Multiple Regression Analysis with multiple years I am a fresher to statistics. Currently I am doing my project in Finance which is a kind of research. 
I have got 30 companies with 1 dependent variable and 3 independent variables. I have data for 5 years. Which method should I use to do regression analysis incorporating data of 5 years? Which software can be used for this?
Thanks in advance for your response.
 A: What you have is called panel data, and so all your regressions should be taking this structure into account. Depending on the type of outcome (dependent variable) you are considering (binary, numerical, etc.) you need to run a type of regression or another (linear, binary response, etc.).
I strongly recommend Wooldridge's book (Econometric Analysis of Cross Section and Panel Data). There you will be able to find many, many methods that will be appropriate for different contexts.
A: I would use multi-level modeling since you have hierarchical data (several data-points for each company). Sometimes called mixed models, since they mix fixed and random effects. You can use the lme4 package or the nlme package in R for this. (R is a popular and free statistics software.)
A: depending on what you suppose the error-structure for your model is you can run 3 basic models to start with when you have a panel-data structure (I am assuming you want to run OLS):


*

*Pooled OLS (just use all the years in one regression)

*Fixed Effects Model (FE)

*Random Effects Model (RE)


You might also want to fit dynamic panel data models like the Anderson-Hsiao estimator. (AH-EST).
You should read about these in a good textbook, like Wooldridge's book on panel data models (JW). Specifically you need to find out what the error structure of the model should be and run the model that is appropriate.
You can estimate all of these models for your data with various kinds of software. My personal choice is Stata and this is also what the book I recommended uses. But I'm pretty sure you can use open-source alternative like R.
