Home > Statistics Every Writer Should Know > The Stats Board > Discusssion
MANCOVA
Message posted by Holly on October 15, 1999 at 12:00 AM (ET)
What is an ANCOV/MANCOVA and what specific question will this answer?
Thanks!
READERS RESPOND:
(In chronological order. Most recent at the bottom.)
Re: MANCOVA
Message posted by Bill on October 15, 1999 at 12:00 AM (ET)
Holly, ANOVA is analysis of variance and MANOVA is multivariate analysis of variance. ANOVA is used when there is one dependent variable and MANOVA is used when there are more than one dependent variables. There are many forms of both procedures but each tests for a "relationship" between a nominal independent variable e.g., Male vs Female, Drug A vs Drug B vs Drug C and a interval scale dependent variable, eg. Weight. So, we might test for a relationship between the type of diet drug used and the amount of weight lost (did one drug produce more weight loss that the other drug(s)). This is the short answer to your question. The long answer is usually covered in 2-3 chapters of a statistical text. You may get a better response if you focused your question.
Re: MANCOVA
Message posted by Bill on October 18, 1999 at 12:00 AM (ET)
Opps, you said ANCOVA, not ANOVA. ANCOVA (MANCOVA) is a statistical procedure used to control for a confounding variable. Confounding variables most often occur when subjects can not be randomly assigned to groups. Suppose you analyzed ventricular wall motion abnormalities in smokers and nonsmokers. You can not assign subjects to the smoker and nonsmoker group. You could use a t-test (or one-way ANOVA) to determine if the number of abnormalities are different between the two groups. However, researchers know that wall motion abnormalities are related to coronary stenosis and that smokers generally have more stenosis. So, any differences in abnormalities may be due to stenosis. We want to remove (partial out) the variance in the number of abnormalities attributed to stenosis. After this variance has been removed then we can test for the relationship between smoking and abnormalities while adjusting or controlling for stenosis. Number of abnormalities is the dependent variable, smoke vs nonsmoke the independent variable, and stenosis is the covariate. See Basic and Clinical Biostatistics. Dawson-Saundrs, B. Lange Books, 1990.
Your $5 contribution helps cover part the $500 annual cost of keeping this site online.
