“There are lies, damned lies, and statistics…”

Sharing is Caring!

by Trivirus

With the recent media propaganda and “gaslighting” over the 3.9% figure, I’ve been thinking…

What other figures relevant to the job market are not only misleading, but possibly corrupt?

Let me give you an example of questionable data (that isn’t everyone’s second favorite target aka inflation): Salary/wage data.

The BLS compiles data about worker pay across jobs and industries. They use the NCS (National Compensation Survey) to gather the numbers, and then use various statistical formulas to calculate percentiles and “adjust for seasonality.”

But here’s the problem, and I quote…


Participation in the NCS is voluntary; therefore, a company official may refuse to participate in the initial survey or may be unwilling or unable to update previously provided data for one or more occupations during subsequent contact.

And here lies the heart of the problem.

The PROBLEM is, what employers say employees get paid is not the same as what employees actually get paid.

Not only can employers lie and “exaggerate” the figure to make themselves look better, but employers who don’t pay their employees as much are (in my estimation) less likely to participate in the NCS in the first place.

If this psychological scenario is too difficult for you to envision, let me give you an example.

See also  When There’s Too Much Money, There’s Never Enough Stuff

Let’s say a local health group commissions a survey to random adults in your area. They ask you to estimate how many grams of sugar you take in every day, on average.

The people who don’t consume a lot of sugar (<30 g) are not only more likely to answer, but are less likely to fudge their answer.

The people who DO consume a lot of sugar are more likely to lie on the survey or simply refuse to respond, because answering honestly would hurt their egos tremendously.


Likewise, I don’t think many employers want to “admit” that they pay employees below “market rate,” so they are more prone to fabrication.

Recall how a lot of companies jumped on the “Trump tax cut wagon” and promised wage/salary increases that did not materialize. (static.seekingalpha.com/uploads/2018/3/1/saupload_median-household-income-in-21st-century-yoy-growth-rate-200101-thru-201801_thumb1.png)

It’s MUCH easier to claim you do [X] than to actually do [X].

Whereas a company like Google would happily disclose that they pay their junior software engineers $140K, some 5 man shack on the edge of Silicon Valley would probably avoid embarrassing themselves by admitting that they only pay their code monkeys $80,000.

Now imagine this “effect” (where the Googles of the country happily and accurately disclose employee pay, and the 5 man shacks lie or refuse to participate), and do you see what the consequence is?

See also  New England Journal of Medicine now admit that there is no evidence that the vaccines are safe for pregnant women.

So why is this “employee salary data” an issue? Or how did this thought come across my mind?

Often times, I will subtly prod people to reveal how much they make.

There is a pretty wide discrepancy (thanks to market arbitrage/market failure…a basic concept in economics) in pay due to employers keeping employees in the dark.

Transparency would allegedly “destroy morale,” and American culture has made it taboo to talk about how much you earn (at least to people you know); this resulting information asymmetry greatly hurts workers.

(Thankfully I don’t give a shit about many social norms…a personal characteristic that has served me quite well tbh)

Due to my prodding behavior, I sometimes uncover bits of information, all of which suggest that the variance is higher and the “median” for most industries is slightly lower than BLS suggests.

Proposed solution (not that it would ever be implemented) to reconcile and fix corrupted data:

IMO, the best way to obtain the wage data is to actually sort through and collect numbers from tax returns, and use W-2’s. Employers are legally required to submit W-2’s on their end as well, which would eliminate the possibility of sampling/survey bias.


Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.