March 2018 Archives

Target your damned survey report

StackOverflow has released their 2018 Developer Hiring Landscape report. (alternate source)

This is the report that reportedly is about describing the demographics and preferences of software creators, which will enable people looking to hire such creators to better tailor their offerings.

It's an advertising manual, basically. However, they dropped the ball in a few areas. One of which has been getting a lot of traction on Twitter.

It's getting traction for a good reason, and it has to do with how these sorts of reports are written. The section under discussion here is "Differences in assessing jobs by gender". They have five cross-tabs here:

  1. All respondents highest-ranked.
  2. All respondents lowest-ranks (what the above references).
  3. All men highest-ranked.
  4. All women highest-ranked.
  5. All non-binary highest-ranked (they have this. This is awesome).

I took this survey, and it was one of those classic questions like:

Rank these ten items from lowest to highest.

And yet, this report seems to ignore everything but the 1's and 10's. This is misguided, and leaves a lot of very valuable market-segment targeting information on the floor. Since 92% of respondents were men, the first and third tabs were almost identical, differing only by tenths of a percent. The second tab is likewise, that's a proxy tab for "what men don't want". We don't know how women or non-binary differ in their least-liked preferences.

There is some very good data they could have presented, but chose not to. First of all, the number one, two and three priorities are the ones that people are most conscious of and may be willing to compromise one to get the other two. This should have been presented.

  1. All respondents top-3 ranked.
  2. All men top-3 ranked.
  3. All women top-3 ranked.
  4. All non-binary top-3 ranked.

Compensation/Benefits would probably be close to 100%, but we would get interesting differences in the number two and three places on that chart. This gives recruiters the information they need to construct their pitches. Top-rank is fine, but you also want to know the close-enoughs. Sometimes, if you don't hit the top spot, you can win someone by hitting everything else.

I have the same complaint for their "What Developers Value in Compensation and Benefits" cross-tab. Salary/Bonus is the top item for nearly everyone. This is kind of a gimmie. The number 2 and 3 places are very important because they're the tie-breaker. If an applicant is looking at a job that hits their pay rank, but misses on the next two most important priorities, they're going to be somewhat less enthusiastic. In a tight labor market, if they're also looking at an offer from a company that misses the pay by a bit and hits the rest, that may be the offer that gets accepted. The 2 through 9 rankings on that chart are important.

This is a company that uses proportional voting for their moderator elections. They know the value of ranked voting. Winner-takes-all surveys are missing the point, and doing their own target market, recruiters, a disservice.

They should do better.