ANCOVA question

[lazure]

---

Members don't see this ad.

Having a momentary brain fart here 🙂
When reporting the means for an ANCOVA do I report the raw means or the means adjusted for the covariate?
Thanks 🙂

 

[PublicHealth]

Having a momentary brain fart here 🙂
When reporting the means for an ANCOVA do I report the raw means or the means adjusted for the covariate?
Thanks 🙂

Adjusted means are typically reported. For example, if you're comparing groups and adjusting for age, you would report means adjusted for age. Make sure you note this in a footnote or in the title of the table/figure: "Table 1. Age-adjusted means on the [insert title of instrument]."

 

[lazure]

That's what I thought but I need confirmation, thanks 🙂

How do I get SPSS to spit out the standard deviations with the adjusted means, though? The output just gives standard errors and I'd prefer to avoid calculating the standard deviation by hand...

 

[PublicHealth]

That's what I thought but I need confirmation, thanks 🙂

How do I get SPSS to spit out the standard deviations with the adjusted means, though? The output just gives standard errors and I'd prefer to avoid calculating the standard deviation by hand...

Where possible, you should report standard errors. This differs by journal, though, so if you need standard deviations, simply compute means and use the SDs reported in the output. Steps: ANALYZE...COMPARE MEANS...MEANS. Make sure you select Standard Deviation from the OPTIONS box.

 

[lazure]

but aren't the standard deviations for the adjusted means also adjusted? Therefore they would be different from the standard deviations of the raw means....
Thanks,

 

[psychgeek]

but aren't the standard deviations for the adjusted means also adjusted? Therefore they would be different from the standard deviations of the raw means....
Thanks,

Doesn't the transformation involve only addition or subtraction to account for variation in the value of the covariate btw groups? If so, the SD remains the same.

 

[PublicHealth]

but aren't the standard deviations for the adjusted means also adjusted? Therefore they would be different from the standard deviations of the raw means....
Thanks,

In a nutshell, the family of ANOVA tests compute F's, which are estimates of the ratio of between- to within-group variance, taking into account how many subjects there are in the groups. Therefore, standard error, which takes sample size into consideration, is the most appropriate measure of variability in ANOVA. Click here for more information: http://www.sportsci.org/resource/stats/ttest.html I would also recommend consulting your stats professor.

If you click "Descriptive Statistics" under "Options" and run your ANCOVA. The output will be identical to the output produced if you simply computed means and SDs.