Interpretation of Regression Results

[Daniel Kosoko]

Hello esteemed members of Statalist!

I am conducting a cross-sectional analysis of UK firms examining the prevalence of ethnic 'presence' (a dummy variable to indicate if boards have at least 1 ethnic minority director) on firm financial outcomes such as 'ebitda1' (natural log). I have grouped and labelled my firms by industry. Below are the results of an outlier robust regression with no controls

rreg ebitda1 presence i.industry, nolog

Robust regression Number of obs = 90
F( 12, 77) = 3.92
Prob > F = 0.0001

----------------------------------------------------------------------------------------
ebitda1 | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-----------------------+----------------------------------------------------------------
presence | .2476538 .2250689 1.10 0.275 -.2005156 .6958233
|
industry |
Manufacturing | -1.609817 .4452493 -3.62 0.001 -2.496422 -.7232123
Utilities | -2.038274 .6915941 -2.95 0.004 -3.415414 -.6611347
Construction | -2.525203 .5318272 -4.75 0.000 -3.584206 -1.4662
Retailing | -1.624857 .5092477 -3.19 0.002 -2.638899 -.6108155
Leisure | -2.334223 .6885354 -3.39 0.001 -3.705273 -.9631742
Transport | -2.171283 .6284394 -3.46 0.001 -3.422666 -.9199008
Telecoms | -1.762717 .5499304 -3.21 0.002 -2.857769 -.667666
Finance | -1.703792 .4439282 -3.84 0.000 -2.587766 -.8198184
Prof/Support Services | -2.525983 .4528735 -5.58 0.000 -3.427769 -1.624197
Health & Public | -2.570481 .6284394 -4.09 0.000 -3.821863 -1.319098
Entertainment | -3.050315 .8044537 -3.79 0.000 -4.652186 -1.448443
|
_cons | 15.98618 .3860808 41.41 0.000 15.21739 16.75496
----------------------------------------------------------------------------------------

Perhaps this is trivial of me but I am not sure what the coefficients of the industries indicate. Do they represent the effect of 'presence' on 'ebitda1' for each industry or do they represent a direct relationship between each industry and 'ebitda1', if so, what exactly does that infer?

Thank you for responding.

Kind regards,
Daniel

[Daniel Kosoko]

Pardon me, this is cross-sectional data of FTSE 100 companies in 2017

[Carlo Lazzaro]

Daniel:
some comments about your query:
1) nowadays -rreg- seems a bit side-tracked. Hence, you should have (very) sound reasons to justify it in papers/research reports;
2) -rreg- coefficients have the usual meaning: they show the contribution of a given regressor (when adjusted for other predictors) to variation of the dependent variable (ie, the regressand; -ebitda1- in your case).
3) if you have such kind of doubts in interpreting the regression coefficients (at large), I would take the liberty to point you out to any decent statistics/econometrics textbook.
4) as an aside, please note that the outcome of -rreg- that you posted is really difficult to read. For the future, please use CODE delimiters to share with the list what you typed and what Stata gave you back (as per FAQ). Thanks.