I'm currently working on my Bachelor thesis relating to behavioral finance. My topic requires me to run a regression analysis to spot the presence of herd behavior in the Vietnamese stock market. The model for testing is as follows:
CSAD=c+b1*abs(rm)+b2(rm)^2

Here are the results generated by Eviews 8:

However, due to my lack of knowledge in statistics, I am unable to interpret those results in a meaningful way. Could anyone please give me some basic explanations on the information listed in the table? Which ones are important and which are less? My friend said the Durbin Watson stat and R-squared of my analysis are too low, is it true and what is considered to be "normal" or "statistically correct" here? I would be grateful for any help if possible.

Re: Please help with the regression analysis result

Hi,
just a few quick thoughts: you have very low p-values which mean all your coefficients are significantly different from zero . This could be good news, bzut in your case you have a really large dadtaset, so basically every small effect will be significant, and your effect is very small as per the R-Squared(adjusted). This says that roughly 3% of the variation is explained by your model, the rest of about 93% is not. So, I would not worry about Durbin-Watson and the like - practically there is no effect that your model could show. I would look at other explanatory variables if possible.

I hope this helps a bit, though not really good news.

Re: Please help with the regression analysis result

Hello rogojel,

Thank you very much for your quick reply. Since I took the model from one of the herd behavior research that I believe to be credible enough, I think there are some problems with my handling of the data set. It would be nice if you could suggest something that I could do to improve the situation? Furthermore, how high should the R-squared be to be explanatory enough regarding to the data? I know these are some silly questions but I'm a bit lost with all these things, and would really appreciate some help.

Re: Please help with the regression analysis result

R^2 interpretation is context based. In some situations 3% would be great and in others 3% would be way too low. Have you gone to your fields literature to see what you can find as comparisons. Is C your constant (i.e., intercept)? So you only have a term and its square in the model. As rogojel wrote, perhaps there are more predictors available to help with the explanation.

Re: Please help with the regression analysis result

Hi All,

Please help me to understand my multiple regression analysis result. I have one dependent variable and 7 independent variables. I tried multiple regression using SPSS enter method entering all independent variable and one dependent variable, which showed only two independent variables significant with dependent variable. However, when i did pearson correlation i found six out of seven independent variables significant also, there exist correlation between predictor variables. So, somewhere i found that we need to exclude those correlated and insignificant variables in mulitple regression analysis to get a good output and i found Stepwise regression method in SPSS for that. So, i tried with SPSS stepwise regression analysis. Here, is the attached below the multiple regression analysis result. Please have a look and let me know is this correct analysis approach and whether the analysis result is good enough to proceed in my thesis work or not. I am a beginner in stat, your help will contribute a lot in my study. Thanks.