There are one dependent variable and five dependent variables in the model. Responses are collected for both dependent and independent variable using five point Likert scale. Can hierarchical regression analysis be performed ? Please Respond
I don't see why you would want to do it here. Note that with a Likert DV(ordinal) you should be thinking about Ordinal logistic Regression. See the link:
https://www.google.com/search?client=firefox-b-1-d&q=Likert+DV+ordinal+logistic+regression
While you can enter variables any way you wish, the usual hierarchical stats are pretty meaningless away from least squares. Best, David Booth
The main assumptions of hierarchical linear regression are the same as for other
forms of regression analyses. This includes that multicollinearity does not exist or is only present at very low levels (Tabachnick & Fidell, 2013); this can be
assessed using the Durbin–Watson test for independence of residuals (Field,
2013). Researchers should also check plots of standardized residuals to evaluate
for assumptions of normality, homoscedasticity, linearity, independence of errors, and absence of outliers.
Thanks to David Eugene Booth for mentioning the appropriateness of multiple ordinal regression and Imran Anwar for clarifying on regression assumptions.
Hierarchical regression is more appropriate for model comparison for nested data when the researcher needs to account (or control) for the effect of certain IVs on the DV, before the impact of other IVs on the DV is measured. The traditional multiple linear regression presumes that all cases are independent of each other, however, in the case of nested data, the researcher determines the order of entering variables into the regression equation. For example, if x1, x2, x3, x4 and x5 are IVs and y is the DV, and the researcher needs to measure the impact of x4 and x5 on y after controlling for the effect of x1, x2 and x3, then hierarchical multiple regression would be the most appropriate method. From your dataset, if you need to compare alternative models after controlling for a-priori identified variables (from literature review), effect of which need to be controlled before the rest of IVs are entered to measure the impact on DV, then hierarchical multiple regression would be needed.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey