I have the data of about 30 patients on monthly visits over half a year. Each patient filled out the same tests every month (for example, the BDI for depression). Some of these tests may have a correlation, for example, the BDI-depression test might have an influence on the score for overall stress etc. Now I want to analyze the relationship of these test scores with the overall quality of life. I have been advised to use "multivariate panel regression with R". But after a first glance at the available literature (mostly econometric, this doesn't make it easier for me coming from a neuroscience background), I am not sure whether this is really the best way to analyze my data. I assume that the within-unit variation dominates the analysis, so I think a fixed-effect estimator would be the appropriate choice.
closed as off-topic by Michael Chernick, Peter Flom♦ Feb 10 at 10:56
This question appears to be off-topic. The users who voted to close gave this specific reason:
- "This question appears to be off-topic because EITHER it is not about statistics, machine learning, data analysis, data mining, or data visualization, OR it focuses on programming, debugging, or performing routine operations within a statistical computing platform. If the latter, you could try the support links we maintain." – Michael Chernick, Peter Flom
The advise you got was right. I suggest that an easier way to start is using Excel rather than R. Yes, Excel can also do simple linear regression models. You first need to activate the "Data Analysis" function by going to Options -> Add-Ins (Goolge should give you plenty of advice on the details). Once done, you can go to the Data tab and just follow the steps shown below. Let me know if this suggestions works for your.
As you have clustered data (each patient has multiple observations), the simple linear model will not produce correct p-values, but the coefficients are correct to get an initial understanding what goes on. I suspect, what you need is random effect or fixed effect regression modelling at the end, but that isn't a beginner's topic anymore. In R, you would need to employ a packing called plm to do that - but first you would need to understand the difference between fixed and random effect. I would therefore first just start with Excel and see whether there is anything interesting worth exploring further. If you don't find something interesting there, it may not be worth your time to read up about panel data (unless you have to).