Blog Archive The General Linear Model, Analysis of Covariance, a...

http://www.analysisfactor.com/statchat/the-general-linear-model-analys...

We are Making our Online Statistics Workshops Even Better! Using Adjusted Means to Interpret Moderators in Analysis of Covariance

The General Linear Model, Analysis of Covariance, and How ANOVA and Linear Regression Really are the Same Model Wearing Different Clothes
Just recently, a client got some feedback from a committee member that the Analysis of Covariance (ANCOVA) model she ran did not meet all the assumptions. Specifically, the assumption in question is that the covariate to be uncorrelated to the independent variable. This committee member is, in the strictest sense, correct. Analysis of Covariance was developed for experimental situations in which the independent variables are categorical and usually manipulated, not observed. The covariatecontinuous and observedis considered a nuisance variable. There are no research questions about how this covariate itself affects the dependent variable. The only hypothesis tests of interest are about the independent variables, controlling for the effects of the nuisance covariate. A typical example would be to compare the math scores of students who were enrolled in three different learning programs at the end of the school year. The only research question would be about whether the math scores differed on average among the three programs. It would be useful to control for a covariate like IQ scores, but we are not really interested in the relationship between IQ and math scores. But thats really just one application of a linear model with one categorical and one continuous predictor. The research question of interest doesnt have to be about the categorical predictor, and the covariate doesnt have to be a nuisance variable. A regression model with one continuous and one dummy variable is the same model (actually, youd need two dummy variables to cover the three categories, but thats another story). The focus of that model may differperhaps the main research question is about the continuous predictor. But its the same model. And your software will run it the same way. YOU may focus on different parts of the output or select different options, but its the same model. And thats where the model names can get in the way of understanding the relationships among your variables. The model itself doesnt care if the categorical variable was manipulated. It doesnt care if the categorical independent variable and her continuous covariate are mildly correlated. If those ANCOVA assumptions arent met, it does not change the analysis at all. It only affects how parameter estimates are interpreted and the kinds of conclusions you can draw. In fact, those assumptions really arent about the model. Theyre about the design. Its the design that affects the conclusions. It doesnt matter if a covariate is a nuisance variable or an interesting phenomenon to the model. Thats a design issue. So what do you do instead of labeling models? Just call them a General Linear Model. Its hard to think of regression and ANOVA as the same model because the equations look so different. But it turns out they arent.

If you look at the two models, first you may notice some similarities. Both are modeling Y, an outcome. Both have a fixed portion on the right with some parameters to estimatethis portion estimates the mean values of Y at the different values of X.

1 de 4

16/10/2010 02:16 a.m.

 Blog Archive The General Linear Model, Analysis of Covariance, a...

http://www.analysisfactor.com/statchat/the-general-linear-model-analys...

Both equations have a residual, which is the random part of the modelthe variation in Y that is not affected by the Xs. But wait a minute, Karen, are you nut?there are no Xs in the ANOVA model! Actually, there are. Theyre just implicit. Since the Xs are categorical, they have only a few values, to indicate which category a case is in. Those j and k subscripts? Theyre really just indicating the values of X. (And for the record, I think a couple Xs are a lot easier to keep track of than all those subscripts. Ever have to calculate an ANOVA model by hand? Just sayin.) So instead of trying to come up with the right label for a model, focus instead on understanding (and describing in your paper) the measurement scales of your variables, if and how much theyre related, and how that affects the conclusions.

Tags: ANOVA, General Linear Model, Linear Regression

This entry was posted on Friday, September 17th, 2010 at 11:24 am and is filed under ANOVA, Linear Regression. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

Leave a Reply
Name Mail (will not be published) Website

Add Me to Your Home Page

Enter your email address to receive automatic notice of new posts:

Search for:

Survival Analysis - An Introduction to Time-to-Event Data

2 de 4

16/10/2010 02:16 a.m.

 Blog Archive The General Linear Model, Analysis of Covariance, a...

http://www.analysisfactor.com/statchat/the-general-linear-model-analys...

Assumptions of the General Linear Model and How to Check Them

Visit my web page:

Subscribe! Monthly Tips, Resources, and News. Plus our 10-page report of The Top Resources for Learning 13 Statistical Methods. FREE! Your First Name: Your Email:

We hate spam as much as you do. We will never share your email address with anyone. Never.

Data Analysis with SPSS (3rd Edition) by Stephen Sweet and Karen Grace-Martin

Authors
Karen Grace-Martin

Categories
ANOVA (39) Confusing Statistical Terms (5) Factor Analysis (5) General (35) Missing Data (37) Mixed and Multilevel Models (26) Other regression models (86) Linear Regression (60) Logistic Regression (35) Poisson and Negative Binomial Regression Models (15) Power and Sample Size (5) Statistical Consulting (2)

3 de 4

16/10/2010 02:16 a.m.

 Blog Archive The General Linear Model, Analysis of Covariance, a...

http://www.analysisfactor.com/statchat/the-general-linear-model-analys...

Statistical Resources (32) Statistical Software (58) R (2) SAS (23) SPSS (47) Stata (6) Statistics Workshops (18) The Analysis Factor Monthly Teleseminars (18) Uncategorized (48)

Add me as your friend in Facebook is proudly powered by WordPress Entries (RSS) and Comments (RSS).

4 de 4

16/10/2010 02:16 a.m.