Considering that the company I work for tried to hide their gender pay discrimination by "releveling" all the women who were getting systematically underpaid, you should in fact report these data without disaggregation.
"Those people make less because they are all lower level" is not a mitigating factor for pay discrimination.
As someone who’s done a bunch of DEI analyses for companies, I don’t really believe you understood what you’re saying happened. That really does not make sense. It would be a massive organizational change for some niche cherry picked reporting stats that few people look at. Does not sound plausible at all. This stuff is genuinely hard to implement.
You absolutely need to normalize for level when determine pay discrepancies. As well as hiring, retention, and promotion rates. It’s silly to suggest otherwise. Frankly a lot of companies fail to do this and are convinced they have huge problems when in fact things are pretty fair at the individual level and on track to gradually diversify over time. But you can’t just snap your fingers and diversify the top of an org. It takes a long time.
I work for a federal defense contractor that is required to prove in Department of Labor audits that pay is not discriminatory. The possibility of losing those federal contracts is an existential threat to the company. Definitely not "niche cherry picked reporting stats that few people look at."
The significant restructuring (affected several hundred people) that you describe did in fact occur.
Hmm, no, still don’t believe it. That’s just absurd. I just refuse to believe that they systematically demoted women / promoted men to try and obfuscate pay differences at original levels.
This idea is surely much more expensive to implement than just paying the women more. It’s also fairly easily discovered by an audit of any competence or whistleblower. And the women would surely have a strong negative reaction to this which is an operational risk of them just leaving.
What you are proposing happened is illegal discrimination to begin with, but it’s also really difficult to do. Seems more likely a willing company would prefer to simply lie
Most of the time it’s companies concerned that they’re not doing a good job on equal opportunity stuff. Most of the time (mostly east coast white collar jobs) these orgs are doing just fine. A lot are concerned because top level stats suggest minorities are doing worse. But when you look at hiring, retention, promotions, comp, employee engagement/happiness surveys, they tend to be pretty equal when normalized. Companies that are interested in doing these analyses that I’ve looked at are generally doing the right thing and are on track to diversify to the maximum extent that the candidate pole sensibly allows over time, which is pretty slow, but it’s there. Generally speaking if you’re in a liberal cohort of people you’re not going to see a lot of impact structural biases that hinder folks at the organizational stat level. You very well might at the individual manager level and that’s basically impossible to catch aside from personal conversations and astute leadership.I wouldn’t generalize this across all areas or industries mind you.
To some extent imo, provided an org isn’t overtly discriminatory, the most important thing is to make sure that hiring channels include diverse sources for candidates.
There's no such thing as downleveling or demotion at Google, and hiring & promotion rates are tracked by race and gender.
I would bet that the pay discrepancy here is entirely due to differing levels, which in turn is because Google's DEI efforts are ~5-7 years old and when new hires come in at L3, it takes that long or more to get to the high-paid levels. That and people who were hired a while ago are sitting on lots of appreciation in their stock grants.
Yeah, there was downslotting back when L5+ came in as MTS.
They got rid of it. I think they changed it to "everyone comes in at L3/L4, and if you're good you'll rapidly get promoted to your true level" around 2012, and then sometime between 2014-2020 it changed to "sure, you can come in at L6+ and we don't really care if you're incompetent".
Reporting without disaggregation means that companies would be discouraged from hiring junior women, and that would just make the issue worse not better. Be careful with what you wish for.
This has been a mantra of many since the 1990s, and I continues to get proven false (generally). There are so many variables here.
If you want to read a book that directly talks about this, check out Thomas Sowell's "Economic facts and fallacies". There are chapters for both gender and race economic differences. Even though the book is ~14 years old, it counters these kinds of arguments.
How is anyone underpaid? If they agree to the negotiated compensation, isn't that equitable?
Let's say there is an employer who systematically wants to underpay women to save money by exploiting the pay gap. If there are no women in agreeing to the pay the employer will adjust upward in an attempt to either hire women or ignore the disparity, pay more, and appear bias.
I would argue that women who want more pay must discriminate more than their potential employer.
I can only assume you don’t actually want your question answered. Because if you did want an answer you could have asked an ai or searched as there is plenty written on the topic.
I see this pattern all over the place. Often with conservatives. Asking a reasonable question that puts the fundamental concept in doubt. But never actually wanting to know the answer because a simple search would have revealed lots of thorough and existing discussions answering the question.
I'm not joking. This is an area that AI would answer well. It's an issue that's been around for decades and is frequently written about by reputable sources. Like Wikipedia, it's probably a good start for someone like the parent who hasn't given this area much thought.
That's fair. Then don't read the search result from the economist. Read the result from the Census, or the Forbes article, or HuffPo if you don't like their stance.
My point is simply that your question is valid, but has been asked and answered a thousand times before. Nothing in your question is specific to this audience that we'd have a twist on the answer that isn't represented in any of the thousand existing answers.
I'm curious, since you didn't get much discussion on your question here, did you take a minute to answer your question elsewhere?
I am almost entirely surrounded by either professional people who do not discuss their income or people who share my position. The few discussions I have with people is they tend to agree that people are not conscripted or enslaved so they end up agreeing that they pay gap is caused by the acceptance of an offered compensation.
I am not discussing whether this is fair, equitable, or even kind. I just do not care for the misrepresentation of the situation.
I could argue I am both overpaid or underpaid based upon my perspective. Having been on the other side of the negotiations with friends and strangers the situation is the same. They want as much value as they can extract. I too want as much value as I can extract - within agreement. Now that I am back to being an employee I have requested more compensation. They were open and asked why: apart from desire, I have not provided additional value (yet) and thus am not entitled nor deserving.
Often people take what’s offered, whether it is a fair deal or not. Everybody isn’t out there negotiating their best possible rate all the time. Corporations have long taken advantage of whomever they can to earn a buck - it’s a time honored “tradition”.
That’s not underpaid. To be underpaid normally means one received less than was agreed upon. That scenario is offering a discount, maybe.
If it is true that women are likely to offer discounts, it questions why for-profit companies hire men at all? Meanwhile we worry about companies not being willing to hire women, not men.
Wiktionary - Underpaid: "Getting too little financial compensation for one's work."
Meriam Webster - Underpay: "to pay less than what is normal or required"
Cambridge Dictionary - underpay: "to pay (a person) too little"
Dictionary.com - underpaid: "not paid enough"
Only the Cambridge version of those could imply (to me) only the definition of receiving less than what was agreed upon.
I personally like the Meriam Webster definition. If a software engineer is normally paid $ X, and you are paying $ 0.5*X, then that person is underpaid.
> Wiktionary - Underpaid: "Getting too little financial compensation for one's work." Meriam Webster - Underpay: "to pay less than what is normal or required" Cambridge Dictionary - underpay: "to pay (a person) too little" Dictionary.com - underpaid: "not paid enough"
Exactly. There is nothing in there about agreeing to take less than you might have been able to get if you chose to charge more.
> If a software engineer is normally paid $ X, and you are paying $ 0.5X, then that person is underpaid.*
The general opinion is that people are not fungible. Why do you believe that they are?
> “Those people make less because they are all lower level"
Of course it is. People with more experience and higher performance get paid more.
If the accusation is instead that certain groups are being promoted less often because of their immutable characteristics, it requires some evidence to substantiate.
That also has some flaws if not combined with gender/race ratio in those levels. Underpaying for the same level isn't the only potential issue, but bias in hiring and promotions for more senior roles could be too.
In theory it is not possible to report 100% accuracy here, as I do not have to reveal my gender or race to the company I work for, correct ? That is unless not specified is a category.
I personally feel like it is not the companies business to know this about me. Is there a legal requirement around this ?
I am questioning how anyone can decision anything based off any data regarding this topic accurately , including governments or courts. Seems like it is all good faith but no way to validate the results.
Reporting is entirely voluntary, so there are no concerns around that as far as the original collections go. No one is forced to report anything.
The person leaking it is acting unethically though. It has no whistleblower value of any kind, and betrays those who filled it out thinking it would be internal only.
Company wide things were being leaked at G within hours of announcement nearly a decade ago. :( Anyone who didn’t expect it just really wasn’t paying attention. Not that it excuses it, but c’mon.
The data is entirely self and voluntarily reported by employees. There are no double checks for accuracy or typos or attempts to make it representative beyond, “If everyone fills this out, it will be helpful for everyone.”
It is useful as far as it goes, but one can only draw limited conclusions from what is there.
Business Insider doesn’t really caveat it’s reporting very well here. And it should.
This is very poor reporting indeed. BI claims this spreadsheet "gives us our best look yet at how much the tech giant is paying" despite the data being entirely self-reported by employees and completely unvetted. The odds that sample in the sheet is representative of the broader population are vanishingly small, even if the data were accurate.
Perfect example is the report that "The employee with the highest base salary reported being a level 7 software engineer with a base salary of $718,000 and a total compensation package of just under $800,000."
I suppose there might be some engineer making $720k in base and $80k in bonus/equity combined, but I've never heard of such a thing. There are obviously errors in the data. Yet BI doesn't question the data at all.
Disclosure: I work at Google. Opinions are my own and do not reflect those of my employer.
Legally, the company can ask your gender and race, but you aren't required to answer. It's illegal to use either of those in the hiring process, so they avoid even asking until you're already hired. And imo they shouldn't ask period.
I can't access the article, even via the archive link (infinite captchas), so idk what what BI is reporting. But the original sheet is employee-created, so identity stats are probably not coming from the company's own database.
Does BI not have anyone on staff familiar with statistics? The map with “minimum salaries” seems to just prove that a more populated office location will contain a broader range and more extreme outliers of self-reported salaries. A previous article on this topic claimed that Software Engineer had the highest max salary but there are probably just more people with that title at the company than any other title.
The person whose ethnicity is labeled "Indigenous software engineer" cracked me up. I pictured a race of humans born with ctrl + c + v buttons on their fingers
Less than 10% of the company self reported salary data over an unspecified time window, and a limited set of aggregate statistics are reported from this distribution. Certainly interesting information, but unfortunately I don’t think you can make honest inferences from these figures. Any comments or interpretations are little more than speculations and are basically deceitful.
> "We compensate Googlers based on what they do, not who they are," said Google spokesperson Tamani Jayasinghe in a statement.
Isn't this famously false, given that they pay large disparities depending on someone's country in which they work? This only holds true within a specific country.
Doesn't Google also try to dock your pay if you move out of a HCOL city, too?
What passport you are able to receive, and where you can resultantly live and work, is a function of who you are. Nationality is a protected class for purposes of discrimination, at least in US laws. (Not that US law applies, just as a reflection of the moral context.)
Yet Google happily uses this opportunity to pay wildly different amounts to people doing the exact same work.
In any case, this clearly invalidates the first claim, that people are paid based on what they do. They're paid actually based on their replacement cost in the place they are legally restricted to do it.
To make a contrived example, this means that a woman in a LCOL country/city will earn less than a man in an HCOL country/city, for the exact same work.
All of this pay equity stuff is total nonsense and posturing as long as Google is taking advantage of labor market arbitrage (which their staff cannot easily do, not being multinational themselves).
If they truly believe in equal pay for equal work, it should plainly apply across national borders that staff cannot cross. Anything else is lip service.
Your nationality determines where you can live and work. That's not conflating anything.
The fact that some tiny lucky subset of people with very specific credentials can get permission to move across borders does not change this at all. (In fact, H1Bs generally making less than others for the same work reinforces my point.)
> We compensate Googlers based on how little we can get away with in the given job market where the employee lives, remote-working or otherwise.
Whether that’s fair or not is a matter of belief system.
Google’s Bay Area compensation isn’t a result of some intrinsically just meritocracy - they’re a wealthy business competing with other wealthy businesses over an extremely high-demand labor pool.
If I remember correctly a previous version of this article basically just copy/pasted the highest numbers they could, which meant that they focused on random typos. There was a running joke and a bunch of memes about some entry level engineer making 700k base, because I think it was very clear that they mistyped “178k” or “187k” as “718k”. For an L7 engineer bases go a bit higher but numbers that high, with that little equity, are unheard of. It’s highly likely nobody with an understanding of how compensation works looked through this table.
What strikes me about this story (and "reporting" in general) is how eager people writing stories are to stir up the worst kind of racial/gender-based implications without any further responsibility for analyzing or self-doubting the top level findings they're throwing in front of the audience.
You would think that anyone presenting statistics that call into question some kind of differential treatment by race, ethnicity, or gender would take a moment to point out the cautions in the data. Or even express some doubt about writing the story to begin with.
But no, for some reason, we are fascinated by the "0th order analysis" of inequities by race, almost want to see it blown up into incorrect interpretations and incentivize reporters to produce such stories.
I think there should be a code of ethics where reporters should be questioned about what they seek to accomplish with half-assed analyses like this, if they fail to accompany it with huge warnings about what it hasn't taken into account that could explain differences:
-- age
-- experience
-- selection biases in employees tenure (both hiring and departures)
etc.
Without this kind of info, all these stories do is shout an alarmist call that "this isn't fair" and then fail to produce any backup for the claim, but rile everyone up.
81 comments
[ 3.0 ms ] story [ 190 ms ] threadI was asking about the leaked spreadsheet which has the data on Google US's 200k employees.
"Those people make less because they are all lower level" is not a mitigating factor for pay discrimination.
You absolutely need to normalize for level when determine pay discrepancies. As well as hiring, retention, and promotion rates. It’s silly to suggest otherwise. Frankly a lot of companies fail to do this and are convinced they have huge problems when in fact things are pretty fair at the individual level and on track to gradually diversify over time. But you can’t just snap your fingers and diversify the top of an org. It takes a long time.
The significant restructuring (affected several hundred people) that you describe did in fact occur.
This idea is surely much more expensive to implement than just paying the women more. It’s also fairly easily discovered by an audit of any competence or whistleblower. And the women would surely have a strong negative reaction to this which is an operational risk of them just leaving.
What you are proposing happened is illegal discrimination to begin with, but it’s also really difficult to do. Seems more likely a willing company would prefer to simply lie
To some extent imo, provided an org isn’t overtly discriminatory, the most important thing is to make sure that hiring channels include diverse sources for candidates.
I would bet that the pay discrepancy here is entirely due to differing levels, which in turn is because Google's DEI efforts are ~5-7 years old and when new hires come in at L3, it takes that long or more to get to the high-paid levels. That and people who were hired a while ago are sitting on lots of appreciation in their stock grants.
There used to be. In the “member of technical staff” era, most new hires were down-slotted within a year.
They got rid of it. I think they changed it to "everyone comes in at L3/L4, and if you're good you'll rapidly get promoted to your true level" around 2012, and then sometime between 2014-2020 it changed to "sure, you can come in at L6+ and we don't really care if you're incompetent".
Leveling is likely the major component, but thats also a legitimate concern!
If you want to read a book that directly talks about this, check out Thomas Sowell's "Economic facts and fallacies". There are chapters for both gender and race economic differences. Even though the book is ~14 years old, it counters these kinds of arguments.
Has Sowell, or maybe one his acolytes, updated his rhetoric to incorporate the findings of all the ongoing research into the gender pay gap?
Let's say there is an employer who systematically wants to underpay women to save money by exploiting the pay gap. If there are no women in agreeing to the pay the employer will adjust upward in an attempt to either hire women or ignore the disparity, pay more, and appear bias.
I would argue that women who want more pay must discriminate more than their potential employer.
I see this pattern all over the place. Often with conservatives. Asking a reasonable question that puts the fundamental concept in doubt. But never actually wanting to know the answer because a simple search would have revealed lots of thorough and existing discussions answering the question.
You're joking, right? Otherwise why stop at AI? Why not go directly to the source and read Twitter?
What makes you think it's a joke?
Would this be the fallacy of an appeal to authority?
My point is simply that your question is valid, but has been asked and answered a thousand times before. Nothing in your question is specific to this audience that we'd have a twist on the answer that isn't represented in any of the thousand existing answers.
I'm curious, since you didn't get much discussion on your question here, did you take a minute to answer your question elsewhere?
I am not discussing whether this is fair, equitable, or even kind. I just do not care for the misrepresentation of the situation.
I could argue I am both overpaid or underpaid based upon my perspective. Having been on the other side of the negotiations with friends and strangers the situation is the same. They want as much value as they can extract. I too want as much value as I can extract - within agreement. Now that I am back to being an employee I have requested more compensation. They were open and asked why: apart from desire, I have not provided additional value (yet) and thus am not entitled nor deserving.
If it is true that women are likely to offer discounts, it questions why for-profit companies hire men at all? Meanwhile we worry about companies not being willing to hire women, not men.
Wiktionary - Underpaid: "Getting too little financial compensation for one's work." Meriam Webster - Underpay: "to pay less than what is normal or required" Cambridge Dictionary - underpay: "to pay (a person) too little" Dictionary.com - underpaid: "not paid enough"
Only the Cambridge version of those could imply (to me) only the definition of receiving less than what was agreed upon.
I personally like the Meriam Webster definition. If a software engineer is normally paid $ X, and you are paying $ 0.5*X, then that person is underpaid.
Exactly. There is nothing in there about agreeing to take less than you might have been able to get if you chose to charge more.
> If a software engineer is normally paid $ X, and you are paying $ 0.5X, then that person is underpaid.*
The general opinion is that people are not fungible. Why do you believe that they are?
Of course it is. People with more experience and higher performance get paid more.
If the accusation is instead that certain groups are being promoted less often because of their immutable characteristics, it requires some evidence to substantiate.
I personally feel like it is not the companies business to know this about me. Is there a legal requirement around this ?
Besides , is it anyone’s business as to what I identify as ? Is it ethical to force me to reveal this ?
The person leaking it is acting unethically though. It has no whistleblower value of any kind, and betrays those who filled it out thinking it would be internal only.
Maybe not the exact same spreadsheet, but the first one. It would be surprising if it was different though.
It is useful as far as it goes, but one can only draw limited conclusions from what is there.
Business Insider doesn’t really caveat it’s reporting very well here. And it should.
Perfect example is the report that "The employee with the highest base salary reported being a level 7 software engineer with a base salary of $718,000 and a total compensation package of just under $800,000."
I suppose there might be some engineer making $720k in base and $80k in bonus/equity combined, but I've never heard of such a thing. There are obviously errors in the data. Yet BI doesn't question the data at all.
Disclosure: I work at Google. Opinions are my own and do not reflect those of my employer.
I can't access the article, even via the archive link (infinite captchas), so idk what what BI is reporting. But the original sheet is employee-created, so identity stats are probably not coming from the company's own database.
Isn't this famously false, given that they pay large disparities depending on someone's country in which they work? This only holds true within a specific country.
Doesn't Google also try to dock your pay if you move out of a HCOL city, too?
Yet Google happily uses this opportunity to pay wildly different amounts to people doing the exact same work.
In any case, this clearly invalidates the first claim, that people are paid based on what they do. They're paid actually based on their replacement cost in the place they are legally restricted to do it.
To make a contrived example, this means that a woman in a LCOL country/city will earn less than a man in an HCOL country/city, for the exact same work.
All of this pay equity stuff is total nonsense and posturing as long as Google is taking advantage of labor market arbitrage (which their staff cannot easily do, not being multinational themselves).
If they truly believe in equal pay for equal work, it should plainly apply across national borders that staff cannot cross. Anything else is lip service.
The fact that some tiny lucky subset of people with very specific credentials can get permission to move across borders does not change this at all. (In fact, H1Bs generally making less than others for the same work reinforces my point.)
Whether that’s fair or not is a matter of belief system.
Google’s Bay Area compensation isn’t a result of some intrinsically just meritocracy - they’re a wealthy business competing with other wealthy businesses over an extremely high-demand labor pool.
You would think that anyone presenting statistics that call into question some kind of differential treatment by race, ethnicity, or gender would take a moment to point out the cautions in the data. Or even express some doubt about writing the story to begin with.
But no, for some reason, we are fascinated by the "0th order analysis" of inequities by race, almost want to see it blown up into incorrect interpretations and incentivize reporters to produce such stories.
I think there should be a code of ethics where reporters should be questioned about what they seek to accomplish with half-assed analyses like this, if they fail to accompany it with huge warnings about what it hasn't taken into account that could explain differences:
-- age
-- experience
-- selection biases in employees tenure (both hiring and departures)
etc.
Without this kind of info, all these stories do is shout an alarmist call that "this isn't fair" and then fail to produce any backup for the claim, but rile everyone up.