e-Articles (Chapter 16)
Here are some full-length research articles that illustrate the use of bivariate, multiple, and logistoc regression. To view any article, simply click on its title. (NOTE: No claim is made that these articles are perfect in all respects. By carefully reviewing them, you will hone your skills at being able both to decipher and to critique statistically-based research reports.)
A simultaneous multiple regression was used to investigate which of several characteristics of tutors explained why some tutors were evaluated more positively than others by students. See the last section of the report's "Method" section, the section of "Results" entitled "The relationship between student perceptions of tutor performance and tutorsí background in PBL," and Table 5.
Illustrates the use of hierarchical multiple regression. Table 2 beautifully displays the results at each stage of the multiple regression (with these "stages" referred to as "models" by the researchers). This table shows which independent variables (IVs) were involved at each stage, as well as each IV's beta weight and its associated p-level. This same table shows, for each stage of the hierarchical multiple regression, how well the set of IVs worked as predictors. This was done by showing, for each "model," the size of the F-ratio (and its associated p-level), R2, DR2, and an F-ratio to assess the increment in explained variability.
In this study, two multiple regression models were compared, each having the same dependent variable: percentage of 9th-graders who were overweight. One model had just 3 independent variables: the proximities of fast food restaurants, convenience stores, and supermarkets. The second model included those 3 indendent variables along with several demographic and socioeconomic variables. The differing results illustrate nicely the point that regression results are "context dependent" and can vary dramatically depending on what is put into or left out of the model. See Tables 3 and 4.
This study illustrates the use of a hierarchical multiple regression to identify potential moderator variables. The authors of this article explain nicely the logic and procedures of this kind of regression analysis. See sections 2.3 ("Data Analysis"), 3.2 ("Hierarchical Multiple Regression Analysis"), 3.3 ("Tests for Moderation"), and 5 ("Conlusion").
A logistic regression was used in this study to identify predictors of mental distress. Table 4 presents the Odds Ratio for each predictor, along with 95% confidence intervals and p-values.
Illustrates the use of logistic regression to estimate odds ratios. Read the next-to-last sentence of the "Results" section of the Abstract, the next-to-last sentence in the "Method" section, and the final paragraph of the main "Results" section. Also take a look at the middle column of numbers in Table 2.
Illustrates the computation of Nagelkerke's R2 for a logistic regression model. See the 10th sentence in the first paragraph of the article's "Results" section.
Copyright © 2012
Schuyler W. Huck