ANCOVA

[deleted_user]

I am not very statistically inclined, so please forgive my ignorance. I am trying to run an ANCOVA in EG 4.2. I cannot find any reference to where in any procedure one would add a covariate. I have done these analyses in Statistica before and were quite simple and intuitive to set up.

I suspect I'm looking for the wrong term or am in the wrong procedure, but I'm also surprised that it hasn't revealed itself to me more immediately.

Any help or thoughts would be greatly appreciated. Thanks!

[TheChadd]

proc reg will do ANCOVA for you.
example:
dependent variable: z
independent categorical variable: x
independent continuous variable: y

proc REG;
model z= x y;

This assumes equal slopes. You can modify this to have unequal slopes, too.

Cheers,
Chadd

[deleted_user]

Thank you for the response. However, I am using Enterprise Guide, and really have no idea on how to input code. Do you have any ideas on how to "point and click" my way through setting up an ANCOVA in EG?

[TheChadd]

Honestly, I gave up on EG after playing around with it for a month or so :(. I found it easier to just do my analyses in SAS or JMP. I'm not sure if EG has a pre-coded solution for ANCOVA.

[Doc_Duke]

ANCOVA is a type of linear model. So, go to
Tasks --> ANOVA --> Linear Models
and you can put in your classification variables and you continuous covariates.

Doc Muhlbaier
Duke

[deleted_user]

OK...

Let's say I have an independant variable (A), a bunch of dependant variables (1,2,3, and 4) and a covariate (C). How do I set that up in the Model screen? Is that where I assign the covariate? Or is the covariate the "Quantitative Variable" on the Data screen?

Once again, I'm not smart, and really appreciate the help. Thanks!

[Doc_Duke]

ANCOVA is designed for one dependent variable at a time. Although the general linear model will handle multiple dependent variables simultaneously, that is not part of the EGuide interface and is a rather advanced topic.

Continuous variables are the quantitative variables on the Data tab. In ANCOVA terms they would be covariates.
Classification variables are the grouping variables, and would be labeled the independent variables in ANCOVA.

After assigning the variable types in the data tab, their roles are assigned in the Model tab. In classic one-way ANCOVA, they would both be main effects.

You can get more details with the context sensitive help key (F1) in EGuide.

[ChrisHemedinger]

The Linear Models interface in EG doesn't facilitate ANCOVA easily, but you can accomplish it with just a bit of code. Here is an example from one of our educators. Note that it uses ODS graphics and PLOTS=ALL, so it will generate the Analysis of Covariance plot as well.

/* ANCOVA */
ods graphics on;
proc glm data=sashelp.cars PLOTS=all;
class type origin;
model invoice = type | origin | mpg_highway;
run;