Interaction terms

[Samuel Welch]

Dear Users,

I am investigating the impacts of climate variability on the uptake of farming practices. One of my practices of interest is fertiliser use, and in previous literature, access to information through extension programmes is found to significantly impact the uptake of this practice in response to climate change.

Consequently, I wanted to include an interaction term between my climate variability variable, and my Extension Programme Dummy Variable. When I run the regression including this interaction term, it has a negative coefficient. Does this infer that farmers with access to extension services use less fertiliser in response to increased climate variability?

Or am I interpreting this coefficient incorrectly?

Any help would be greatly appreciated!

[Clyde Schechter]

Does this infer that farmers with access to extension services use less fertiliser in response to increased climate variability?

I'm not really comfortable with commenting on the interpretation of a regression result that you haven't shown, since it is possible that your model included complications that would change the interpretation, or errors that would invalidate it.

But assuming that your model explanatory variables are climate variation, extension program, their interaction, and no others, and assuming you have coded everything correctly and your data are appropriately organized to support this analysis, your interpretation is not correct. The interpretation would be: the effect of climate variation on fertilizer use is lower among farmers with access to extension services than among farmers who do not.

So it's not that the farmers with access use less fertilizer. It's that their use of fertilizer is not increased as much by greater climate variation as that of farmers without access.