Unfortunately the answer is not one that the “gender pay gap” addresses. i.e. it’s complicated and it is not necessarily because women are treated worse *because *they are women, their freely-made choices may well be a factor and that, ultimately, seeing a zero “gender pay gap” may not be possible nor even desirable because as it is currently defined and calculated it may drive behaviours and decisions that are ultimately harmful to women.
That, unfortunately, is a less snappy title and requires us all to think a little more deeply about what factors are *actually *at work and what will *actually *help.
Williams was not paid scale for the rest of the movie. She made $625K for the movie. The only reason she was paid scale for the reshoots is that union rules mandate that actors can not work for free on reshoots.
Well, no I don’t really know how it works. I assume it works by the people doing the study plugging in some number somewhere that represent the affect that different controls have. How do they come up with the numbers they used? The study doesn’t say.
That would seem problematic if data suggests that men work an average of 100 hours a year more than women (not saying that’s true, but it seems to me would be relevant)
They describe their approach starting on p8 and reference their 2016 study which has more detailed methodology. Let us know which part has you stumped.
Well, for instance this: “The estimated coefficient on the male dummy term β1
tells us the salary advantage experienced by males over females once all other factors are statistically adjusted for”
Does ‘estimated’ in statistics mean something different than ‘estimated’ in other fields?
If you knew how Hollywood works you would know that stars get paid more than other actors and that it has nothing to do with male or female. That Mark Wahlberg is a much bigger star than Michelle Williams and is thus paid more.
…again: I’ve talked about this. I’ve literally talked about this. If you had read my posts you would have known this. Why don’t you read my posts first before continuing to lecture me?
It’s an estimate of the coefficient for the population because they are fitting a sample of that population. I estimate the fairness of a coin by flipping it a bunch of times.
I’m not very well versed in statistics, but my understanding with OLS regression is basically like fitting a line to a scatter plot in such a way that it minimizes error, in this case with multiple variables. Thus, “estimated” in this case could possibly replaced by “calculated best fit” (with “best fit” being specifically minimizing the square of the errors). It’s potentially possible for them to have “overestimated” the effect of a particular variable (eg. education) if there were significant outliers in their data set - but I think it’s just as possible that they could have “underestimated” - basically their estimates are going to be totally a function of their data set, with no room for interpretation in their methodology (unless they are cherry-picking data points).
I agree, and it’s a bit unclear to me reading the report how this is addressed, if at all. They have footnotes indicating that any “All amounts reported as hourly wages are expressed in annual terms assuming a 2,000-hour full-time work year” - which I interpret to mean that only the base $/hr wage matters for this report, not number of hours worked for workers paid by the hour (ie. $20/hr = $40,000/yr, regardless of how many hours worked). If that is the case, any difference in # of hours worked between men and women wouldn’t show up as a pay gap.
However, as I mentioned previously, it’s possible that for salaried employees, men might work more “unpaid” overtime hours than women, and might actually get compensated with higher salary for the subsequent higher productivity. So if there’s data out there on hours worked in salary positions, it would be good to correct the data for that as well.