______. a. essentially .00 b. weak to moderate c. moderate d. moderate to strong Ans: B Cognitive Domain: Knowledge Answer Location: 8.3: Screening for Violations of Assumptions Difficulty Level: Easy 8. The assumption that there is no interaction between a treatment variable and a covariate is referred to as ______. a. homogeneity of effect sizes b. the absence of multicollinearity c. heterogeneity of variance d. homogeneity of regression Ans: D Cognitive Domain: Knowledge

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Answer Location: 8.3: Screening for Violations of Assumptions Difficulty Level: Easy 9. When used in discussions of ANCOVA, the term homogeneity of regression indicates that the ______. a. variance of treatment variable is approximately equal across all levels of the treatment b. distributions of the treatment variable and the covariate are approximately normal c. slope of the line predicting the treatment variable from the covariate is the same within each treatment group d. covariate must have a linear relationship to the treatment variable Ans: C Cognitive Domain: Comprehension Answer Location: 8.3: Screening for Violations of Assumptions Difficulty Level: Medium 10. In a typical ANCOVA analysis, variance is partitioned by ______. a. entering the treatment variable and covariate simultaneously b. first entering the covariate and then the treatment variable c. first entering the treatment variable and then the covariate d. entering the treatment variable without the covariate Ans: B Cognitive Domain: Knowledge Answer Location: 8.4: Variance Partitioning in ANCOVA Difficulty Level: Easy 11. Even in well-designed experiments, inclusion of covariates is useful as a way to ______. a. increase SSresidual b. reduce SSbetween c. increase SSwithin d. reduce SSwithin Ans: D Cognitive Domain: Knowledge Answer Location: 8.5: Issues in Planning a Study Difficulty Level: Easy 12. The results of ANCOVA are potentially misleading unless the covariate is ______. a. reliable and valid b. uncorrelated with the treatment variable c. strongly correlated with the treatment variable d. reliable, but not necessarily valid Ans: A Cognitive Domain: Knowledge Answer Location: 8.5: Issues in Planning a Study Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 13. Researchers should be skeptical of the adjusted means resulting from ANCOVA when ______. a. the rank order of the treatment group means is substantially different b. the main effect of the treatment increases c. the main effect of the treatment decreases d. if the analysis yields nonsignificant differences between the treatment group means Ans: A Cognitive Domain: Knowledge Answer Location: 8.8: Conceptual Basis: Factors That Affect the Magnitude of SSAadj and SSresidual and the Pattern of Adjusted Group Means Difficulty Level: Easy 14. Effect sizes in ANCOVA can be indexed by ______. a. β b. adjusted R2 c. XC2 d. partial η2 Ans: D Cognitive Domain: Knowledge Answer Location: 8.9: Effect Size Difficulty Level: Easy 15. When ANCOVA has increased power compared to ANOVA, the reason is that ______. a. SStotal is higher for ANCOVA b. SSresidual is higher for ANCOVA c. SSresidual is lower for ANCOVA d. SSAadj is lower for ANCOVA Ans: C Cognitive Domain: Knowledge Answer Location: 8.10: Statistical Power Difficulty Level: Easy True/False 1. Traditional ANCOVA is useful when an interaction between the treatment and a covariate is present. Ans: F Cognitive Domain: Knowledge Answer Location: 8.1: Research Situations for Analysis of Covariance Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 2. ANCOVA is equally effective as random assignment in controlling the composition of groups. Ans: F Cognitive Domain: Knowledge Answer Location: 8.1: Research Situations for Analysis of Covariance Difficulty Level: Easy 3. ANCOVA can be used to adjust for nonequivalent groups when the data were collected in a true experiment. Ans: T Cognitive Domain: Knowledge Answer Location: 8.1: Research Situations for Analysis of Covariance Difficulty Level: Easy 4. If the treatment-by-covariate interaction is not statistically significant, and important assumption of ANCOVA has been violated. Ans: F Cognitive Domain: Knowledge Answer Location: 8.3: Screening for Violations of Assumptions Difficulty Level: Easy 5. ANCOVA models can include two or more covariates. Ans: T Cognitive Domain: Knowledge Answer Location: 8.4: Variance Partitioning in ANCOVA Difficulty Level: Easy 6. Statistical control of confounds using ANCOVA is as effective as random assignment of participants to groups. Ans: F Cognitive Domain: Knowledge Answer Location: 8.5: Issues in Planning a Study Difficulty Level: Easy 7. Partial η2 can be expected to larger than η2. Ans: F Cognitive Domain: Knowledge Answer Location: 8.9: Effect Size Difficulty Level: Easy 8. Adding covariates may reduce the power of ANCOVA. Ans: T Cognitive Domain: Knowledge Answer Location: 8.10: Statistical Power Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Essay 1. What are two situations in which ANCOVA is preferred over ANOVA? Ans: (1) When the covariate is confounded with the group membership variable, and the unique effects of the grouping variable can only be assed when the effects of the covariate are controlled or partialled out, and/or (2) when the covariate is strongly predictive of the outcome variable and including the covariate yields a smaller error term and larger F ration when assessing the main effect of the treatment variable. Cognitive Domain: Comprehension Answer Location: 8.1: Research Situations for Analysis of Covariance Difficulty Level: Medium 2. How are ANCOVA and regression related? Ans: ANCOVA is a special case of regression in which one or more quantitative predictors are used as covariates and one or more categorical variables represent membership in treatment groups. Cognitive Domain: Comprehension Answer Location: 8.6: Formulas for ANCOVA Difficulty Level: Medium 3.What is the primary goal of ANCOVA? Ans: Assess whether the main treatment effects remain significant when one or more covariates are controlled. Cognitive Domain: Knowledge Answer Location: 8.6: Formulas for ANCOVA Difficulty Level: Easy 4. What are four possible changes in the means of treatment groups when they are adjusted by controlling for the possible effects of covariates? Ans: (1) main effects may become nonsignificant or decrease, (2) main effects may change from nonsignificant in ANOVA to significant, (3) statistical significance and effect size may not be different for ANOVA and ANCOVA, and rank order of group means may change following ANCOVA Cognitive Domain: Comprehension Answer Location: 8.8: Conceptual Basis: Factors That Affect the Magnitude of SSAadj and SSresidual and the Pattern of Adjusted Group Means Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 9: Mediation Test Bank Multiple Choice 1. When used in the context of multiple regression, mediation refers to ______. a. an interaction between two predictor variables b. a curvilinear relationship between two predictor variables c. a causal relationship between two or more predictor variables d. collinearity between two predictor or more predictor variables Ans: C Cognitive Domain: Knowledge Answer Location: 9.1: Definition of Mediation Difficulty Level: Easy 2. In a path model, the assumption that X and Y are unrelated would be represented by ______. a. the absence of an arrow between X and Y b. a unidirectional arrow from X to Y c. a unidirectional from Y to X d. a bidirectional arrow between X and Y Ans: A Cognitive Domain: Knowledge Answer Location: 9.1.1: Path Model Notation Difficulty Level: Easy 3. In a path model, the assumption that X causes Y would be represented by ______. a. the absence of an arrow between X and Y b. a unidirectional arrow from X to Y c. a unidirectional from Y to X d. a bidirectional arrow between X and Y Ans: B Cognitive Domain: Knowledge Answer Location: 9.1.1: Path Model Notation Difficulty Level: Easy 4. In a path model, a bidirectional arrow between X and Y would indicate ______. a. the absence of a confounding relationship between X and Y b. the absence of a correlation between X and Y c. a reciprocal causal relationship between X and Y d. a noncausal association between X and Y Ans: D Cognitive Domain: Knowledge Answer Location: 9.1.1: Path Model Notation

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Difficulty Level: Easy 5. Which of the following models suggests that X1 causes of influences X2, which in turn causes or influences Y? a. X1 → X2 = Y b.X1 ↔ X2 = Y c. X1 → X2 → Y d. X1 ↔ X2 ↔ Y Ans: C Cognitive Domain: Comprehension Answer Location: 9.1.1: Path Model Notation Difficulty Level: Medium 6. Although path models are called causal models, “causal” is rarely appropriate because data used to estimate relationships in path models ______. a. are almost always obtained in nonexperimental studies b. involve significant multicollinear relationships c. are based on data obtained using unreliable measures d. test hypotheses of little practical significance Ans: A Cognitive Domain: Knowledge Answer Location: 9.3: Limitations of “Causal” Models Difficulty Level: Easy 7. Given the mediation model equation, c = ab + c’, and X1 → X2 → Y, ab represents the ______. a. direct effect of X1 on Y b. direct effect of X2 on Y c. indirect effect of X1 on Y d. indirect effect of X2 on Y Ans: C Cognitive Domain: Knowledge Answer Location: 9.4: Questions in a Mediation Analysis Difficulty Level: Easy 8. Given the mediation model equation, c = ab + c’, and task clarity → group cohesiveness → job satisfaction, ab represents the ______. a. direct effect of task clarity on job satisfaction b. direct effect of group cohesiveness on job satisfaction c. indirect effect of task clarity on job satisfaction d. indirect effect of group cohesiveness on job satisfaction Ans: C Cognitive Domain: Application Answer Location: 9.4: Questions in a Mediation Analysis Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 9. Given the mediation model equation, c = ab + c’, and X1 → X2 → Y, c’ represents the ______. a. direct effect of X1 on Y b. direct effect of X2 on Y c. indirect effect of X1 on Y d. indirect effect of X2 on Y Ans: A Cognitive Domain: Knowledge Answer Location: 9.4: Questions in a Mediation Analysis Difficulty Level: Easy 10. Given the mediation model equation, c = ab + c’, and task clarity → group cohesiveness → job satisfaction, c’ represents the ______. a. direct effect of task clarity on job satisfaction b. direct effect of group cohesiveness on job satisfaction c. indirect effect of task clarity on job satisfaction d. indirect effect of group cohesiveness on job satisfaction Ans: A Cognitive Domain: Application Answer Location: 9.4: Questions in a Mediation Analysis Difficulty Level: Medium 11. Which of the following represents a plausible mediation pathway? a. mental activity → age → memory loss b. age → mental activity → memory loss c. memory loss → mental activity → age d. memory loss → age → mental activity Ans: B Cognitive Domain: Comprehension Answer Location: 9.5: Issues in Designing a Mediation Analysis Study Difficulty Level: Medium 12. Which of the following variables potentially would be acceptable as X1 in a standard mediation analysis, but not as X2 or Y? a. happiness b. stress c. age d. relationship status Ans: D Cognitive Domain: Comprehension Answer Location: 9.5.1: Type of Variables in Mediation Analysis Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 13. If a mediation model that hypothesized that Stress → Conflict → Anxiety yielded the following standardized path coefficients: bstress-anxiety = .6, bstress-conflict = .4, bconflict-anxiety = .5, the value of the mediated effect is ______. a. .20 b. .09 c. .24 d. .30 Ans: C Cognitive Domain: Application Answer Location: 9.8.1: The Mediated or Indirect Path: ab Difficulty Level: Medium 14. If a mediation model that hypothesized that Stress → Conflict → Anxiety yielded the following standardized path coefficients: bstress-anxiety = .6, bstress-conflict = .4, bconflict-anxiety = .5, the value of the total effect is ______. a. .15 b. .80 c. .29 d. .12 Ans: B Cognitive Domain: Application Answer Location: 9.8.2: Mediated and Direct Path as Partition of Total Effect Difficulty Level: Medium 15. The test of significance for mediation models tests H0: ab = 0 as its null hypothesis and uses an estimate of the standard error of the ab product is ______. a. a causal steps approach b. the joint significance test c. the Sobel test d. a bootstrapped confidence interval Ans: C Cognitive Domain: Knowledge Answer Location: 9.9.3: Sobel Test of H0: ab = 0 Difficulty Level: Easy 16. A test of significance for mediation models that uses several thousand repeated samples to estimate the standard error of the ab product ______. a. uses a causal steps approach b. is the joint significance test c. is the Sobel test d. results in a bootstrapped confidence interval Ans: D Cognitive Domain: Knowledge Answer Location: 9.9.4: Bootstrapped Confidence Interval for ab Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 True/False 1. If X1 causes or influences X2, X1 will have no additional effect on Y. Ans: F Cognitive Domain: Knowledge Answer Location: 9.1: Definition of Mediation Difficulty Level: Easy 2. If X1 is hypothesized to cause X2, X1 must occur before X2. Ans: T Cognitive Domain: Knowledge Answer Location: 9.1.2: Circumstances in which Mediation May Be a Reasonable Hypothesis Difficulty Level: Easy 3. Large, statistically significant path coefficients provide proof that one variable causes another. Ans: F Cognitive Domain: Knowledge Answer Location: 9.3.2: Possible Interpretations for Statistically Significant Paths Difficulty Level: Easy 4. Single mediation analyses provide preliminary nonexperimental evidence that causal models are plausible. Ans: T Cognitive Domain: Knowledge Answer Location: 9.3.2: Possible Interpretations for Statistically Significant Paths Difficulty Level: Easy 5. The role of temporal precedence in establishing causality can be overcome by mediation analysis. Ans: F Cognitive Domain: Understanding Answer Location: 9.5.2: Temporal Precedence or Sequence of Variables in Mediation Studies Difficulty Level: Medium 6. Mediation analysis may provide useful information when bivariate analyses reveal no relationships between any pairs of variables. Ans: F Cognitive Domain: Knowledge Answer Location: 9.6: Assumptions in Mediation Analysis and Preliminary Data Screening Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 7. If one is using the causal-steps approach to determine the statistical significance of a mediation analysis, a nonsignificant c’ indicates that the model reflects complete mediation. Ans: T Cognitive Domain: Knowledge Answer Location: 9.9.1: Causal-Steps Approach Difficulty Level: Easy 8. Only unstandardized path coefficients should be reported in research reports. Ans: F Cognitive Domain: Knowledge Answer Location: 9.10: Effect-Size Information Difficulty Level: Easy 9. Mediation and moderation can occur together. Ans: T Cognitive Domain: Knowledge Answer Location: 9.12: Additional Examples of Mediation Models Difficulty Level: Easy 10. More complex mediation models can be analyzed using structural equation modeling. Ans: T Cognitive Domain: Knowledge Answer Location: 9.13: Note About Use of Structural Equation Modeling Programs to Test Mediation Models Difficulty Level: Easy Essay 1. What are the four possible path models involving two variables, and how would they be denoted in a path diagram? Ans: (1) X and Y are not related (X [no arrow] Y), (2) X causes Y (X → Y), (3) Y causes X (Y → X), and (4) X and Y are correlated, but not because of any causal influence (XY) Cognitive Domain: Knowledge Answer Location: 9.1.1: Path Model Notation Difficulty Level: Easy 2. What are three reasons that some path coefficients may not be statistically significant? Ans: (1) there is no causal or noncausal association between the variables, (2) sampling error is present, and (3) there are serious violations of the assumptions of regression analysis Cognitive Domain: Knowledge

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Answer Location: 9.3.1: Reasons Why Some Path Coefficients May Be Not Statistically Significant Difficulty Level: Easy 3. If the role of stress (X1) in predicting depression following disasters (Y) is mediated by the self-esteem (X2), how would the mediated effect of stress on depression through self-esteem be calculated? Ans: The standardized coefficients for stress predicting self-esteem and self-esteem predicting depression would be multiplied. Cognitive Domain: Comprehension Answer Location: 9.8.1: The Mediated or Indirect Path: ab Difficulty Level: Medium 4. If the role of stress (X1) in predicting depression following disasters (Y) is mediated by the self-esteem (X2), how would the total effect of stress on depression through selfesteem be calculated? Ans: The standardized coefficients for stress predicting self-esteem and self-esteem predicting depression would be multiplied, and the standardized coefficient for stress predicting depression would be added to the product. Cognitive Domain: Comprehension Answer Location: 9.8.2: Mediated and Direct Path as Partition of Total Effect Difficulty Level: Medium 5. How might suppression occur in a mediation model and what is its effect? Ans: Positive direct effects and negative indirect effects tend to cancel each other out, resulting in a small, nonsignificant total effect. Cognitive Domain: Knowledge Answer Location: 9.9.1: Causal-Steps Approach Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 10: Discriminant Analysis Test Bank Multiple Choice 1. Discriminant analysis differs from multiple regression in that discriminant analysis ______. a. uses only categorical variables to predict a quantitative Y variable b. uses only categorical variables to predict a categorical Y variable c. involves a categorical Y that is predicted by quantitative or categorical variables d. involves two or more quantitative Y variables that are predicted by quantitative or categorical variables Ans: C Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 2. The discriminant function obtained in a discriminant analysis can be thought of as the ______. a. best possible linear combination of the X predictor variables that maximizes SSbetween b. maximum correlation of a quantitative Y variable with the linear composite of the X variables c. values assigned to the categorical Y variables d. maximum classification accuracy index of the model Ans: A Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 3. Discriminant functions are said to be optimal in the sense that they ______. a. have the largest possible value of SSwithin b. yield the highest possible percentage of group classifications c. yield the lowest possible value of SSresidual d. have the maximum possible correlation with Y Ans: B Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 4. If a student applying to a graduate program who would have succeeded in the program is denied admission based on the results of a discriminant analysis, the admissions committee has made a ______.

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 a. logistic error b. nonlinear error c. falsification error d. classification error Ans: D Cognitive Domain: Comprehension Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Medium 5. Scores on a discriminant function do not matter because they have been scaled to have a mean of ______. a. 0 and a within groups variance of 1 b. 1 and a within groups variance of 1 c. 0 and a between groups variance of 1 d. of 1 and a between groups variance of 1 Ans: A Cognitive Domain: Application Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Medium 6. If the outcome variable in a discriminant analysis includes 4 groups and the number of X predictors is 5, the number of possible discriminant functions is ______. a. 4 b. 3 c. 5 d. 2 Ans: B Cognitive Domain: Application Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Medium 7. The goal of discriminant analysis is to determine the weighted linear combination of scores on standardized predictor variables that yields the ______. a. largest possible variance within groups b. smallest possible variance within groups c. smallest possible variance between groups d. largest possible variance between groups Ans: D Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 8. A territorial map in discriminant function is essentially a plot of the ______. a. number of predictor scores for each discriminant function b. error scores and predicted Y scores for each discriminant function c. normality curves for each of the predictor variables

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 d. predicted Y scores for each discriminant function Ans: D Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 9. In the omnibus test for discriminant analysis, Wilk’s Λ is defined as the ______. a. degree of fit between the model’s predicted Y group membership and the actual group membership b. correlation of a predictor variable and its associated discriminant function c. proportion of error variance or unexplained variance in Y group membership scores d. change in R2 between two discriminant functions Ans: C Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 10. The process of examining the statistical significance of two or more discriminant functions to identify those to be retained for further analysis and interpretation is conducted using ______. a. dimension reduction analysis b. classification assessment c. coefficient ranking d. dimension screening Ans: A Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 11. When conducting a dimension reduction analysis, researchers should be aware that ______. a. the proportion of between-groups variance decreases as the number of functions decreases b. classification accuracy increases with each successive function c. discriminant function coefficients remain the same for each successive function d. the size of the correlation between each function decreases with each successive function Ans: A Cognitive Domain: Comprehension Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Medium 12. The assumptions underlying discriminant analysis are similar to those underlying multiple regression except that ______. a. predictor scores cannot have negative values b. no interactions between or among predictor variables may be present

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. the variance/covariance matrices must be homogeneous across groups d. the correlations between the discriminant functions must be equal Ans: C Cognitive Domain: Knowledge Answer Location: 10.3: Screening for Violations of Assumptions Difficulty Level: Easy 13. A potential problem with using Box’s M test to assess the homogeneity of variance/covariance matrices following a discriminant analysis is that ______. a. it can be used following analyses involving only one discriminant function b. unproblematic results may be indicated following analyses involving large samples c. it does not control for Type I errors when more than one discriminant function is derived d. levels of significance cannot be adjusted based on the purpose of the analysis Ans: B Cognitive Domain: Comprehension Answer Location: 10.3: Screening for Violations of Assumptions Difficulty Level: Medium 14. An important consideration when planning the sample size for a discriminant analysis is that ______. a. violations of assumptions are more likely with larger samples b. each dependent variable group should include the same number of subjects c. a maximum of 30 subjects should be identified for each predictor variable d. the number of subjects should exceed the number of predictor variables Ans: D Cognitive Domain: Knowledge Answer Location: 10.4: Issues in Planning a Study Difficulty Level: Easy 15. If Wilk’s Λ is relatively small and the omnibus F is statistically significant, it is reasonable to conclude that considering all discriminant functions as a set ______. a. group membership can be predicted better than chance b. the between-groups variance is very small c. the homogeneity of variance/covariance assumption has been violated d. the within-groups variance is very large Ans: A Cognitive Domain: Comprehension Answer Location: 10.5: Equations for Discriminant Analysis Difficulty Level: Medium 16. When calculated as as a component of discriminant analysis, eigenvalues are the basis for ______. a. determining the statistical significance of individual discriminant functions b. calculating a measure of effect size called canonical correlation c. determining the Wilk’s Λ for individual discriminant functions

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 d. testing the homogeneity of variance/covariance assumption for the analysis Ans: B Cognitive Domain: Comprehension Answer Location: 10.5: Equations for Discriminant Analysis Difficulty Level: Medium 17. The structure coefficient for a discriminant function is the ______. a. standardized discriminant function coefficient b. nonstandardized discriminant function coefficient c. correlation of a specific predictor variable with that function d. variance within that function explained by a specific predictor variable Ans: C Cognitive Domain: Comprehension Answer Location: 10.5: Equations for Discriminant Analysis Difficulty Level: Medium 18. In order to determine the effect size of a discriminant analysis using a discriminating variables and discriminant functions, the overall Wilk’s Λ is used to calculate ______. a. r2 b. β c. R2 d. η2 Ans: D Cognitive Domain: Knowledge Answer Location: 10.7: Effect Size Difficulty Level: Easy 19. The effect size of each discriminant function is based on ______. a. ANOVA b. canonical correlation c. standardized regression coefficients d. dimension reduction estimates Ans: B Cognitive Domain: Knowledge Answer Location: 10.7: Effect Size Difficulty Level: Easy 20. The first decision to make when interpreting the results of a discriminant analysis is to ______. a. label each discriminant function in meaningful terminology b. identify common classification errors that may alter the underlying theory c. examine Wilk’s Λ to see if the overall model is statistically significant d. determine the contribution each predictor makes to the overall prediction Ans: C Cognitive Domain: Application

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Answer Location: 10.9: Follow-Up Tests to Assess What Pattern of Scores Best Differentiates Groups Difficulty Level: Medium True/False 1. Like multiple regression, discriminant analysis uses optimal linear combinations of predictor variables to predict and to explain outcome variables. Ans: True Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 2. When both the predictor and dependent variables are categorical, log linear analysis is more appropriate than discriminant analysis. Ans: True Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 3. The number of different discriminant functions in any discriminant analysis is the maximum of the number of groups –1and the number of discriminating variables. Ans: False Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 4. If the categorical variable to be predicted in a discriminant analysis includes three groups, there will be three functions. Ans: False Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 5. A discriminant function combining scores from two or more variables may distinguish two or more groups, even if the groups do not differ significantly on any of the variables. Ans: True Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 6. In a discriminant analysis involving five groups, discrimination function #4 will be the most predictive function. Ans: False Cognitive Domain: Knowledge

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 7. Wilk’s Λ is interpreted as the proportion of variance in discriminant function scores that is predictable from group membership. Ans: False Cognitive Domain: Knowledge Answer Location: 10.5: Equations for Discriminant Analysis Difficulty Level: Easy 8. A small Wilk’s Λ is to be expected when a territorial map shows clear separation of groups. Ans: True Cognitive Domain: Knowledge Answer Location: 10.6 Conceptual Basis: Factors That Affect the Magnitude of Wilk’s Λ Difficulty Level: Easy 9. There is no effect-size index for individual discriminating variables that can be interpreted as a proportion of variance. Ans: True Cognitive Domain: Knowledge Answer Location: 10.7: Effect Size Difficulty Level: Easy 10. When planning a discriminant analysis, smaller expected effect sizes suggest a need for larger sample sizes. Ans: True Cognitive Domain: Knowledge Answer Location: 10.8: Statistical Power and Sample Size Recommendations Difficulty Level: Easy Essay 1. Describe the two uses of discriminant analysis. Ans: (1) Make predictions about group membership outcomes and (2) describe the nature of the differences between two (or more) groups. Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 2. What three possible reasons that a researcher conducting a study to identify predictors of successful completion of a graduate program might have for conducting a discriminant analysis using characteristics of admitted applicants?

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Ans: To (1) accurately predict those applicants who will complete the program, (2) obtain a description of the nature of differences between admitted applicants who will complete the program and those who will not, or (3) to obtain both kinds of information Cognitive Domain: Application Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Medium 3. Why is discriminant analysis an improvement over a series of univariate ANOVAs? Ans: Discriminant analysis takes intercorrelations among predictor variables into account, which reduces the risk of Type I errors. Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 4. What are three factors that must be considered when interpreting the results of a discriminant analysis involving three groups? Ans: (1) The first function must have the largest possible between-groups differences, (2) scores on the second function must be uncorrelated with scores on the first function, and (3) scores on the second function must predict the largest possible between-groups variance not explained by the first function Cognitive Domain: Comprehension Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Medium 5. List four possible research questions that can be addressed when assessing the results of a discriminant analysis. Ans: (1) Is the overall model significant? (2) How many discriminant functions are useful in differentiating groups? (3) Which variables are useful in discriminating among groups or predicting group membership? (4) Which groups can (or cannot) be differentiated? (5) What meaning (name or label) can be given to each discriminant function? (6) What type of classification errors is the most common? Cognitive Domain: Knowledge Answer Location: 10.1: Research Situations and Research Questions Difficulty Level: Easy 6. What are two potential problems that may arise when using Box’s M test to assess the assumption of homogeneity of variance/covariance matrices in a discriminant analysis? Ans: (1) Box’s M is very sensitive to nonnormality in the distribution of scores, and (2) when the number of cases is very large, Box’s M may indicate statistically significant violations of homogeneity of variance/covariance matrices, even when departures are not sufficiently serious to raise questions about the analysis results. Cognitive Domain: Knowledge Answer Location: 10.3: Screening for Violations of Assumptions Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 7. How can researchers determine the discriminating variables that provide useful information about group membership? Ans: By examining the (1) sizes and (2) signs of the discriminant function coefficients. Cognitive Domain: Application Answer Location: 10.5: Equations for Discriminant Analysis Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 11: Multivariate Analysis of Variance Test Bank Multiple Choice 1. MANOVA can be considered an extension of ANOVA because ______. a. the outcome variable can be categorical or quantitative b. more than two independent variables are included in an analysis c. group differences are compared on two or more outcome variables d. three-way and four-way interactions can be analyzed Ans: C Cognitive Domain: Comprehension Answer Location: 11.1: Research Situations and Research Questions Difficulty Level: Medium 2. The null hypothesis for MANOVA is that the ______. a. means of all outcome variables are equal across all groups b. means for at least outcome variable is equal across all groups c. covariance of each pair of outcome variables is equal across all groups d. cross-product of each pair of outcome variables is equal across all groups Ans: A Cognitive Domain: Knowledge Answer Location: 11.1: Research Situations and Research Questions Difficulty Level: Easy 3. MANOVA may be useful when ______. a. the outcome variables are theoretically unrelated or unsupported by previous research b. group separation is sought among highly correlated outcome variables c. the aim is to eliminate the risk of Type I and Type II errors completely d. simultaneous consideration of outcome variables may reveal differences not observed on individual outcome variables Ans: D Cognitive Domain: Comprehension Answer Location: 11.3: Why Include Multiple Outcome Measures? Difficulty Level: Medium 4. A difference between discriminant analysis and MANOVA is that ______. a. discriminant analysis requires inclusion of the cross-product variance of all pairs of predictor variables, but MANOVA does not b. MANOVA can include multiple categorical variables, but discriminant analysis is usually limited to one categorical variable

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. discriminant analysis requires homogeneity of variance/covariance, but MANOVA does not d. discriminant analysis can be adapted to include quantitative outcome variables, but MANOVA cannot Ans: B Cognitive Domain: Comprehension Answer Location: 11.4: Equivalence of MANOVA and DA Difficulty Level: Medium 5. The primary characteristic of general linear models (GLM) is that ______. a. at least one pair of variables in the model must be linearly related b. all pairs of quantitative variables in the model must be linearly related c. only categorical outcome variables are permitted d. all analyses must be conducted using standardized scores Ans: B Cognitive Domain: Knowledge Answer Location: 11.5: The General Linear Model Difficulty Level: Easy 6. For MANOVA to be appropriate, the magnitude of correlations among the outcome variables should be ______. a. near zero b. small, that is, .10–.30 c. moderate, that is, .30–.70 d. large, that is, above .80 Ans: C Cognitive Domain: Knowledge Answer Location: 11.5: The General Linear Model Difficulty Level: Easy 7. When assessing whether the assumptions for MANOVA have been met, researchers ______. a. can assume that outliers will have little effect on the results b. be sure that the variance/covariance matrices for the outcome variables are heterogeneous c. be sure that pairs of outcome variables are linearly related d. transform scores so any two outcome variables are dependent Ans: C Cognitive Domain: Comprehension Answer Location: 11.6: Assumptions and Data Screening Difficulty Level: Medium 8. A researcher conducting a study of anxiety who administers a paper-and-pencil inventory to participants at the beginning of the study, videotapes some sessions to look for behavioral signs of anxiety, and randomly measures the heart rate of participants is using ______.

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 a. triangulation b. qualitative assessment c. social desirability monitoring d. desensitization Ans: A Cognitive Domain: Knowledge Answer Location: 11.7: Issues in Planning a Study Difficulty Level: Easy 9. When the assumptions of MANOVA are seriously violated, the multivariate test statistic most frequently chosen is ______. a. Wilk’s lambda b. Hotelling’s trace c. Pillai’s trace d. Roy’s largest root Ans: C Cognitive Domain: Knowledge Answer Location: 11.9: Multivariate Test Statistics Difficulty Level: Easy 10. The multivariate test that can be used when the researcher wants to test only the significance of the first discriminant function is ______. a. Wilk’s lambda b. Hotelling’s trace c. Pillai’s trace d. Roy’s largest root Ans: D Cognitive Domain: Knowledge Answer Location: 11.9: Multivariate Test Statistics Difficulty Level: Easy 10 An estimate of the effect size for each effect in a MANOVA can be estimated by ______. a. Λ b. 1 – Λ c. Λ2 d. 1 – Λ2 Ans: B Cognitive Domain: Knowledge Answer Location: 11.11: Effect Size for MANOVA Difficulty Level: Easy 11. An important reason for using discriminant analysis as the follow-up analysis when the MANOVA omnibus test is significant is that discriminant analysis ______. a. overcomes violations of the homogeneity of the variance/covariance assumption b. is more robust in situations involving small cell sizes

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. can help overcome certain theoretical omissions d. takes intercorrelations of individual outcome variables into account Ans: D Cognitive Domain: Comprehension Answer Location: 11.16: Comparison of Univariate and Multivariate Follow-Up Analyses Difficulty Level: Medium True/False 1. Even when a MANOVA is significant, Type I errors can still occur in follow-up test results. Ans: T Cognitive Domain: Knowledge Answer Location: 11.1: Research Situations and Research Questions Difficulty Level: Easy 2. Compared to ANOVA, MANOVA reduces the risk of Type II errors. Ans: F Cognitive Domain: Knowledge Answer Location: 11.1: Research Situations and Research Questions Difficulty Level: Easy 3. One-way MANOVA can be conceptualized as the reverse of discriminant analysis. Ans: T Cognitive Domain: Knowledge Answer Location: 11.4: Equivalence of MANOVA and DA Difficulty Level: Easy 4. MANOVA can include more than one categorical predictor. Ans: T Cognitive Domain: Knowledge Answer Location: 11.4: Equivalence of MANOVA and DA Difficulty Level: Easy 5. The basic statistical analysis at the core of general linear models is correlation. Ans: T Cognitive Domain: Knowledge Answer Location: 11.5: The General Linear Model Difficulty Level: Easy 6. MANOVA is appropriate even when the multiple outcome variables are highly correlated. Ans: F Cognitive Domain: Knowledge Answer Location: 11.5: The General Linear Model

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Difficulty Level: Easy 7. Because MANOVA is a robust analysis, outcome variables are protected against violations of assumptions. Ans: F Cognitive Domain: Knowledge Answer Location: 11.6: Assumptions and Data Screening Difficulty Level: Easy 8. Compared to ANOVA, MANOVA allows for a larger number of groups. Ans: F Cognitive Domain: Knowledge Answer Location: 11.7: Issues in Planning a Study Difficulty Level: Easy 9. MANOVA can correct for overlap among test scores of several different constructs. Ans: T Cognitive Domain: Knowledge Answer Location: 11.7: Issues in Planning a Study Difficulty Level: Easy 10. A large value of Wilk’s Λ indicates large group differences. Ans: F Cognitive Domain: Knowledge Answer Location: 11.8: Conceptual Basis of MANOVA Difficulty Level: Easy 11. Small sample sizes in MANOVA increase the likelihood of unequal variances and covariances across groups. Ans: T Cognitive Domain: Knowledge Answer Location: 11.12: Statistical Power and Sample Size Decisions Difficulty Level: Easy 12. Because using MANOVA as an omnibus test controls for Type I errors, Type I errors are also controlled in univariate follow-up tests like Tukey’s test. Ans: F Cognitive Domain: Knowledge Answer Location: 11.13: One-Way MANOVA: Career Group Data Difficulty Level: Easy Essay 1. What are three advantages of MANOVA when compared to ANOVA?

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Ans: A series of ANOVAs (1) inflates the risk of Type I errors, (2) does not take intercorrelations into account, and (3) in situations in which groups do not differ on any one outcome variable, differences among groups when two or more outcome variables are considered jointly. Cognitive Domain: Comprehension Answer Location: 11.3: Why Include Multiple Outcome Measures? Difficulty Level: Medium 2. How do researcher interests in supplemental or follow-up analyses tend to differ for MANOVA and discriminant analysis? Ans: Researchers using MANOVA tend to be primarily interested in identifying groups that differ significantly, while researchers using discriminant analysis tend to be primarily interested in determining the relative predictive usefulness of individual discriminating variables. Cognitive Domain: Comprehension Answer Location: 11.4: Equivalence of MANOVA and DA Difficulty Level: Medium 3. Compare variance partitioning in ANOVA and MANOVA. Ans: In ANOVA, the total variance (SStotal) is partitioned into two parts, SSbetween and SSwithin so that SStotal = SSbetween + SSwithin. Because MANOVA includes multiple outcome variables, variance must include both the variance of the outcome variables and the covariance of all possible pairs of outcome variables, which is represented by a sum of cross-products matrix (SCP). Analogous to ANOVA partitioning of variance, MANOVA variance can be partitioned as SCPtotal = SCPbetween + SCPwithin. Cognitive Domain: Analysis Answer Location: 11.8: Conceptual Basis of MANOVA Difficulty Level: Hard 4. How might sample size affect the alpha values used for tests like Box’s M or Levene’s test when these tests are being used to assess assumptions of equality of variances and covariance in MANOVA? Ans: When N is small, large alpha values make detection of violations of these assumptions easier. However, when N is large, small alpha values may make detection of violations of these assumptions more likely. Cognitive Domain: Comprehension Answer Location: 11.12: Statistical Power and Sample Size Decisions Difficulty Level: Medium 5. What are some problems that may arise when univariate analyses are used following a MANOVA? Ans: (1) Such analyses may result in an inflated risk of Type I errors. (2) Because they do not take intercorrelations among individual outcome variables into account, they may lead to mistaken decisions about significant differences among pairs of outcome variables that are so highly correlated that they may be measures of the same underlying construct. (3) Suppression may occur, and an outcome variable that is

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 nonsignificant in a univariate analysis may be important in forming a discriminant function. (4) Univariate and multivariate analyses may yield different rank orders of means on an outcome variable. Cognitive Domain: Comprehension Answer Location: 11.16: Comparison of Univariate and Multivariate Follow-Up Analyses Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 12: Exploratory Factor Analysis Test Bank Multiple Choice 1. When conducting principal components analysis (PC), the term, component, is defined as ______. a. the total variance shared among the variables b. a latent variable that represents an underlying characteristic not directly measured c. the total amount of shared variance among the variables d. any observed variable that can be directly measured Ans: B Cognitive Domain: Knowledge Answer Location: 12.1: Research Situations Difficulty Level: Easy 2 When conducting principal axis factor analysis (PAF), the term, factor, is defined as______. a. the total variance shared among the variables b. a latent variable that represents an underlying characteristic not directly measured c. the total amount of shared variance among the variables d. any observed variable that can be directly measured Ans: B Cognitive Domain: Knowledge Answer Location: 12.1: Research Situations Difficulty Level: Easy 3. Factor analysis is designated as exploratory when the ______. a. number of factors is specified in advance b. model will not be orthogonal c. number of variables expected to be associated with a factor is not specified in advance d. data analysts have hypotheses about the underlying factor structure Ans: C Cognitive Domain: Knowledge Answer Location: 12.1: Research Situations Difficulty Level: Easy 4. If scores on a reading achievement test and a test of ability to solve analogies are believed to explain mental ability, then mental ability would be viewed as ______. a. latent variable b. a predictor variable

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. an outcome variable d. an explanatory variable Ans: A Cognitive Domain: Knowledge Answer Location: 12.1: Research Situations Difficulty Level: Easy 5. An important decision in factor analysis is determining the underlying factors or components which is the process of _______. a. rotation b. extraction c. reproduction d. iteration Ans: B Cognitive Domain: Knowledge Answer Location: 12.8.2: Computation of the Initial Factor Loading Matrix A Difficulty Level: Knowledge 6. If a factor analysis reveals that two variables, X1 and X2, are significantly correlated, the appropriate interpretation is that ______. a. X1 causes X2 so that they can be thought of as one construct or factor b. X1 or X2 are leading to the same scores on some underlying latent variable c. an underlying variable may account for the correlation between X1 and X2 d. X1 and X2 are measured without error Ans: C Cognitive Domain: Knowledge Answer Location: 12.2: Path Model for Factor Analysis Difficulty Level: Easy 7. If a factor analysis indicates that eight measured variables can be explained by three latent variables, the analysis has achieved the goal of ______. a. proof of the latent variable existence b. data reduction c. outcome prediction d. complete extraction Ans: B Cognitive Domain: Knowledge Answer Location: 12.3: Factor Analysis as a Method of Data Reduction Difficulty Level: Easy 8. If a factor analysis involves four variables, a, b, c, and d, which of the following interpretations would be correct? a. If none of the variables are significantly correlated with each another, a one-factor solution is indicated. b. If a and c are significantly correlated, but b and d are not significantly correlated with any other variable, a two-factor solution is indicated.

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. If all the variables are significantly correlated with each another, a four-factor solution is indicated. d. If all the variables are significantly correlated with one another, a one-factor solution is indicated. Ans: D Cognitive Domain: Comprehension Answer Location: 12.3: Factor Analysis as a Method of Data Reduction Difficulty Level: Medium 9. When screening a dataset prior to conducting a factor analysis, analysts should ______. a. recognize that factor analysis can adjust for nonlinear associations among variables b. verify that the off-diagonal correlations in the correlation matrix are all near 0 c. understand that scores on the variables to be included in the analysis may be dichotomous d. ensure that a fairly large number of the correlations among the variables to be included are > .30 Ans: D Cognitive Domain: Knowledge Answer Location: 12.5: Screening for Violations of Assumptions Difficulty Level: Easy 10 When planning a factor analysis, researchers should remember that ______. a. the general rule is that the sample size should always be greater than 100 b. a ratio of 5-8 subjects per variable is generally adequate c. lower variance among the variables yields more accurate results d. correlations between pairs of variables should be low Ans: A Cognitive Domain: Knowledge Answer Location: 12.6: Issues in Planning a Factor-Analytic Study Difficulty Level: Easy 11. The elements in a factor loading matrix A correspond to the ______. a. correlation of each variable with all the other variables b. correlation of each factor with the other factors c. reproduced predicted X scores from the factor loadings d. correlation of each variable with each factor Ans: D Cognitive Domain: Knowledge Answer Location: 12.7: Computation of Factor Loadings Difficulty Level: Easy 12. The total variance of any X variable in a factor has a value of 1 and is called a(n) ______. a. rotation b. communality

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. extraction d. iteration Ans: B Cognitive Domain: Knowledge Answer Location: 12.7: Computation of Factor Loadings Difficulty Level: Easy 13. The multiple steps that may be involved in principal axis factoring are called ______. a. rotations b. communalities c. extractions d. iterations Ans: D Cognitive Domain: Knowledge Answer Location: 12.7: Computation of Factor Loadings | 12.8.2: Computation of the Initial Factor Loading Matrix Difficulty Level: Easy 14. When all the factor loadings for a factor (or component) are squared and then added together, the resulting value corresponds to the _______. a. communality b. structure coefficient c. eigenvalue d. unique variance Ans: C Cognitive Domain: Knowledge Answer Location: 12.8.3: Limiting the Number of Components or Factors Difficulty Level: Easy 15. The factor loading that a data analyst would find easiest to interpret is ______. a. .03 b. .25 c. .40 d. .55 Ans: A Cognitive Domain: Knowledge Answer Location: 12.8.4: Rotation of Factors Difficulty Level: Easy 16. The correlation between two factors is equal to the ______. a. sine of the angle between the two factors b. cosine of the angle between the two factors c. angle between the two factors expressed in radians d. angle between the two factors divided by 90 Ans: B Cognitive Domain: Knowledge

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Answer Location: 12.13: Geometric Representation of Factor Rotation Difficulty Level: Easy 17. Which of the following statements characterizes two orthogonal factors? a. The correlation between the two factors is greater than 0. b. The two factors have equal eigenvalues. c. The two factors have variance values greater than 1.0. d. The angle between the two factors is 90 degrees. Ans: D Cognitive Domain: Knowledge Answer Location: 12.13: Geometric Representation of Factor Rotation Difficulty Level: Easy 18. Generally, the percentage of variance explained by retained factors should be ______. a. 10% – 20% b. 20% – 30% c. 40% – 70% d. 50% – 80% Ans: C Cognitive Domain: Knowledge Answer Location: 12.16.2: How “Important” Are the Factors or Components? How Much Variance Does Each Factor or Component Explain? Difficulty Level: Easy True/False 1. In factor analytic terms, a latent variable is an unobservable variable suggested by the intercorrelations of measured or observed variables. Ans: T Cognitive Domain: Knowledge Answer Location: 12.1: Research Situations Difficulty Level: Easy 2. In principal component analysis, the term “component” refers to a measured variable. Ans: F Cognitive Domain: Knowledge Answer Location: 12.1: Research Situations Difficulty Level: Easy 3. In principal axis factoring, the term “factor” refers to a latent variable. Ans: T Cognitive Domain: Knowledge Answer Location: 12.1: Research Situations Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 4. A goal of reducing the number of measured variables from 12 variables to a model including 1-5 factors would be considered reasonable. Ans: T Cognitive Domain: Knowledge Answer Location: 12.3: Factor Analysis as a Method of Data Reduction Difficulty Level: Easy 5. Exploratory factor analysis provides proof that latent variables underlie variables included in a dataset. Ans: F Cognitive Domain: Knowledge Answer Location: 12.3: Factor Analysis as a Method of Data Reduction Difficulty Level: Easy 6. If all the off-diagonal correlations are 0, no reduction in the number of factors is possible. Ans: T Cognitive Domain: Knowledge Answer Location: 12.5: Screening for Violations of Assumptions Difficulty Level: Easy 7. If all the off-diagonal correlations are 1, a theoretically infinite number of factors is possible. Ans: F Cognitive Domain: Knowledge Answer Location: 12.5: Screening for Violations of Assumptions Difficulty Level: Easy 8. In factor analysis, the number of factors and the nature of those factors depends entirely on the set of variables selected by the researcher. Ans: T Cognitive Domain: Knowledge Answer Location: 12.6: Issues in Planning a Factor-Analytic Study Difficulty Level: Easy 9. Researchers primarily interested in reducing information from a large set of variables to scores on a smaller number of components would prefer principal axis factoring. Ans: F Cognitive Domain: Knowledge Answer Location: 12.8.2: Computation of the Initial Factor Loading Matrix A Difficulty Level: Easy 10. A rule-of-thumb is that at least three indicator variables should be associated with every factor. Ans: T

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Cognitive Domain: Knowledge Answer Location: 12.8.3: Limiting the Number of Components or Factors Difficulty Level: Easy Essay 1. What is the basic purpose of exploratory factor analysis? Ans: To see if it makes sense to interpret a set of measured variables as measures of some smaller number of underlying constructs, or latent variables. Cognitive Domain: Knowledge Answer Location: 12.1: Research Situations Difficulty Level: Easy 2. What are three potential advantages of factor analysis? Ans: Factor analysis can be helpful (1) in theory development and testing, (2) in reducing large numbers of predictors to a more manageable number, and (3) developing multiple-item scales. Cognitive Domain: Knowledge Answer Location: 12.3: Factor Analysis as a Method of Data Reduction Difficulty Level: Easy 3. What are two potential disadvantages of factor analysis? Ans: Factor analysis (1) may be used in attempts to summarize datasets in which the selection of variables was not considered carefully, and (2) may be mistakenly viewed as providing proof of the presence of latent variables. Cognitive Domain: Knowledge Answer Location: 12.3: Factor Analysis as a Method of Data Reduction Difficulty Level: Easy 4. What is the key difference between principal component analysis and principal axis factoring? Ans: In principal axis factoring, the initial set of loadings for p variables on p factors accounts for all the variance in each of the p variables. Principal axis factoring attempts to reproduce only the variance in each measured variable that is shared with or predictable from other variables in the analysis. Cognitive Domain: Analysis Answer Location: 12.8.2: Computation of the Initial Factor Loading Matrix A Difficulty Level: Hard 5. What are four questions that should be addressed when interpreting a factor analysis? Ans: (1) How many factors (or components or latent variables) are needed to account for (or reconstruct) the pattern of correlations among the measured variables? (2) How important are the factors or components? (3) How much variance does each factor or component explain? How can the factors (or components) be named or labeled?

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 (4)How well do the retained factors (or components) reproduce the structure of the original dataset? Cognitive Domain: Knowledge Answer Location: 12.16: Questions to Address in the Interpretation of Factor Analysis Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 13: Reliability, Validity, and Multiple-item Scales Test Bank Multiple Choice 1 The concept of reliability refers to the ______. a. content of a measure b. clarity of a measure c. consistency of a measure d. variability of a measure Ans: C Cognitive Domain: Knowledge Answer Location: 13.1.1: Reliability Difficulty Level: Easy 2 The concept describing the extent to which a test or inventory measures the construct or characteristic it purports to measure is ______. a. validity b. clarity c. consistency d. sensitivity Ans: A Cognitive Domain: Knowledge Answer Location: 13.1.2: Validity Difficulty Level: Easy 3. An anxiety inventory that enables clinicians to make distinctions among their clients based on the level of anxiety each client is experiencing is said to have ______. a. validity b. clarity c. consistency d. sensitivity Ans: D Cognitive Domain: Application Answer Location: 13.1.3: Sensitivity Difficulty Level: Medium 4. Measurements are said to be biased when they ______. a. yield inconsistent scores at different times b. are systematically too high or too low relative to the true value of the construct c. are more expensive in certain settings d. vary in level of invasiveness

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Ans: b Cognitive Domain: Knowledge Answer Location: 13.1.4: Bias Difficulty Level: Easy 5. If responding to an anxiety inventory increases the respondent’s level of anxiety, the respondent is experiencing ______. a. reactivity b. clarity c. consistency d. sensitivity Ans: A Cognitive Domain: Application Answer Location: 13.2.3: Reactivity of Measurement Difficulty Level: Medium 6. A researcher who wants to know if a cognitive ability test yields consistent scores when children are assessed at 3-month intervals is interested in ______. a. alternate forms reliability b. internal consistency reliability c. test-retest reliability d. interobserver reliability Ans: C Cognitive Domain: Comprehension Answer Location: 13.3.1: Definition of Reliability Difficulty Level: Medium 7. Clinical psychologists making decisions about whether clients’ levels of depression are sufficiently severe to warrant a recommendation of hospitalization would want the test-retest reliability of the depression inventory to be ______. a. .75 b. .80 c. .85 d. .95 Ans: D Cognitive Domain: Comprehension Answer Location: 13.3.2: Test-Retest Reliability Assessment for a Quantitative Variable Difficulty Level: Medium 8. A developmental psychologist who asks two colleagues to categorize 60 families as authoritative, authoritarian, or permissive could assess the reliability of the ratings using ______. a. alternate forms reliability b. test-retest reliability c. Cronbach’s alpha d. Cohen’s kappa

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Ans: D Cognitive Domain: Comprehension Answer Location: 13.3.3: Interobserver Reliability Assessment for Scores on a Categorical Variable Difficulty Level: Medium 9. In classical measurement theory, if a person obtains a GRE score of 165, but the person’s true GRE score is 170, the error score is ______. a. +5 b. –5 c. 167.5 d. 162.5 Ans: b Cognitive Domain: Comprehension Answer Location: 13.4: Concepts from Classical Measurement Theory Difficulty Level: Medium 10. When interpreting a test-retest reliability coefficient, reliability can be defined as the ______. a. average of all observed scores b. variance of the observed scores minus the variance of the error scores c. ratio of the true score variance to the observed score variance d. ratio of the error score variance to the observed score variance Ans: C Cognitive Domain: Knowledge Answer Location: 13.4.1: Reliability as Partition of Variance Difficulty Level: Easy 11. If a test has a reliability of .70, the proportion of variance in observed scores due to error is ______. a. 70% b. 49% c. 15% d. 30% Ans: D Cognitive Domain: Comprehension Answer Location: 13.4.1: Reliability as Partition of Variance Difficulty Level: Medium 12. A potential advantage of using summated scales is that scores on such scales may ______. a. be less sensitive to individual differences b. be more reliable than single scores c. have less variance than scores based on single items d. be less likely to result in normal distributions Ans: b

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Cognitive Domain: Comprehension Answer Location: 13.5: Use of Multiple-Item Measures to Improve Measurement Reliability Difficulty Level: Medium 13. A researcher who wants to combine information from scales measuring employee ratings of work-group functioning, task-role clarity, supervisory behavior, and overall job satisfaction, all of which have different scoring scales, could address the problem by using ______. a. the sum of the scores because the differences are unlikely to be problematic b. the average or mean score of all the measures c. the z-scores of all the measures d. a log transformation of the measures Ans: C Cognitive Domain: Knowledge Answer Location: 13.6.7.2: Unit-Weighted Sum of z Scores Difficulty Level: Easy 14. A reliability estimate obtained to indicate the degree to which all scale items measure the same construct or variable is an index of ______. a. internal consistency reliability b. test-retest reliability c. split-half reliability d. interobserver reliability Ans: A Cognitive Domain: Knowledge Answer Location: 13.7: Assessment of Internal Homogeneity for Multiple-Item Measures: Cronbach’s Alpha Reliability Coefficient Difficulty Level: Easy 15. Cronbach’s alpha can be interpreted as the ______. a. average of several thousand bootstrapped test-retest reliabilities b. correlation of one half of the items with the other half c. average of all possible split-half reliabilities d. ratio of error variance to true score variance Ans: C Cognitive Domain: Knowledge Answer Location: 13.7.1: Conceptual Basis of Cronbach’s Alpha Difficulty Level: Easy 16. If a college instructor writes test items that clearly relate to information provided in the textbook and lecture notes, the instructor is attempting to ensure that the test has ______. a. concurrent validity b. predictive validity c. content validity d. construct validity

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Ans: C Cognitive Domain: Application Answer Location: 13.8.1: Content and Face Validity Difficulty Level: Medium 17. If the items on a scale designed to measure anxiety appear relevant to the students completing the scale, the scale has ______. a. concurrent validity b. predictive validity c. content validity d. face validity Ans: D Cognitive Domain: Application Answer Location: 13.8.1: Content and Face Validity Difficulty Level: Medium 18. If a psychologist who developed a 12-item scale to assess student stress obtained a strong correlation between scores on the new scale and the Maslach Burnout Inventory, the psychologist would have evidence of ______. a. convergent validity b. discriminant validity c. concurrent validity d. predictive validity Ans: A Cognitive Domain: Application Answer Location: 13.8.2: Criterion-Oriented Validity Difficulty Level: Medium 19. If a psychologist who developed a 12-item scale to assess student stress obtained a weak correlation between scores on the new scale and the Crown-Marlowe Social Desirability scale, the psychologist would have evidence of ______. a. convergent validity b. discriminant validity c. concurrent validity d. predictive validity Ans: b Cognitive Domain: Application Answer Location: 13.8.2: Criterion-Oriented Validity Difficulty Level: Medium 20. If a psychologist who developed a 12-item scale to assess student stress obtained a high correlation between scores on the new scale and the likelihood that students would set appointments at the school’s counseling center, the psychologist would have evidence of ______. a. convergent validity b. discriminant validity

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 c. concurrent validity d. predictive validity Ans: C Cognitive Domain: Application Answer Location: 13.8.2: Criterion-Oriented Validity Difficulty Level: Medium 21. If a psychologist who developed a 12-item scale to assess student stress obtained a high correlation between scores on the new scale at the beginning of the term and the likelihood that students would set appointments at the school’s counseling center during the term, the psychologist would have evidence of ______. a. convergent validity b. discriminant validity c. concurrent validity d. predictive validity Ans: D Cognitive Domain: Application Answer Location: 13.8.2: Criterion-Oriented Validity Difficulty Level: Medium True/False 1. Measures with good evidence of reliability can be expected to also have high levels of measurement error. Ans: False Cognitive Domain: Knowledge Answer Location: 13.1.1: Reliability Difficulty Level: Easy 2. Validity is more difficult to assess than reliability. Ans: True Cognitive Domain: Knowledge Answer Location: 13.1.2: Validity Difficulty Level: Easy 3. Measurement items with fewer numbers of response categories tend to be more sensitive to individual differences. Ans: False Cognitive Domain: Knowledge Answer Location: 13.1.3: Sensitivity Difficulty Level: Easy 4. In basic research situations, increasing test-retest reliability beyond .80 may not be necessary. Ans: True

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Cognitive Domain: Knowledge Answer Location: 13.3.2: Test-Retest Reliability Assessment for a Quantitative Variable Difficulty Level: Easy 5. When measurement errors are summed across many items, the assumption that the sum of the errors will be 1 is reasonable. Ans: False Cognitive Domain: Knowledge Answer Location: 13.5: Use of Multiple-Item Measures to Improve Measurement Reliability Difficulty Level: Easy 6. Items that are reverse-worded on rating scales so that higher values reflect lower amounts of the characteristic must be recoded to match the scored direction of the other items. Ans: True Cognitive Domain: Knowledge Answer Location: Recoding Scores for Reverse-Worded Items Difficulty Level: Easy 7. Cronbach’s alpha is the most widely used method of assessing the reliability of multiple-item scales. Ans: True Cognitive Domain: Knowledge Answer Location: 13.7.1: Conceptual Basis of Cronbach’s Alpha Difficulty Level: Easy Essay 1. List the Likert scale response categories for the following item: “I am satisfied with my job.” Ans: (1) Strongly disagree, (2) Disagree, (3) Neutral/No opinion/Don’t know, (4) Agree, (5) Strongly agree Cognitive Domain: Application Answer Location: 13.1.3: Sensitivity Difficulty Level: Medium 2. What is the difference between a ceiling effect and a floor effect? Ans: Ceiling effect: majority of the scores massed at the high end of the possible score range. Floor effect: most of the scores massed at the low end of the possible score range. Cognitive Domain: Knowledge Answer Location: 13.1.3: Sensitivity Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 3. What are two negative consequences of poor reliability? Ans: (1) measures must be reliable to be valid; therefore, measures with low reliability will have low validity, and (2) low reliability tends to yield lower correlations among variables, which can lead to errors in statistical results. Cognitive Domain: Comprehension Answer Location: 13.4.2: Attenuation of Correlations Due to Unreliability of Measurement Difficulty Level: Medium 4. What are two ways to improve the reliability of a measure? Ans: (1) Increase the variance in true scores by using samples that have more variability on the characteristic being measured, and (2) use more than one measure of the same characteristic in order control for extraneous variables, which may be expected to reduce the magnitude of measurement errors. Cognitive Domain: Comprehension Answer Location: 13.4.2: Attenuation of Correlations Due to Unreliability of Measurement Difficulty Level: Medium 5. What are four potential advantages of using summated scales instead of single measures? Ans: Summated scales (1) are generally more reliable than single scores, (2) generally have more variance than scores based on single items (3) are more likely to result in normal score distributions, and (4) are more sensitive to individual differences, Cognitive Domain: Knowledge Answer Location: 13.5: Use of Multiple-Item Measures to Improve Measurement Reliability Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 14: More about Repeated Measures Test Bank Multiple Choice 1. Another term for interaction is ______. a. additivity b. nonadditivity c. sphericity d. unreliability Ans: B Cognitive Domain: Knowledge Answer Location: 14.4: Test for Participant by Time or Participant by Treatment Interaction Difficulty Level: Easy 2 An index of the extent to which data analyzed using a one-way repeated measures ANOVA violate the sphericity assumption is provided by ______. a. α b. π c. µ d. ε Ans: D Cognitive Domain: Knowledge Answer Location: 14.5: One-way Repeated Measures Results for Heart Rate/Social Stress Data Difficulty Level: Easy 3. Values of ε can be used as a multiplier to ______. a. reduce the degrees of freedom for a univariate repeated measures F ratio b. adjust data analyzed using a one-way repeated measures ANOVA to remove interaction effects c. adjust data analyzed using a one-way repeated measures ANOVA to increase the likelihood of obtaining a statistically significant result d. calculate the degrees of freedom for a univariate repeated measures F ratio necessary to obtain a statistically significant result Ans: A Cognitive Domain: Knowledge Answer Location: 14.5: One-way Repeated Measures Results for Heart Rate/Social Stress Data Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 4. The Greenhouse-Geisser correction is recommended when using a one-way repeated measures ANOVA because it is robust to the violation of ______. a. collinearity b. sphericity c. linearity d. unequal group Ns Ans: B Cognitive Domain: Knowledge Answer Location: 14.5: One-way Repeated Measures Results for Heart Rate/Social Stress Data Difficulty Level: Easy 5. The number of contrasts possible following a study in which a variable was measured at four separate times is ______. a. 3 b. 4 c. 5 d. 8 Ans: A Cognitive Domain: Knowledge Answer Location: 14.6: Testing the Sphericity Assumption Difficulty Level: Easy 6. The advantage of using contrasts in MANOVA when sphericity is violated is that ______. a. the variance of person effects is incorporated into the error variance b. fewer subjects are required to obtain adequate statistical power c. the epsilon (ε) values are larger, which results in a less conservative adjustment to the df values d. the requirements for equality of variances and covariances in MANOVA are less stringent than in ANOVA Ans: D Cognitive Domain: Knowledge Answer Location: 14.7: MANOVA for Repeated Measures Difficulty Level: Easy 7. The most widely used multivariate test statistic used to interpret the results of a MANOVA for repeated measures is ______. a. Pillai’s trace b. Wilk’s Λ c. Hotelling’s trace d. Roy’s largest root Ans: B Cognitive Domain: Knowledge Answer Location: 14.8: Results for HR and Social Stress Analysis Using MANOVA Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 8. A study involving measures of stress, anxiety, heart rate at four points in time would be considered a doubly multivariate repeated measures study if ______. a. randomly selected pairs of the variables are measured at the end of predetermined intervals b. the order in which the pairs of variables are measured is determined by random assignment c. the length of the intervals is varied based on the results of prior measurements d. all three variables are measured at the end of each measurement interval Ans: D Cognitive Domain: Comprehension Answer Location: 14.9: Doubly Multivariate Repeated Measures Difficulty Level: Medium 9. Adding sex as a between-subjects factor in a study involving measuring stress, at four points in time would change the study to a ______. a. repeated measures MANOVA b. doubly multivariate repeated measures MANOVA c. mixed model ANOVA d. mixed model MANOVA Ans: C Cognitive Domain: Comprehension Answer Location: 14.10: Mixed Model MANOVA: Between-S and Within-S Factors Difficulty Level: 10. Lord’s paradox refers to the fact that when comparing pretest and posttest scores ______. a. order effects across the treatment levels may affect different analyses differently b. different statistical analyses can lead to different results c. different methods of counterbalancing treatment order lead to different ANOVA results d. the effects of nonadditivity may differ across factorial repeated-measures Ans: B Cognitive Domain: Knowledge Answer Location: 14.10.5: Comparison of Repeated Measures ANOVA to Difference Score and ANCOVA Approaches Difficulty Level: Easy 11. A researcher interested in the degree of frustration in infants asked to perform three novel tasks, A, B, and C. The tasks were presented in order (A then B then C), the frustration measure was taken immediately after each task, and infants were given a 3minute “cooling down” period between tasks. If repeated-measures ANOVA showed highest frustration levels on C and lowest on A, a potential confound is ______. a. contrast effects b. sensitization c. carryover effects

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 d. task complexity Ans: C Cognitive Domain: Comprehension Answer Location: 14.11: Order and Sequence Effects Difficulty Level: Medium True/False 1. If the sphericity assumption is violated, the statistical significance of F ratios obtained using univariate repeated measures ANOVA should be assessed using degrees of freedom that have been adjusted downward. Ans: T Cognitive Domain: Knowledge Answer Location: 14.5: One-way Repeated Measures Results for Heart Rate/Social Stress Data Difficulty Level: Easy 2. Analysts using paired samples t tests to make comparisons among levels of a repeated measures factor should take the increased risk of Type II errors into account. Ans: F Cognitive Domain: Knowledge Answer Location: 14.5: One-way Repeated Measures Results for Heart Rate/Social Stress Data Difficulty Level: Easy 3. If the variances of contrasts analyzed following a one-way repeated measures ANOVA differ significantly, the risk of Type I errors for an F ratio and its corresponding p value may be underestimated. Ans: T Cognitive Domain: Knowledge Answer Location: 14.6: Testing the Sphericity Assumption Difficulty Level: Easy 4. As the number of levels included in a MANOVA increases, the suggested number of participants to be involved in the study decreases. Ans: F Cognitive Domain: Knowledge Answer Location: MANOVA for Repeated Measures Difficulty Level: Easy 5. Follow-up tests to a statistically significant omnibus test in doubly multivariate repeated measures study may still be subject to Type I errors. Ans: T Cognitive Domain: Knowledge Answer Location: 14.9: Doubly Multivariate Repeated Measures

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Difficulty Level: Easy 6. A potential problem with change (or gain) scores in repeated measures is that the pretest score is often negatively correlated with the posttest score. Ans: T Cognitive Domain: Knowledge Answer Location: 14.10.5: Comparison of Repeated Measures ANOVA to Difference Score and ANCOVA Approaches Difficulty Level: Easy 7. ANCOVA eliminates the correlation between pretest and posttest scores. Ans: T Cognitive Domain: Knowledge Answer Location: 14.10.5: Comparison of Repeated Measures ANOVA to Difference Score and ANCOVA Approaches Difficulty Level: Easy 8. A researcher who uses a Latin square design is primarily interested in sphericity violations. Ans: F Cognitive Domain: Knowledge Answer Location: 14.11: Order and Sequence Effects Difficulty Level: Easy 9. Order effects may be of interest in some studies. Ans: T Cognitive Domain: Knowledge Answer Location: 14.13: Second Example: Order Effect is of Interest Difficulty Level: Easy Essay 1. When the sphericity assumption is violated in a one-way repeated measures ANOVA, why is MANOVA a better option? Ans: In MANOVA, each contrast is one of multiple dependent variables, and in MANOVA, dependent variables do not need to have equal variances. Cognitive Domain: Knowledge Answer Location: 14.6: Testing the Sphericity Assumption Difficulty Level: Easy 2. How could a study involving measuring of life satisfaction at three points in time be converted to a mixed model study? Ans: By adding a between-subjects factor like sex, relationship status, or other categorical variable to the design Cognitive Domain: Application

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Answer Location: 14.10: Mixed Model MANOVA: Between-S and Within-S Factors Difficulty Level: Medium 3. How might converting a repeated measures design to a mixed model design be an advantage? Ans: Adding a between-subjects factor might help identify specific participant characteristics that might explain any interaction effects identified. Cognitive Domain: Knowledge Answer Location: 14.10: Mixed Model MANOVA: Between-S and Within-S Factors Difficulty Level: Easy 4. What are two reasons that repeated measures designs may be ineffective? Ans: (1) If any of the treatments have irreversible or long lasting effects or (2) if the effect of a treatment may differ depending on the treatment that preceded it. Cognitive Domain: Knowledge Answer Location: 14.11: Order and Sequence Effects Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 15: Structural Equation Modeling with Amos: A Brief Introduction Test Bank Multiple Choice 1. The general analytic method that evaluates relations among variables based on the information in the variance/covariance matrix is ______. a. regression analysis b. mediation analysis c. path analysis d. structural equation modeling Ans: D Cognitive Domain: Knowledge Answer Location: 15.1: What is Structural Equation Modeling? Difficulty Level: Easy 2 When planning a structural equation analysis, one recommendation about minimum sample size is ______. a. 5–10 observations per estimated parameter b. a minimum sample size of 50 cases c. 5 cases per variable d. a minimum of 25 observations per estimated parameter Ans: A Cognitive Domain: Knowledge Answer Location: 15.6: Screening and Preparing Data for SEM Difficulty Level: Easy 3. Structural equation modeling programs provide confidence intervals obtained using an empirical resampling method known as ______. a. multilevel sampling b. stratified sampling c. bootstrapping d. clustering Ans: C Cognitive Domain: Knowledge Answer Location: 15.8: Specify the Analysis Properties Difficulty Level: Easy 4. Distribution characteristics that include the mean, variance, and covariance are referred to as ______. a. error terms

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 b. moments c. latent variables d. fit indexes Ans: B Cognitive Domain: Knowledge Answer Location: 15.12: Selected SEM Model Terminology Difficulty Level: Easy 5. Compared to the use of the correlation matrix in regression analysis, determining model fit in structural equation modeling is based on the ______. a. ratio of the number latent to observed variables b. sum of the path coefficients c. number of free parameters d. variance/covariance matrix Ans: D Cognitive Domain: Knowledge Answer Location: 15.12: Selected SEM Model Terminology Difficulty Level: Easy 6. A structural equation model in which the model fits the data perfectly is a(n) ______. a. just-identified model b. over-identified model c. under-identified model d. recursive model Ans: A Cognitive Domain: Knowledge Answer Location: 15.12: Selected SEM Model Terminology Difficulty Level: Easy 7. A structural equation model in which the parameters and paths will generally not be sufficient to perfectly reproduce the variance/covariance matrix a(n) ______. a. just-identified model b. over-identified model c. under-identified model d. nested model Ans: B Cognitive Domain: Knowledge Answer Location: 15.12: Selected SEM Model Terminology Difficulty Level: Easy 8. A structural equation model in which the number of pieces of information in the data is smaller than the number of parameters to be estimated is a(n) ______. a. just-identified model b. over-identified model c. under-identified model d. nested model

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Ans: C Cognitive Domain: Knowledge Answer Location: 15.12: Selected SEM Model Terminology Difficulty Level: Easy 9. A structural equation model that contains causal loops is a(n) ______. a. just-identified model b. over-identified model c. under-identified model d. recursive model Ans: D Cognitive Domain: Knowledge Answer Location: 15.12: Selected SEM Model Terminology Difficulty Level: Easy 10. If a structural equation model is a second model obtained by dropping one or more paths from another, the new model is a(n) ______. a. just-identified model b. over-identified model c. under-identified model d. nested model Ans: D Cognitive Domain: Knowledge Answer Location: 15.12: Selected SEM Model Terminology Difficulty Level: Easy 11. When structural equation modeling includes only measurement models for latent variables and does not include causal paths, it is called ______. a. exploratory factor analysis b. confirmatory factor analysis c. logistic regression analysis d. multiple correlation analysis Ans: B Cognitive Domain: Knowledge Answer Location: 15.14: Second Example: Confirmatory Factor Analysis Difficulty Level: Easy

True/False 1. Structural equation modeling is based primarily on information contained in the correlation matrix R. Ans: F Cognitive Domain: Knowledge Answer Location: 15.1: What Is Structural Equation Modeling?

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Difficulty Level: Easy 2. When structural equation modeling is used to test a path model, the structural equation model may include more than one dependent variable. Ans: T Cognitive Domain: Knowledge Answer Location: 15.2: Review of Path Analysis Difficulty Level: Easy 3. Most users of structural equation modeling are interested in just-identified models. Ans: F Cognitive Domain: Knowledge Answer Location: 15.12: Selected SEM Model Terminology Difficulty Level: Easy 4. When chi-square is used as a goodness of fit index in structural equation modeling, nonsignificant values are preferred. Ans: T Cognitive Domain: Knowledge Answer Location: 15.13: SEM Goodness of Fit Indexes Difficulty Level: Easy 5. When using structural equation modeling for confirmatory factor analysis, no causal paths are present. Ans: T Cognitive Domain: Knowledge Answer Location: 15.14: Second Example: Confirmatory Factor Analysis Difficulty Level: Easy 6. In confirmatory factor analysis, the number of factors is specified prior to the analysis. Ans: T Cognitive Domain: Knowledge Answer Location: 15.14: Second Example: Confirmatory Factor Analysis Difficulty Level: Easy 7. Confirmatory factor analysis does allow for oblique factors. Ans: F Cognitive Domain: Knowledge Answer Location: 15.14: Second Example: Confirmatory Factor Analysis Difficulty Level: Easy Essay 1. Why is structural equation modeling preferred over regression analysis? Ans: Regression analysis implicitly assumes that all variables are measured with perfect

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 reliability. Structural equation modeling overcomes this problem by developing measurement models that use multiple observed variables for each construct and shows their relation to latent variables. Cognitive Domain: Comprehension Answer Location: 15.1: What is Structural Equation Modeling? Difficulty Level: Medium 2. What are two advantages of using structural equation modeling instead of ordinary least squares regression or mediation analyses? Ans: Structural equation modeling provides better information about confidence intervals and statistical significance than regression analysis when normality assumptions are violated. Structural equation modeling permits simultaneous analysis of mediation models instead of running separate regression analyses and then combining the results. Cognitive Domain: Knowledge Answer Location: 15.1: What is Structural Equation Modeling? Difficulty Level: Easy 3. What is the difference between a just-identified model, an over-identified model, and an under-identified model? Ans: (1) In a just-identified model, the number of pieces of information in the data is equal to the number of parameters to be estimated. The model fits the data perfectly, so that the variance/covariance matrix can be reproduced perfectly. (2) In an overidentified model, the number of pieces of information in the data is greater than the number of parameters to be estimated. The parameters and paths in the model generally are not sufficient to reproduce the variance/covariance matrix perfectly. (3) In an under-identified model, the number of pieces of information in the data is less than the number of parameters to be estimated. Cognitive Domain: Comprehension Answer Location: 15.12: Selected SEM Model Terminology Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 16: Binary Logistic Regression Test Bank Multiple Choice 1. Binary logistic regression may be useful when the ______. a. outcome variable is normally distributed b. variance of the outcome variable is equal across all levels of the predictor variables c. outcome variable is dichotomous d. participants can be members of both categories of the outcome variable at the same time Ans: C Cognitive Domain: Knowledge Answer Location: 16.1: Research Situations Difficulty Level: Easy 2. Ordinary linear regression is inappropriate when the outcome variable is dichotomous because ______. a. the dichotomous outcome variable requires that probabilities lie between 0 and 1 b. the error scores from the prediction are normally distributed c. the calculated odds ratio values may be negative d. only categorical predictors can be used Ans: A Cognitive Domain: Knowledge Answer Location: 16.3.1: Why Ordinary Linear Regression Is Inadequate When Outcome is Categorical Difficulty Level: Easy 3. If a school system finds that of 100 students who have been suspended, 20 have been suspended a second time during the term, the odds of a student receiving a second suspension are ______. a. 0.20 b. 0.25 c. 4.00 d. 5.00 Ans: B Cognitive Domain: Application Answer Location: 16.4: Definition and Interpretation of Odds Difficulty Level: Medium 4. The inverse of the natural logarithm is the ______. a. logit

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 b. odds ratio c. exponential function d. log-likelihood Ans: C Cognitive Domain: Knowledge Answer Location: 16.4: Definition and Interpretation of Odds Difficulty Level: Easy 5. A logit is defined as the ______. a. base 10 log of the X scores b. natural log of the odds ratio of the outcome c. inverse of the probability that the outcome will be 0 d. base 10 log of the probability that the outcome will be 0 Ans: B Cognitive Domain: Knowledge Answer Location: 16.5: A New Type of Dependent Variable: The Logit Difficulty Level: Easy 6. An advantage of using the logit in binary logistic regression is that ______. a. it represents the proportion of variance explained by the model b. it transforms odds to values that are linearly related to quantitative predictor variables c. its value can only range from 0 to 1 like the dichotomous outcome d. the values are not normally distributed and match the sigmoid curve Ans: B Cognitive Domain: Knowledge Answer Location: 16.5: A New Type of Dependent Variable: The Logit Difficulty Level: Easy 7. The logit (Li) calculated in binary logistic regression by using ______. a. maximum likelihood estimation to obtain the best predictors of outcome values b. ordinary least squares regression involving the predictor variables c. the sums of squares values for the variance explained by the model d. chi-square to determine the odds ratios for membership in each outcome group Ans: A Cognitive Domain: Comprehension Answer Location: 16.6.1: Estimation of Coefficients for a Binary Logistic Regression Model Difficulty Level: Medium 8. The log-likelihood function in binary logistic regression is analogous to the multiple regression construct of ______. a. error variance b. explained variance c. β weight d. sum of squared residuals Ans: D

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Cognitive Domain: Knowledge Answer Location: 16.6.2: Assessment of Overall Goodness of Fit for a Binary Logistic Regression Model Difficulty Level: Easy 9. Multiplying a log likelihood function by -2 makes it comparable to a variable that follows a ______. a. F distribution b. t distribution c. Χ2 distribution d. z distribution Ans: C Cognitive Domain: Knowledge Answer Location: 16.6.2: Assessment of Overall Goodness of Fit for a Binary Logistic Regression Model Difficulty Level: Easy 10. The -2 log-likelihood in binary logistic regression is analogous to the multiple regression measure of ______. a. error variance b. explained variance c. b or β weight d. R2 change Ans: D Cognitive Domain: Knowledge Answer Location: 16.6.2: Assessment of Overall Goodness of Fit for a Binary Logistic Regression Model Difficulty Level: Easy 11. The individual B coefficient for each predictor X in binary logistic regression represents the ______. a. change in Y for a one-unit change in X b. the odds of Y for a one-unit of change in X c. number of units the log odds ratio increases for a one-unit change in X d. the change in X for a one-unit of change in the odds of Y Ans: C Cognitive Domain: Knowledge Answer Location: 16.6.4: Information About Predictive Usefulness of Individual Predictor Variables Difficulty Level: Easy True/False 1. Logistic regression is a special case of the general linear model. Ans: F

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Cognitive Domain: Knowledge Answer Location: 16.1: Research Situations Difficulty Level: Easy 2. The assumptions of binary logistic regression are more restrictive than the restrictions for discriminant analysis. Ans: F Cognitive Domain: Knowledge Answer Location: 16.1.3: Assumptions Required for Linear Regression Versus Binary Linear Regression Difficulty Level: Easy 3. Like chi-square analysis, binary logistic regression requires a minimum of 5 for cell frequency in the outcome category. Ans: T Cognitive Domain: Knowledge Answer Location: 16.1.3: Assumptions Required for Linear Regression Versus Binary Linear Regression Difficulty Level: Easy 4. An of odds ratios over probabilities is that odds ratios have a fixed upper limit. Ans: F Cognitive Domain: Knowledge Answer Location: 16.3.2: Modifying the Method of Analysis to Handle a Binary Categorical Outcome Difficulty Level: Easy 5. Logits have a direct intuitive interpretation. Ans: F Cognitive Domain: Knowledge Answer Location: 16.5: A New Type of Dependent Variable: The Logit Difficulty Level: Easy 6. A null model generates a predicted odds value for all participants in a study. Ans: T Cognitive Domain: Knowledge Answer Location: 16.5: A New Type of Dependent Variable: The Logit Difficulty Level: Easy 7. Larger absolute values of the log-likelihood function indicate better model fit. Ans: F Cognitive Domain: Knowledge Answer Location: 16.6.2: Assessment of Overall Goodness of Fit for a Binary Logistic Regression Model Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 8. Results of binary logistic regression analyses are more easily interpreted when members of the group with a more negative outcome are coded as 1. Ans: T Cognitive Domain: Knowledge Answer Location: 16.8.3: Coding Scores on Binary Variables Difficulty Level: Easy 9. An advantage of using maximum likelihood estimation for deriving the coefficients in logistic regression is that the values are relatively stable even in small samples. Ans: F Cognitive Domain: Knowledge Answer Location: 16.11: Comparison of Discriminant Analysis to Binary Logistic Regression Difficulty Level: Easy Essay 1. List three reasons why ordinary linear regression is inadequate when outcome is categorical. Ans: Linear regression: (1) could yield estimated probabilities are negative or greater than 1, (2) the probability of being in the target group may be nonlinear, and (3) the probability of accurately predicting membership in the target group is better at the extremes. Cognitive Domain: Comprehension Answer Location: 16.3.1: Why Ordinary Linear Regression Is Inadequate When Outcome is Categorical Difficulty Level: Medium 2. What are three advantages of using logits as outcome variables in binary linear regression? Ans: Logits have (1) fixed upper or lower limit, (2) values that tend to be normally distributed, and (3) values that are linearly related to quantitative predictor variables in many research situations Cognitive Domain: Knowledge Answer Location: 16.5: A New Type of Dependent Variable: The Logit Difficulty Level: Easy 3. How is the log-likelihood (LL) function used in evaluating logistic regression models? Ans: The LL is similar to the residual SS in multiple regression. Therefore, higher values indicate poorer model fit while lower values represent better fit, with fit meaning how well the predicted outcomes of the model match the actual group membership. Because the values of LL will always be negative, multiplying by -2 results in a chi-square distribution. The LL for the null model includes only the intercept and represents the maximum amount of error. Each time a predictor variable is entered into the model, a new LL is calculated and the resulting difference (LLnew – LLnull) is the chi-square value.

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 A significant value indicates that the added predictors have resulted in a better fitting model. Cognitive Domain: Comprehension Answer Location: 16.6.2: Assessment of Overall Goodness of Fit for a Binary Logistic Regression Model Difficulty Level: Medium 4. Explain how exp (B) is interpreted for values less than 1.0, equal to 1.0, and greater than 1.0. Ans: For each one-unit change in predictor X, the predicted odds of Y will change by eB units. When exp(B) is less than 1.0, the odds of membership in the targeted group (1) will decrease as X increases. When exp(B) equals 1.0, the odds of membership does not change as X increases. When exp(B) is greater than 1.0, the odds of membership in the targeted group increases as X increases. Cognitive Domain: Understanding Answer Location: 16.6.4: Information About Predictive Usefulness of Individual Predictor Variables Difficulty Level: Medium 5. Define the minimal recommended sample and cell sizes recommended for binary logistic regression. Ans: At least 10 times as many cases as predictor variables few cells with expected frequencies less than 5. Cognitive Domain: Knowledge Answer Location: 16.8.2: Design Decisions Difficulty Level: Easy

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021

Chapter 17: Additional Statistical Techniques Test Bank Multiple Choice 1. If a clinical psychologist placed each client into one of three categories based on their scores on an optimism scale at the beginning of therapy and tracked the number of clients in each category who completed the therapy program, the psychologist was using ______. a. cluster analysis b. factor analysis c. survival analysis d. discriminant analysis Ans: C Cognitive Domain: Application Answer Location: 17.3: Survival Analysis Difficulty Level: Medium 2 If a school system assesses each student’s cognitive ability, reading achievement, and math achievement at the beginning the school year and places them in different groups according to their scores so that different programs can be designed for different types of students, the school system is using ______. a. cluster analysis b. factor analysis c. survival analysis d. discriminant analysis Ans: A Cognitive Domain: Application Answer Location: 17.4: Cluster Analysis Difficulty Level: Medium 3. If a university student counseling center interested in determining the day of the week on which students come to the center in person to request services recorded all such visits daily for two months, the center was using ______. a. cluster analysis b. time series analysis c. survival analysis d. discriminant analysis Ans: B Cognitive Domain: Application Answer Location: 17.5: Time Series Analysis Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 4. If a time series analysis reveals an exponential decline in the values of correlations across lags, the model can be described as a(n) ______. a. autoregressive model b. integrated model c. moving average model d. cyclic model Ans: A Cognitive Domain: Knowledge Answer Location: 17.5.1: Describing a Single Time Series Difficulty Level: Easy 5. If a time series analysis reveals today’s measure is equal to yesterday’s measure plus some amount of random error, the model can be described as a(n) ______. a. autoregressive model b. integrated model c. moving average model d. cyclic model Ans: B Cognitive Domain: Knowledge Answer Location: 17.5.1: Describing a Single Time Series Difficulty Level: Easy 6. If a time series analysis predicts today’s observation from one or more lagged residuals from previous days, the model can be described as a(n) ______. a. autoregressive model b. integrated model c. moving average model d. cyclic model Ans: C Cognitive Domain: Knowledge Answer Location: 17.5.1: Describing a Single Time Series Difficulty Level: Easy 7. In order to predict the probability that college students will continue to use online counseling services that has never been used by 70% of students, but that has been used by 10% more than 5 times, the most appropriate analysis researcher could use would be ______. a. logistic regression b. Poisson regression c. bivariate regression d. multiple regression Ans: B Cognitive Domain: Application Answer Location: 17.6: Poisson and Binomial Regression for Zero-Inflated Count Data Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 8. If a questionnaire in a popular magazine indicates that a person has a depressive disorder, but further examination and testing by a clinical psychologist indicates that the person does not have such a disorder, the initial result is a ______. a. true positive b. false positive c. true negative d. false negative Ans: B Cognitive Domain: Comprehension Answer Location: 17.7: Bayes’ Theorem Difficulty Level: Medium 9. If an external examination of a person’s knee leads to a diagnosis of overexertion, but an MRI exam reveals a meniscal tear, the initial result is a ______. a. true positive b. false positive c. true negative d. false negative Ans: D Cognitive Domain: Comprehension Answer Location: 17.7: Bayes’ Theorem Difficulty Level: Medium 10. The percentage of patients who are tested for Alzheimer’s disease who have Alzheimer’s disease is the ______. a. effect size b. hazard ratio c. risk ratio d. base rate Ans: D Cognitive Domain: Comprehension Answer Location: 17.7: Bayes’ Theorem Difficulty Level: Medium 11. Following a study designed to assess the remission rates at two hospitals that included 10 patients seen by five doctors in each of five departments at each hospital, an appropriate analysis of the data collected could be ______. a. multilevel modeling b. factor analysis c. logistic regression d. analysis of variance for repeated measures Ans: A Cognitive Domain: Comprehension Answer Location: 17.8: Multilevel Modeling Difficulty Level: Medium

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 True/False 1. Observations in time series analyses are uncorrelated. Ans: F Cognitive Domain: Knowledge Answer Location: 17.5: Time Series Analysis Difficulty Level: Easy 2. Binomial regression is appropriate when a dataset includes a large number of zero responses. Ans: T Cognitive Domain: Knowledge Answer Location: 17.6: Poisson and Binomial Regression for Zero-Inflated Count Data Difficulty Level: Easy 3. If a questionnaire in a popular magazine indicates that a person has a depressive disorder, and further examination and testing by a clinical psychologist indicates that the person does have such a disorder, the initial result is a true positive. Ans: T Cognitive Domain: Comprehension Answer Location: 17.7: Bayes’ Theorem Difficulty Level: Medium 4. Bayesian analyses are based on frequentist methods. Ans: F Cognitive Domain: Knowledge Answer Location: 17.7: Bayes’ Theorem Difficulty Level: Easy 5. Multilevel is useful in studies involving nested units or cases. Ans: T Cognitive Domain: Knowledge Answer Location: 17.8: Multilevel Modeling Difficulty Level: Easy Essay 1. Using the concept of ‘mood on awakening’, describe the concept of ‘lag one autocorrelation’ in time series analysis and its use. Ans: Lag refers to the correlation between mood measured one day with mood measured on the previous day, it indicates how well today’s mood can be predicted from yesterday’s mood. Cognitive Domain: Comprehension

Instructor Resource Warner, Applied Statistics II, 3e SAGE Publishing, 2021 Answer Location: 17.5.1: Describing a Single Time Series Difficulty Level: Medium 2. Why is an interrupted time series analysis superior to a simple pretest-posttest design? Ans: A single pretest measure may not establish a stable baseline and researchers cannot know whether the dependent variable may have been changing even in the absence of the intervention. A single posttest measure does not indicate the duration of the intervention effect, or a temporary intervention effect that occurred at an earlier or later time may be unobserved. With only one pretest and one posttest measure, researchers cannot determine whether the observed effect is a continuing trend that was not affected by the intervention or other temporal processes like maturation or fatigue. Cognitive Domain: Comprehension Answer Location: 17.5.2: Interrupted Time Series: Evaluating Intervention Impact Difficulty Level: Medium 3. What are the principles underlying Bayes’ Theorem? Ans: (1) Prior estimates of the probability of events can be revised and updated as new evidence becomes available, and (2) the prior evidence should continue to be taken into account as the new evidence becomes available. Cognitive Domain: Knowledge Answer Location: 17.7: Bayes’ Theorem Difficulty Level: Easy