It has been suggested that I analyze my data with an ANCOVA instead, and have ethnicity as my IV and political skill as an additional co-variate (rather than doing different regressions, have one data set). The ANCOVA is useful - From the output, i can see which covariate is more related to the outcome variable and which explains the bigger proportion of variance.But from the output that I have, I can not see how ethnicity correlates or is associated with any of the co-variates (which is important for assessing my hypothesis). I want to know if there are differences in political skill, perceptions of organizational and informal networks between the ethnicities. Is this possible to do with ANCOVA? If so what am I supposed to be looking at to interpret my data? I would be grateful for any guidance. Thank you.
ANCOVA is also difficult in that if any assumption is not met, the analysis is not useful. I would say to you, as I did someone else with MANOVA issues, to first check your assumptions.
I highly recommend that you purchase books which give you exerciese to learn these procedures. You can get used copies. You may need more than one of such books, but they are available from booksellers.
Dear Ndade,
you can not do this in ANCOVA and reccomending your professor to conduct ANCOVA, NOT mean that you can not use any other analysis before your main analysis. If you want to see this simply, conduct a Pearson correlation first to give information about your covariates and etnicity assocation. let political skill, perceptions of organizational and informal networks your dependent variables separetely and you coded your etnicity variable 1: white, 2: black. IF correlation significant and negative, whites significaly higher politikal skills than blacks. If correlation positive and significant, blackssignificaly higher politikal skills than whites. intepretation all the same. ıf correlation not significant, there are no difference. Also squaring your correlation value give information about explained variance to you. Contrary to Zinn, ANCOVA is robust to violation of some assumptions. Most important assumption is that homogenity of regression slopes, its violation may negatively affects your results. You can watch videos of Todd Grande in youtube.
It sounds like political skill and perceptions of networks are not covariates, but are additional dependent variables. So you can do separate tests comparing the ethnicities on each of these variables, or you can use MANOVA.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA