Hi,
I am reading some papers that use GLM to deal with repeated measure. They often mentioned that they used "GLM repeated measures ANOVA". There are also F values and p values reported.
For example, in Vlčková (2012): "There was a statistically significant difference in protein contamination (A260/A280 ratios) using the three storing methods (GLM repeated measures ANOVA; F2,20 = 7.20, p < < 0.01)."
When I ran a GLM model, where should I look for the F value?
I found one answer online. It suggests that we use avo() function: aov(dv ~ iv, data = mydata, dispersion = NULL, test = NULL). Is this the right way?
Thank you!
Hello Qianhui,
Chances are 99% that, as Jos Feys 's reply indicates, the analysis that was run was really just a standard, repeated measures anova. I suspect that the "GLM repeated measures ANOVA" phrase was due to the user mimicking SPSS's apparent language for selecting a repeated measures anova: Analyze/GLM/repeated measures...
The "GLM" is redundant, as repeated measures is just a special case of the General Linear Model.
Good luck with your work.
I did look up the paper mentioned. They do describe using a general linear model. And they used Statistica software. Statistica --- as well as SAS and SPSS --- uses the abbreviation glm for general linear model. In contrast to this, R uses the lm() function for general linear model. (Or the function aov() can be used.)
Somewhat confusingly, there is a different type of model called a generalized linear model. R uses the function glm() for this, and in a lot of contexts you'll see the abbreviation GLM used for this. However, sometimes, you'll see GLIM or GLiM used.
The upshot? 1) In general, don't assume your reader will know what the abbreviation glm means. Someone coming from a SAS or SPSS background will read that differently than someone with an R background. < EDITED BASED ON FURTHER RESPONSES FROM OP >.
David Morse Sal Mangiafico Thank you so much for your answers! I didn’t realize that when the authors said GLM they mean general linear modeling. Thank you for the clarification.
I plan to use a Generalized Linear Model (GLM) because I have a categorical dv and I used repeated measures. There aren’t many examples of using GLM for repeated measures, I don't know how to report the results (e.g, what to report except the p value). Do you happen to have any suggestions? If not, that’s totally fine, thank you again for your help!
Hello again Qianhui,
While you can reproduce the set of anova tables that a repeated measures design yields, these bits of information will allow anyone sufficiently motivated to largely reproduce the set of results:
1. M & SD of all scores (optional: correlations among measures).
2. F(df1, df2) = x.xxx, p = .yyy (optional: MSE = z.zzz) for each tested effect.
Good luck finishing your project.
Thanks for making explicit you went Generalized Linear Model. Some people interpret glm as just linear model. You might give more detail so people know the link/distribution and any other parameters that may be important, but the anova function works on glm objects, often a chi^2 value being appropriate. Also, for repeated measures GLM the function glmer from lme4 is often used.
Qianhui Ni , I am a little confused. You marked Jos Feys answer as the answer, but you say you want to do generalized linear models. His answer is not about generalized one. Are you interested in doing a generalized linear model or not? I'm confused.
If you really insist, you can model a GLM with the "family=gaussian"
library(lme4)
library(car)
hoops = glmer(dep ~ Days + (1 | subj), data=long,
family=gaussian)
Anova(hoops, test="F", type="III")
hoop = lmer(dep ~ Days + (1 | subj), data=long)
Anova(hoop, test="F", type="III") # F tests (from car package)
These models all yield the same results:
numDF denDF F-value p-value (Intercept) 1 72 3724.109
Qianhui Ni , you mention you have a categorical dependent variable. How many levels are in this categorical variable? Is it nominal categorical or ordinal categorical?
- Badges
- Science topic