Hi,
I have data from an animal experiment where we weight rats over 9 days post surgery. There are two groups - surgery and sham. I only a basic understanding of at statistics, but I believe this is repeated measures data.
Unfortunately, I have many missing values as the rats weren't weighed every day. I also have one group of 15 and one of 14.
What statistical method is the best to use? Are there alternatives to ANOVA?
Thanks
Anova based on a generalized linear mixed model (or mixed model as said by Martín) is not a bad idea. The difference between groups of 1 subject is not critical but how big is "many" missing values? Instead 9 days you can try group days in three phases, in example: 1-3 days; 4-6 days and 7-9 days using the average in same animal as response.
Martin is absolutely right here. This data *should* be analyzed using a mixel model approach, in which your "rat ID" enters as a random factor.
I recommend the book: Gelman & Hill: Data Analysis Using Regression and Multilevel/Hierarchical Models.
And, at least for the start, I also find it a very good idea to consult a local (bio-)statistician.
Absolutely
The way to analyze longitudinal data, including missing values or not, is through the linear model approach to the generalized mixed model approach depending on the nature of the response variable.
InfoStat (www.infostat.com.ar) implements these model using R. It includes a tutorial of these methods.
I'm not statician, but I was dealing with this topic a few year ago, I would suggest to read the West book (Linear Mixed Models) and the Fitzmaurice (Applied Longitudinal Analysis). There are also a lot of examples and tutorials on internet (even on Youtube). Selecting the right covariance model is probably the most difficult part, but I would suggest to use SPSS. Performing these kind of analysis is just a game with this software.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA