4 ms·

The women surveyed could have worked more hours than the men. But by not controlling for it you reduce the studies explanative powers. You don't get to just i

The women surveyed could have worked more hours than the men. But by not controlling for it you reduce the studies explanative powers. You don't get to just ignore it because it is convenient. It doesn't even pass the "more likely than not" standard of evidence.

KirinDave 10y ago
Right, but uh... The vast majority of positions Glassdoor cares about are salaried positions which don't track hours. They're exempt from overtime, as well. It also seems to me that that signal would show up weakly in the data because it wouldn't vary the immediate effect on apples-to-apples salary, only on promotions and possibly referrals. So I think your complaint is specious and the inclusion of a totally unrelated data point from a radically different dataset is either very misguided or profoundly disingenuous.
- sheepmullet 10y ago
  > The vast majority of positions Glassdoor cares about are salaried positions which don't track hours Yes. And this makes any kind of data analysis a lot harder. On paper I work the same number of hours as my Japanese coworkers. In reality they work more than double. It's the same way on-call for many companies is just rolled into salary. I've managed projects as a senior dev while the senior dev sitting next to me was purely technical. Etc etc. The truth is two senior devs sitting right next to each other can have wildly different actual job duties. I suspect glassdoor has taken at least some of those differences in work duties vs stated title into account. > because it wouldn't vary the immediate effect on apples-to-apples salary, Why wouldn't it? Its not uncommon to stay at a certain level/title for 3-5 years. You can't just ignore hours worked because "salary". I come from a strong union background where you would get overtime or be able to take the time off later and where there were fairly rigid job duties. Where interviews and scoring candidates is all done in a very standardised fashion. And oddly enough after adjusting for these explicit factors we had a very small unexplained gender pay gap. Much smaller than average for the tech industry. Perhaps it's in our interest to make these factors explicit rather than implicit?
  - KirinDave 10y ago
    > Why wouldn't it? Because hours worked doesn't affect salary. What's more, you keep assuming that there aren't counter-examples in this dataset at all. It's just a given to you: "I read on Forbes that across all Americans women work slightly less per week" and brought that to this conversation. But in this industry long hours are the norm, single people without family duties are the majority, and many compensation models very much incentivize long hours. > You can't just ignore hours worked because "salary". Unless we're talking consulting, that's exactly what Salary lets you do. > Much smaller than average for the tech industry. The average pay bracket difference is indeed smaller. However, it's still large enough for concern, and the other issues I listed around work culture are problems that disadvantage women more than men in modern society. > Perhaps it's in our interest to make these factors explicit rather than implicit? I am always in favor of transparency and more analysis but I will not let ANY interpretation of the specific data we have on hand be constantly forced off on the grounds that we could collect better data. Saying, "Women get paid less, let's make sure that's not happening" is a call to action, not an accusation leveled at anyone specifically. If women in your organization aren't being paid less, congrats. Job done. Let's be clear: You're effectively trying to stifle and derail the entire conversation on the grounds that it might not be happening everywhere, or that statistics might not be what popular media claims they are. That's counterproductive and frankly unwelcome. If that's your take, your take stifles the discussion here. We can simultaneous recognize that tech is a more complicated compensation environment and also that women are reporting under-compensation. We're all adults capable of multi-tasking.
    - sheepmullet 10y ago
      > Because hours worked doesn't affect salary. > Unless we're talking consulting, that's exactly what Salary lets you do. We are clearly having communication issues and I'm not sure how you are addressing what I'm talking about. My point is: Work longer -> slightly more productive -> slightly higher pay. > You're effectively trying to stifle and derail the entire conversation If the conversation starts from an inconclusive or flimsy premise then it should be derailed. It is far too easy to make bad decisions based on flawed studies. > and also that women are reporting under-compensation. The first step is to work out if it's a real issue we could be, or should be, solving. It could be a problem worth addressing if the gap was driven by sexism. It could be a problem worth addressing if the small pay gap was a major contributor to women leaving the industry. It could be a problem if the small pay gap was a strong contributor to women not entering the field. It could be a problem if we we were seeing a rapidly widening gap. Etc etc. But you have to actually make a case.

KirinDave 10y ago

Right, but uh... The vast majority of positions Glassdoor cares about are salaried positions which don't track hours. They're exempt from overtime, as well. It also seems to me that that signal would show up weakly in the data because it wouldn't vary the immediate effect on apples-to-apples salary, only on promotions and possibly referrals. So I think your complaint is specious and the inclusion of a totally unrelated data point from a radically different dataset is either very misguided or profoundly disingenuous.

sheepmullet 10y ago
> The vast majority of positions Glassdoor cares about are salaried positions which don't track hours Yes. And this makes any kind of data analysis a lot harder. On paper I work the same number of hours as my Japanese coworkers. In reality they work more than double. It's the same way on-call for many companies is just rolled into salary. I've managed projects as a senior dev while the senior dev sitting next to me was purely technical. Etc etc. The truth is two senior devs sitting right next to each other can have wildly different actual job duties. I suspect glassdoor has taken at least some of those differences in work duties vs stated title into account. > because it wouldn't vary the immediate effect on apples-to-apples salary, Why wouldn't it? Its not uncommon to stay at a certain level/title for 3-5 years. You can't just ignore hours worked because "salary". I come from a strong union background where you would get overtime or be able to take the time off later and where there were fairly rigid job duties. Where interviews and scoring candidates is all done in a very standardised fashion. And oddly enough after adjusting for these explicit factors we had a very small unexplained gender pay gap. Much smaller than average for the tech industry. Perhaps it's in our interest to make these factors explicit rather than implicit?
- KirinDave 10y ago
  > Why wouldn't it? Because hours worked doesn't affect salary. What's more, you keep assuming that there aren't counter-examples in this dataset at all. It's just a given to you: "I read on Forbes that across all Americans women work slightly less per week" and brought that to this conversation. But in this industry long hours are the norm, single people without family duties are the majority, and many compensation models very much incentivize long hours. > You can't just ignore hours worked because "salary". Unless we're talking consulting, that's exactly what Salary lets you do. > Much smaller than average for the tech industry. The average pay bracket difference is indeed smaller. However, it's still large enough for concern, and the other issues I listed around work culture are problems that disadvantage women more than men in modern society. > Perhaps it's in our interest to make these factors explicit rather than implicit? I am always in favor of transparency and more analysis but I will not let ANY interpretation of the specific data we have on hand be constantly forced off on the grounds that we could collect better data. Saying, "Women get paid less, let's make sure that's not happening" is a call to action, not an accusation leveled at anyone specifically. If women in your organization aren't being paid less, congrats. Job done. Let's be clear: You're effectively trying to stifle and derail the entire conversation on the grounds that it might not be happening everywhere, or that statistics might not be what popular media claims they are. That's counterproductive and frankly unwelcome. If that's your take, your take stifles the discussion here. We can simultaneous recognize that tech is a more complicated compensation environment and also that women are reporting under-compensation. We're all adults capable of multi-tasking.
  - sheepmullet 10y ago
    > Because hours worked doesn't affect salary. > Unless we're talking consulting, that's exactly what Salary lets you do. We are clearly having communication issues and I'm not sure how you are addressing what I'm talking about. My point is: Work longer -> slightly more productive -> slightly higher pay. > You're effectively trying to stifle and derail the entire conversation If the conversation starts from an inconclusive or flimsy premise then it should be derailed. It is far too easy to make bad decisions based on flawed studies. > and also that women are reporting under-compensation. The first step is to work out if it's a real issue we could be, or should be, solving. It could be a problem worth addressing if the gap was driven by sexism. It could be a problem worth addressing if the small pay gap was a major contributor to women leaving the industry. It could be a problem if the small pay gap was a strong contributor to women not entering the field. It could be a problem if we we were seeing a rapidly widening gap. Etc etc. But you have to actually make a case.