Clarification of Results Interpretation

[Daniel Kosoko]

I am conducting a cross-sectional study of UK firms to investigate the relationship between boardroom diversity (alternately using dummy variable 'presence' and proportion 'div_prop') and firm performance using ('roa' and 'tobinq') while controlling for firm and board characteristics (firm size, board size, average age and average additional directorships). By using dummy variables to represent industries. I use the industry with the highest average proportion of ethnic diversity 'div_prop' as my base group and to compare coefficients of varying industries. To arrive at a conclusion I aim, to find a noticeable contrast between the coefficients of industries with low diversity and industries with higher diversity. I inputted the following code

reg roa presence##i.industry , robust

. Natural resources is my base group and the first group when all are listed. My results are below

My interpretation of these results is 3.302 - 0.560 = 2.742 is the difference between the effect of ethnic presence in the Natural Resources industry and the Construction industry. The positive number, therefore, implying that a unit increase in presence increases Natural Resources roa by 2.742 percentage points (I did not take the natural log).

Your insightful solutions will be highly appreciated.

[Marcos Almeida]

I strongly recommend to take a look at the background knowledge related to regression analysis.

Rest assured that your effort will be exponentially rewarded.

That said, with the exception of just one interaction term (serendipity?) , the model doesn’t provide p-values low enough to back up any inferential proposal.

[Carlo Lazzaro]

Daniel.
as an aside to Marcos helpful advice, you omitted the first part of the Stata outcome (that you should have better reported via CODE delimiters); hence, interested listers cannot say whether you added too much stuff in the right-hand side of your regression equation when contrasted against your sample size.