Quiz (Chapter 15)


The Analysis of Covariance

Introduction

  1. The analysis of covariance is often referred to by means of what abbreviation?
  2. (T/F) The analysis of covariance can't be used in studies involving factors that are within-subjects in nature.
  3. There ___ (is/isn't) an ANCOVA counterpart to any ANOVA.

The Three Different Variables Involved in Any ANCOVA Study
  1. ANCOVA studies contain at least one _____ variable in addition to the independent and dependent variables.
  2. The variable identified in question 4 is more similar to the study's _____ (independent/dependent) variable.
  3. The covariate variable is sometimes referred to as the _____ variable.

The Covariate's Role
  1. If a good covariate is used, the ANCOVA will have ___ (more/less) power than will its ANOVA counterpart.
  2. A good covariate functions to ____ (increase/reduce) the size of "error variance."
  3. While one function of the covariate is connected to "power," the other function is related to "_____."
  4. In Excerpt 15.4, the "control" function of the ANOVA adjusted the MET and SC means of the _____ (baseline/post-treatment) level of smoking .
  5. Will the group with the highest covariate mean have its mean on the dependent variable adjusted up or down?
  6. (T/F) Researchers must choose to use a covariate either to increase power or to use it to exercise control.

Null Hypotheses
  1. How many null hypotheses were associated with the ANCOVAs in Excerpts 15.1? What about Excerpt 15.2?
  2. The ANCOVA null hypothesis always deal with adjusted ____ (sample/population) means on which variable?

The Focus, Number, and Quality of the Covariate Variable(s)
  1. In ANCOVA studies, the covariate variable ___ (can/can't) be different from the dependent variable.
  2. (T/F) Excerpt 15.5 is typical of ANCOVA studies in that only one covariate variable was involved (rather than multiple covariate variables).
  3. The covariate variable ____ (should be/need not be) conceptually related to the dependent variable.

Presentation of Results
  1. If a summary table is built to show the results of Excerpt 15.3's ANCOVA, how many F-values would it contain?
  2. If a summary table is built for Excerpt 15.7's ANCOVA, how many Total df would there be?
  3. Suppose ANCOVA is used in a study on thin people to compare two different diets (A and B) on weight gain over a 4-week period. Suppose further that pre-diet weight is the covariate while post-diet weight is the dependent variable. Finally, suppose that the pre and post means (in pounds) for people in diet A are 140 and 160, respectively, while the pre and post means for people in diet B are 135 and 150. In this study, ANCOVA would cause diet A's adjusted mean to be ____ (higher/lower) than its unadjusted mean.
  4. In Excerpt 15.14, the post hoc investigation focused on _____ (cell/main effrect) means.

The Statistical Basis for ANCOVAšs Power Advantage and Adjustment Feature
  1. In ANCOVA studies, a covariate variable will do a better job of increasing power and providing control if it has a strong relationship with the _____ (independent/dependent) variable.
  2. (T/F) In evaluating the relationship alluded to in Question 22, the researcher should focus on the pooled within-group correlation rather than the correlation based on all comparison groups thrown together.
  3. When multiple covariate variables are involved in an ANCOVA study, those covariate variables work best if they have a _____ (strong/weak) relationship with one another.

Assumptions
  1. The regular assumptions of ANOVA ___ (do/don't) apply to ANCOVA studies.
  2. One unique-to-ANCOVA assumption says that the _____ variable should not affect the _____ variable.
  3. The assumption mentioned in Question 26 was ____ (met/violated) in Excerpt 15.12.
  4. (T/F) With equal sample sizes, ANCOVA is robust to the assumption of equal regression slopes.
  5. (T/F) If the sample sizes are equal, ANCOVA is robust to the assumption of linearity.
  6. You are likely to come across ___ (many/few) ANCOVA studies in which assumptions are not discussed.

ANCOVA When Comparison Groups Are Not Formed Randomly
  1. Are the 2 groups in Excerpt 15.8 "intact groups"? What about the 2 groups in Excerpt 15.15?
  2. When the population means on the covariate differ, ANCOVA's sample-based adjusted means on the dependent variable ____ (will/won't) be biased.
  3. Simply on logical grounds, ANCOVA _____ (should/shouldn't) be viewed as able to equate intact groups.

Related Issues
  1. The Bonferroni adjustment technique ___ (can/can't) be used in ANCOVA studies.
  2. In Excerpt 15.13, two values of eta squared are provided. Based on the criteria for evaluating these indices, the size of the significant Treatment effect (when the dependent variable was "solitary play") would be classified as being
    1. small
    2. medium
    3. large
  3. (T/F) With ANCOVA, only omnibus F-tests and post hoc comparisons are possible, not planned comparisons.

A Few Warnings
  1. Your critique of an ANCOVA article should go ____ (up/down) if it doesn't contain adjusted means.
  2. In testing assumptions, researchers should use an alpha level that's more _____ (lenient/rigorous) than .05.
  3. Which of these studies would yield results that are easier to interpret: (a) a one-way ANCOVA used with intact groups or (b) a one-way ANOVA used with randomly formed groups?


Click here for answers.

 

Copyright © 2012

Schuyler W. Huck
All rights reserved.

| Book Info | Author Info |

Site URL: www.readingstats.com

Top | Site Map
Site Design: John W. Taylor V