1

If my correlation started as r=.132 and became r=-.078 when controlling for a third variable how can I interpret this? I know the third variable obviously accounted for a large portion of the relationship but how do I interpret the relationship turning negative? Thanks

Sydney
  • 11
  • 1
  • The effect is quite small so it may just be noise. – user2974951 Feb 04 '19 at 07:58
  • See https://stats.stackexchange.com/questions/31841/coefficients-change-signs/32237#32237 for an identical question framed in terms of multiple regression. – whuber Feb 04 '19 at 14:25

1 Answers1

1

First, I'd look at the precision of those estimates by examining their standard errors. It might just be random fluctuations.

Assuming you have estimates that are precise enough that the effect is "real" then you can say something, but it's hard to word it properly without context. But it would be something like:

"While there was a modest positive bivariate relationship between A and B, when we controlled for C, the relationship was reversed."

Peter Flom
  • 94,055
  • 35
  • 143
  • 276