2

An independent variable in a linear regression I am working on has positive coefficient but this variable has a negative correlation with the dependent variable. I know that this is mathematically possible given each variables' coefficient in a multiple regression is a results of complex interaction between the relationship of the variable of interest with the other independent variables as well as the dependent variable. But is there any intuitive way I can explain this phenomenon to my business stakeholders without going into the math, maybe using a real life example? Thanks

  • Please visit our [threads on Simpson's Paradox](https://stats.stackexchange.com/search?q=simpson*+paradox+regression) – whuber Oct 11 '19 at 19:42

0 Answers0