The correlation between one variable and its coefficients produced by `lm` output

Asked Jul 19 '22 at 02:07

Active Jul 19 '22 at 02:07

Viewed 22 times

Tried cor(x1,y) and I get a 0.09. However, when I put the x1 in the lm model, its coefficient is negative. Does it make sense?

asked Jul 19 '22 at 02:07

nobodyishere

Was x1 the only predictor used in your lm model? Or did you include others (x2, x3, ...)? – Adrian Jul 19 '22 at 02:32
https://stats.stackexchange.com/questions/116804/coefficient-changes-sign-when-adding-a-variable-in-logistic-regression might be of interest – Adrian Jul 19 '22 at 02:33
https://stats.stackexchange.com/q/44279/9330 could also help – Adrian Jul 19 '22 at 02:36
@Adrian Hi, Adrian, there are other variables such as (x2,x3,...). I am just wondering should the negative number produced by cor(y,x1) be contradict with the positive coefficients of x1 in lm. – nobodyishere Jul 19 '22 at 02:52
Have a look at https://stats.stackexchange.com/questions/91666/sign-of-coefficients-in-linear-regression-vs-the-sign-of-correlation, it's essentially the same question as yours – Adrian Jul 19 '22 at 02:58
2

Probable duplicate of the link @Adrian gave: Sign of coefficients in linear regression vs. the sign of correlation – Dave Jul 19 '22 at 03:00
Thanks a lot, Adrain @Adrian – nobodyishere Jul 19 '22 at 03:15

0 Answers0