Regression Issues!!!

[skuddyb]

Hi guys,

So basically I have to run some regression analysis on a University assignment I am doing.

I have chosen to investigate the efficiency of point kicking in the Aviva Rugby Premier League.

My two independent variables are:

- Points Conceded
- League Points

My independent variable is:

- Points from Kicks

When I run the regression it returns:

- Adjusted R Square : 0.834
- Significance F : 0.000123784
- p-value : 0.48065

The P value is ridiculously high so I'm assuming it means the data is insignificant, but the others aren't too bad. What does this mean?

TIA!

 

[rwmill9716]

I assume that Points from Kicks is your dependent variable so your linear model looks like this: Points from Kicks = bo + b1(Points Conceded) + b2(Leaque Points)

Your adjusted Rsquare = 0.834, so your model is explaining 83.4% of the variability in Points from Kicks

The p-value at 0.48 leads to the conclusion that your model is not significant relative to your estimated error. As you know, the p-value needs to be less than 0.05 to conclude the model is significant at the 95% level of confidence.

The combination of R square and p value here suggests that you do not have many data being used in your model, i.e., an R square can be very high with n=5 and not be significant, while an R square can be very low (say 0.2) if lots of data are involved.

 

[skuddyb]

Hi,

Yes, only 12 DMU's in my data! I have tried it with another Metric and replaced "Points Conceded" with "Wins" which has returned a much better p-value of 2% and a 0 Significance F!

Should I use this model instead?

Thank you for your help!

 

[rwmill9716]

Use the model that proves significant. Those models that are judged not significant are not supported by your data. A good judge of a model's value is its PRESS statistic, which is available in packages like JMP and Minitab. Essentially, when comparing models, the one with the lower PRESS statistic is a better predictive model.

 

[skuddyb]

Been a great help thanks! Will look into those!

 

[pngxl]

Hi guys,

So basically I have to run some regression analysis on a University assignment I am doing.

I have chosen to investigate the efficiency of point kicking in the Aviva Rugby Premier League.

My two independent variables are:

- Points Conceded
- League Points

My independent variable is:

- Points from Kicks

When I run the regression it returns:

- Adjusted R Square : 0.834
- Significance F : 0.000123784
- p-value : 0.48065

The P value is ridiculously high so I'm assuming it means the data is insignificant, but the others aren't too bad. What does this mean?

TIA!

I am also interested in doing the same thing. I'll have to create an analysis template but don't where to start.