Next Upcoming Google+ Hangout: Tuesday, August 27 @ 7PM (CST) - To Participate CLICK HERE
Search For Topics/Content
Top
Discussion Forum Topics
Find a Blog Entry
- Confusing Stats Terms Explained: Internal Consistency
- R Is Not So Hard! A Tutorial, Part 4 (repost)
- Confusing Stats Terms Explained: Heteroscedasticity (Heteroskedasticity)
- How to Create APA Style Graphs and Then Teach SPSS to Do it Automatically!
- Wonderful "How-To" Resources for Learning Structural Equation Modeling (SEM) with AMOS
- Interpreting the Intercept in a Regression Model (repost)
- Why You Shouldn't Conclude "No Effect" from Statistically Insignificant Slopes (repost)
- To syndicate or not to syndicate, that is the question...
- Please Vote on the "Top Confusing Stats Terms"
- How to Conduct a Repeated Measures MANCOVA in SPSS
- The Worst Mistake Made on a Dissertation Is...
- Moderating Effects with Seemingly Uncorrelated Variables
- Using Syntax to Assign 'Variable Labels' and 'Value Labels' in SPSS
- How to make SPSS produce all tables in APA format automatically!
- Confusing Stats Terms Explained: Residual
- Confusing Stats Terms Explained: Multicollinearity
- Confusing Stats Terms Explained: Standard Deviation
- Top Ten Tips for Data Analysis to Make Your Research Life Easier!
- Within-Subject and Between-Subject Effects: Wanting Ice Cream Today, Tomorrow, and The Next Day…
- Bonferroni Correction In Regression: Fun To Say, Important To Do...
- Put aside your fears and be wrong already!
- Trade Your Stats "Truths" for Stats Arguments...
- Data Transformations: statistical voodoo or truth serum for your data?
Is it possible to conduct a MANCOVA with a dummy coded DV in the variate? I have 5 DVs, and 2 of them are extremely skewed (# suspensions and # schools attended in one school year—as you can imagine, the norm is 0 and 1 respectively). I am using a 1/x transformation, but that makes interpretation so much harder! Tabachnik speaks of trying sqrt, LG10, and 1/x transformations, and if all else fails, dichotomize. I know a dummy coded variable works as a predictor in linear regression, and MANCOVA is just sorta doing that backwards, but when I google it, I’m seeing people saying that using a binary variable violates the assumptions of normal distribution and equal variances. So I think I’m convinced I have to transform... even though the distribution is still skewed, it’s dramatically better. I have an IV with three categories, and 5 covariates. The n for the 3 groups is 201, 250 and 363 respectively. Thanks in advance for your advice!