I desperately need your help as I am a German student using jmp for the first time and also never did a regression analysis before (learning by doing and with some books :)
What happens when doing a multiple linear regression but there is positive autocorrelation?
Because first I assumed that there is a linear relationship between the variables.
Also, I get completely different results than expected from my descriptive statistics. I am looking at real estate and especially how long it took to sell them. The dependent variable is the time till sold, and independent variables are several such as price, number of rooms etc. Furthermore, for the different regions I used dummy variables and also for the quarters when they were sold. Especially those dummies seem wrong as I got completely different results in my descriptive statistics. I left out the ones I found best as a reference
Observations over 2’000
R2 = 0.059 but obviously so small because some information was not given such as balcony garden etc.
Autocorrelation refers to the rows of the data table being related in some fashion. This would violate a regression assumption that the rows are independent of each other. I think your problem here is multicollinearity or sometimes just called collinearity. That refers to the variables in your model being related to each other (such as cost and number of rooms). The impact of this is inflated variances on your parameter estimates and models that are not very stable. Look up multicollinearity in your regression texts and you should see a good explanation of the problems as a result of this. Most texts will also discuss some things you might be able to do to help alleviate the problem.