Hello dear forum members!
Consider the following model:
All the estimated coefficients are significant. To examine the effect of interaction, I use:
The result is:
Array
However, when I estimate the model with robust standard errors (as suggested by literature to adjust for clustering):
The interaction term becomes insignificant. Yet, the plot of the marginal effects appears to be similar to that of the "non-robust" estimates:
Array
Is there a plausible explanation for the loss of significance of the interaction term? I suspect this may raise a red flag to the reviewers.
P.S. In the past, I might have come across the notion that in an exponential model the significance of the interaction may not be "observed", and thus marginal effects must be calculated. But I am not 100% sure about this.
Consider the following model:
Code:
xtpoisson y x1 x2 c.x1##c.x2, fe
Code:
margins , at(x1=(1 0) x2 = (1,2,3,4,5,6,7)) marginsplot
Array
However, when I estimate the model with robust standard errors (as suggested by literature to adjust for clustering):
Code:
xtpoisson y x1 x2 c.x1##c.x2, fe vce(robust)
Array
Is there a plausible explanation for the loss of significance of the interaction term? I suspect this may raise a red flag to the reviewers.
P.S. In the past, I might have come across the notion that in an exponential model the significance of the interaction may not be "observed", and thus marginal effects must be calculated. But I am not 100% sure about this.