Linear Regression w/one categorical predictor

[RER39]

Hi everyone,

 

I need to do a simple linear regression with just one categorical predictor.  The predictor (age range) is broken down into three categories; young, middle, old. The DV is a continuous variable. I'm assuming I need to do something different in SPSS because the IV is categorical. Could someone clarify? Thanks so much in advance!! 

[spunky]

 

I need to do a simple linear regression with just one categorical predictor. 

 

or, as it is better known, a one-way ANOVA. 

[RER39]

or, as it is better known, a one-way ANOVA. 

Thanks for the reply. That was my first instinct as well, but I have been instructed by my professor to use regression, which threw me off. Is it possible to use regression for this? 

[spunky]

 Is it possible to use regression for this? 

 

sure it is. maybe your prof is doing this so you can see the connection because Psychology has a tradition of teaching ANOVA and Regression as if they were separate things (cue in Cronbach's 1957 article on the 2 disciplines of scientific psychology) but the emphasis on teaching methods courses has lately been on helping students see the connections (like ANOVA is a special type of multiple regression, the general linear model is a specific instance of the generalIZED linear models and they're all specific instances of structural equation models which is itself a family of techniques under the umbrella of latent variable models). 

 

anyhoo, just make sure you code your things correctly (remember you always need one code less because one of the levels of the factor becomes the reference) and you're good to go. 

[RER39]

sure it is. maybe your prof is doing this so you can see the connection because Psychology has a tradition of teaching ANOVA and Regression as if they were separate things (cue in Cronbach's 1957 article on the 2 disciplines of scientific psychology) but the emphasis on teaching methods courses has lately been on helping students see the connections (like ANOVA is a special type of multiple regression, the general linear model is a specific instance of the generalIZED linear models and they're all specific instances of structural equation models which is itself a family of techniques under the umbrella of latent variable models). 

 

anyhoo, just make sure you code your things correctly (remember you always need one code less because one of the levels of the factor becomes the reference) and you're good to go. 

Thank you so much! Exactly what I needed to know

[juilletmercredi]

That is one thing I wish my early professors had taught the connections about earlier in statistics (connections between the general(ized) linear model).  I kind of vaguely understood the concept, but I didn't really get that they were the same thing until the beginning of graduate school when I kind of pieced it all together and was like "OH!"

 

Like this:

 

like ANOVA is a special type of multiple regression, the general linear model is a specific instance of the generalIZED linear models and they're all specific instances of structural equation models which is itself a family of techniques under the umbrella of latent variable models). 

 

Would've changed my life in college!  I still love its simplicity  I would love to use this as a flow chart in introductory statistics classes.

[spunky]

That is one thing I wish my early professors had taught the connections about earlier in statistics (connections between the general(ized) linear model).  I kind of vaguely understood the concept, but I didn't really get that they were the same thing until the beginning of graduate school when I kind of pieced it all together and was like "OH!"

 

Like this:

 

 

Would've changed my life in college!  I still love its simplicity  I would love to use this as a flow chart in introductory statistics classes.

 

this is the 2nd statistics/methodology question from a psychology (or social science) person that we get in the span of just a few days. if a 3rd question comes along we're gonna need to open shop here!

Edited October 28, 2014 by spunky

[Roll Right]

I wonder if they were trying to get you to understand the concept of dummy variables........

 

Thats the only point I see to such an exercise.

 

You need to include two values from your categorical variable, and one out as a reference group.

[juilletmercredi]

this is the 2nd statistics/methodology question from a psychology (or social science) person that we get in the span of just a few days. if a 3rd question comes along we're gonna need to open shop here! 

 

This would make my life, lol