1

Usually I have small number of subjects (e.g., 4-6 rodents) and I record electrophysiological activity during presentation of a stimulus or performance of a task. A general question that arises when dealing with statistics is: Under what circumstances is it appropriate to pool trials from each subject and do a single level analysis on the pooled data? Due to usually large number of trials per subject and small number of subjects, it is often difficult to show the effects at the second-level across subjects (the effects that you see across individual subjects).

gung - Reinstate Monica
  • 132,789
  • 81
  • 357
  • 650
K Neuro
  • 11
  • 2
  • Why not analyse each animal separately and then use meta-analysis to pool the effects? – mdewey Jan 29 '18 at 17:01
  • what kind of meta-analysis do you mean? – K Neuro Jan 29 '18 at 17:45
  • The standard method, inverse variance weighting. – mdewey Jan 29 '18 at 18:34
  • To my knowledge, these methods are used for pooling small populations and minimizing the variance. Correct me if I am wrong. I am not sure if it is valid to use this method for pooling trials from individual subjects. Let me know about this. i'd also appreciate if you can point me to an ex of similar situation. Thanks! – K Neuro Jan 29 '18 at 19:34
  • Some related threads: (**1**) [Should I run separate regressions for every community, or can community simply be a controlling variable in an aggregated model?](https://stats.stackexchange.com/q/17110/); (**2**) [Multilevel model vs. separate models for each level](https://stats.stackexchange.com/q/63505/); (**3**) [Under what conditions should one use multilevel/hierarchical analysis?](https://stats.stackexchange.com/q/1995/) – gung - Reinstate Monica Jan 30 '18 at 20:19
  • I went through the threads. They were insightful but I believe that they don't address this question. Here there is a single population (small size) but there are much higher number of trials per subject. – K Neuro Jan 31 '18 at 09:41

0 Answers0