0
$\begingroup$

I'm having trouble interpreting my R output. I know when checking for variance no patterns should occur but I have only covered the textbook scenario where it is very easy to interpret. However with my output I was just wondering does the validated of constant variance hold up?enter image description here

$\endgroup$
  • 1
    $\begingroup$ I've tried giving a clearer title - I suspect that you mean "assumption" instead of "validated" in the final line? $\endgroup$ – Silverfish May 10 '16 at 18:44
2
$\begingroup$

It is difficult to judge a plot with such few data points. You could try a formal test of heteroscedasticity such as the Breusch-Pagan test, implemented in the ncvTest function of the car package. Given how small your data set is, I would be surprised if significant heteroscedasticity is found.

The size of your data set is a more fundamental issue. A good rule of thumb is you should have 10-20 observations per parameter in your model, so your data set is borderline too small to fit a linear model with even one predictor.

$\endgroup$
  • $\begingroup$ Hi Slow, thank you for your comment but I have not the Breech-Pagan test therefore it is not required. I have only started linear regression and just want to make some comment on the graph that I got. The other assumptions hold i.e that the residuals are normally distributed and they are independent. But I was hoping to pass some remark on this. $\endgroup$ – user3223190 May 10 '16 at 18:58
  • 1
    $\begingroup$ I think you would be safe saying that it is difficult to tell based on such few data points, but there is perhaps a slight suggestion of non-linearity (U-shaped pattern). Something like this could happen if your response variable grows exponentially but you try to model it with a linear model, e.g., plot(lm(1.3^(1:8)~c(1:8)), which=1) $\endgroup$ – Slow loris May 10 '16 at 20:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.