The Gender Pay Gap & Empirical Facts

“It’s not the things we don’t know that get us into trouble. It’s all the things we know that just ain’t so.” –Unknown (often attributed to Mark Twain)

File this under “Political disagreement due to objective mistakes of empirical fact.” I think most political partisans have no idea how much this happens. As a case in point, I would guess that most feminists assume that their opponents are just bad people who hate women (“misogynists”), while anti-feminists tend to assume that feminists are bad people who hate men.

An interesting alternative to both views is that the main difference between feminists and their opponents is a matter of beliefs about matters of objective, empirical fact.* And wouldn’t it be interesting if there was already enough evidence to resolve most of those questions of empirical fact, but one or both sides simply don’t know the empirical evidence?

*Caveat: It may be that the errors of empirical fact are themselves motivated, so that the political disagreements are indirectly due to differences in personality or values. That is an entirely plausible possibility to combine with this thesis. Whether or not that’s so, the best remedy is probably to just examine the empirical facts.

No doubt there are usually experts (of either ideological persuasion) who know the relevant evidence on both sides of a given issue. But I have in mind the average partisan, not a super-educated partisan.

In the case of feminism, there is a factual question about how much women in U.S. society are disadvantaged due to sexism or “patriarchy”. That question turns on lots and lots of more specific questions. Feminists would cite a multitude of different ways that women are disadvantaged, allegedly due to sexism. It’s impossible to examine all of them.

Aside: The Self-Support Trap: This is an insidious phenomenon that happens with belief systems. Let’s say you already believe sexism is rampant. Then you hear about a new example of something that could be interpreted as sexism, or could be interpreted as something else. Since you’re already convinced that sexism is rampant, you interpret the example as sexism; perhaps this interpretation just strikes you as obvious and no other interpretation even occurs to you. You then add that example to your list of instances of sexism, which further strengthens your general belief that sexism is rampant, which you then use to help interpret the next ambiguous example, and so on. This is how political polarization occurs. People who start with different initial opinions are just going to keep ratcheting up their confidence in whatever are their preferred generalizations.

Let’s just consider one example of the patriarchy. Perhaps the most famous example of the rampant, sexist bias in our society is the gender pay gap. Everybody has heard the statistic that women in the U.S. earn 79 cents for every dollar earned by men. (https://money.cnn.com/2015/04/13/news/economy/equal-pay-day-2015/) (The figure has varied over the years, generally getting smaller.)

Obvious explanation: Men are underpaying women (or overpaying men) out of sheer sexist prejudice. When I first heard the statistic many years ago, that’s what I assumed. I also assumed that the statistic was “Women earn $0.79 for each $1.00 that men earn for the same work.” I bet these are the same things that almost everyone initially assumes when they hear the statistic, and what most people who quote the statistic are also assuming and intend for you to assume.

Much later, I found out that that is, as a matter of objective fact, wrong. The part in bold is wrong – the “79 cents” statistic does not control for what kind of work the men and women are doing. It just takes the average wage of all women and divides by the average wage of all men, regardless of what occupations they’re in, their level of training, their degree of seniority, etc., etc.

Aside: We should have been pretty suspicious about the 21 cent gap to begin with. If it was really what it sounded like, that would mean that a savvy employer could hire all women to staff his firm, get the same work done, and save 21% on labor costs compared to a male-dominated firm. That would be a huge market advantage! Why wouldn’t companies be taking advantage of that?

I then learned that, if you control for things like choice of occupation, hours worked, training, and seniority, the 79 cents goes up to 95 cents.

Now, you may say, “A 5 cent gap is still evidence of sexism. Any gap at all is unjust!” But I think we should first pause for a moment to note how seriously misleading it is, if the actually relevant gap is 5 cents, for people to go around (as they have been doing for decades) reporting a figure that is at least four times higher. Once you find that out, I think you should start distrusting every other factual claim that you hear from the source that gave you the “79 cents” figure. It looks like they’re distorting the facts to try to maximize outrage.

Also, the smaller the gap is, the more likely it is to be due to some other factor (besides sexism) that we haven’t thought of yet and thus haven’t controlled for. So reducing the number reduces the credibility of the qualitative conclusion as well.

I recently learned that there’s evidence that that is in fact the case, that even the 5 cent gap may be overstated. In fact, the sign of the gender pay gap may be misstated; it may be that men actually earn less than women, after we control for everything that needs to be controlled for. The political scientist Warren Farrell found 25 different factors that affect pay rates (including such things as number of hours worked, willingness to do physically dangerous work, willingness to relocate, willingness to do technical work without people contact, etc.). After controlling for all 25 factors, he reports, the gender pay gap not only disappears but reverses: women actually earn more than men do for the same work (https://www.amazon.com/dp/1542751292/). He finds, e.g., that never-married women without children make 13% more than never-married men without children.

In case you’re tempted to dismiss this as some misogynistic men’s rights activist, I note that Warren Farrell began as a hardcore feminist activist, the only man to be elected three times to the Board of Directors of the National Organization for Women. He was initially convinced (just as most people are when they first hear about it) that the gender pay gap was evidence of unacceptable sexism. It was only when he started to do serious research on it that he began to realize that this was simply not true.

Objections

1. “That’s ridiculous; there obviously isn’t sexism against men.”

See my earlier aside about the self-support trap. If you’re tempted by objection 1, that suggests that you’re using your general belief to interpret specific phenomena. If you do that, it’s easy to see how your whole set of beliefs on the subject could be radically mistaken.

“But given the well-known history of (anti-female) sexism, how could it plausibly have turned out that today there’s sexism against men?” First, note that it’s not clear that there is sexism against men either. It’s possible that there are some additional factors we haven’t thought of (besides Farrell’s 25), such that when you control for those, the pay gap either disappears or reverts to favoring men. But obviously, this speculation doesn’t enable one to claim the gender pay gap as evidence for the patriarchy in our society.

But anyway, I don’t think it would be so weird if there was anti-male bias in the labor market. Most people of both sexes like women more than men (https://www.apa.org/monitor/dec04/women). In addition, many people believe that women are systematically underpaid and that this is unjust, so they may try to compensate for that by paying their own women employees a little more than they otherwise would. So these are two ways that a reverse gender pay gap might come about.

2. “It’s unfair or otherwise bad that women don’t do those 25 things that get you more money. Society is oppressing women by teaching them not to do those things.”

Okay, money is good, but it’s not the only good. The 25 things get you more money, not because employers just randomly pick some things to attach extra pay to. They get you more money because they are intrinsically undesirable. The extra pay is compensation; the employers pay extra because they have to pay more in order to get people who are willing to do them.

E.g.: working in dangerous jobs. You have to pay extra to get people to risk their lives on the job. Men do that more, which gets them more money but also causes them to suffer >90% of the workplace fatalities. In this case, it’s pretty clear what the fallacy is in claiming that women are disadvantaged because they’re not doing the income-maximizing thing: they’re not getting the extra money, but they’re also not getting the extra bad thing that that money was there to compensate for.

Second example: Women are more likely to prioritize family life over making money. This is not due to some nefarious brainwashing by society designed to harm women. If anything, it's the men who are brainwashed into sacrificing their best interests. Who says that maximizing profit is the key to a good life?

The same point applies to all the other income-maximizing choices. So it would not benefit women to try to convince them to do all those things so they can get more money.

3. “Maybe those 25 factors increase your pay only because men tend to choose them.”

No, when you look at the factors, it’s intuitively clear why each thing would get you paid more (apart from sexism). E.g., you have to pay people to compensate them for being willing to risk their lives; or to relocate out of their home state; or to work very long hours; or to get special training; etc.

4. “This doesn’t address other examples of sexism.”

No it doesn’t. That’s because it’s impossible to address everything. Each individual example needs to be looked at carefully, which takes time, and I’m not going to write a whole book in a blog post.

But I would just note, (a) the gender pay gap is a significant example. I think I’ve heard it the most often as an illustration of the patriarchy, and it has perhaps the most immediate, intuitive appeal to people. So it’s totally fair to pick this as an example to examine. (b) If the one example you took time to examine carefully turns out to be bogus, you should not be highly confident that all the other examples are just fine and totally non-misleading.

What's the lesson?

I see this as an illustration of a larger phenomenon: Most of the key "facts" that dramatically support a particular political ideology are false. Your ideology is probably based on a bunch of seemingly compelling statistics and anecdotes that are simply false or radically misleading.

Why is this? Because political memes are optimized for provoking emotional reactions and supporting a particular political ideology. It is unlikely that the optimal claim for supporting a particular ideology is also going to be highly accurate; usually, you can find something more effective at supporting your viewpoint if you relax the constraints on truthfulness. It only takes a few people to originate those memes; then there's a kind of evolutionary process in which the most accurate ideas are regularly outcompeted by the less truthful but more provocative memes.