Week 5 - Assignment: Identify Analysis Tools in Published Research

SAGE Research Methods Video

ANCOVA

Video Title: ANCOVA

Originally Published: 2016

Publication Date: Sep. 30, 2016

Publishing Company: SAGE Publications, Inc.

City: Thousand Oaks, United States

ISBN: 9781506359168

DOI: https://dx.doi.org/10.4135/9781506359168

(c) SAGE Publications Inc., 2017

HERSCHEL KNAPP: Welcome to practical statistics for nursing using SPSS. This video shows how to process the ANCOVA test. You can watch the entire video or use the time slider to navigate directly to any time point.

HERSCHEL KNAPP [continued]: The ANCOVA test is similar to ANOVA test. Before proceeding, it's recommended that you first view the video Ch 06 - ANOVA.mp4. In terms of setup and results, the ANCOVA test and the ANOVA test are quite similar.

HERSCHEL KNAPP [continued]: Remember that the ANOVA test assesses three or more groups to detect statistically significant differences between the pairs of groups, specifically the results of ANOVA test are based on the effect that the independent variable, the anti-hypertensive drug, had on the dependent variable, systolic blood pressure.

HERSCHEL KNAPP [continued]: The ANCOVA statistic allows us to include a potentially confounding variable into the model, which we expect may influence the dependent variable, such as smoking rate. The ANCOVA test then adjusts the results in the dependent variable accordingly.

HERSCHEL KNAPP [continued]: The ANCOVA test has two pretest criteria, homogeneity of regression slopes and homogeneity of variance, Levene's test. We'll check for homogeneity of regression slopes now. We'll see the results of the homogeneity of variance test when we run the ANOVA test. This example uses the dataset Ch 07 - Example 01 -ANCOVA.sav.

HERSCHEL KNAPP [continued]: This dataset contains three variables. Group is a categorical variable containing three values, drug A, drug B, and drug C. Systolic BP is a continuous variable that contains the systolic blood pressure of each participant at the conclusion of the study.

HERSCHEL KNAPP [continued]: And smoking is a continuous variable that contains the mean number of cigarettes that each participant smoked on a daily basis. To check for the homogeneity of regression slopes, click on Analyze, General Linear Model, Univariate. Move Group to Fixed Factors.

HERSCHEL KNAPP [continued]: Move Systolic BP to Dependent Variable. And move Smoking to Covariance. Click on Model, select Custom, move Group and Smoking to Model. Next, hold down the Shift key, and click on Group and Smoking.

HERSCHEL KNAPP [continued]: This will select both Group and Smoking together to signify the interaction term. And move them to Model. Click Continue, click OK, and it'll process. We look to the Test Between Subject Effects Table. If the P value for the group smoking interaction term

HERSCHEL KNAPP [continued]: is greater than 0.05, then this would indicate that there is no statistically significant difference in the regression slopes among the variables involved in this model, and the assumption of homogeneity of regression slopes would be satisfied. In this case, the P value is 0.028.

HERSCHEL KNAPP [continued]: Since this is less than or equal to 0.05, this indicates that there is a statistically significant difference between the regression slopes for systolic BP and smoking. This violation makes sense, as the covariate is somewhat atypical. The covariate in this model is smoking,

HERSCHEL KNAPP [continued]: the number of cigarettes that each participant smokes in a typical day. Since more than 90% of the participants are nonsmokers, this finding is not unexpected. Since this pretest criteria is not fully satisfied, this should be noted in the results section.

HERSCHEL KNAPP [continued]: To run the ANCOVA test, click on Analyze, General Linear Model, Univariate. Click on Model, select Full Factorial, click Continue, click Options, move Group to Display

SAGE (c) SAGE Publications Inc., 2017

SAGE Research Methods Video

Page 2 of 3 ANCOVA

Means 4,

HERSCHEL KNAPP [continued]: check the Compare Main Effects checkbox. In the Confidence Interval Adjustment pull down menu, select Bonferroni. In the Display Options, check Homogeneity Tests. Click Continue, click OK, and it'll process.

HERSCHEL KNAPP [continued]: To finalize the pretest checklist, we see that the homogeneity of variance test produced a P value of 0.791. Since this is greater than 0.05, this indicates that there is no statistically significant difference between the variances, hence this criteria is satisfied. Next, we look to the Test of Between Subjects Effects Table.

HERSCHEL KNAPP [continued]: The P value of 0.004 indicates that a statistically significant difference has been detected among the adjusted means for the groups. In the Estimates Table, we see the adjusted means for each group. These means have been adjusted to account for the smoking covariate.

HERSCHEL KNAPP [continued]: These figures will be useful when documenting the results. Finally, we look to the Pairwise Comparisons Table. This table is read in the same way as the Multiple Comparisons table produced by the ANCOVA post-hoc test. To identify the pairs of groups that are statistically significantly different from each other,

HERSCHEL KNAPP [continued]: we look for P values that are less than or equal to 0.05. We see that there is a statistically significant difference in the adjusted means between drug A and drug B and between drug A and drug C. This concludes this video.

SAGE (c) SAGE Publications Inc., 2017

SAGE Research Methods Video

Page 3 of 3 ANCOVA