Why Applicants Don't Volunteer Their Test Scores
By Bryan Caplan
Lots of great feedback in response to last week’s question, “Why don’t applicants volunteer their test scores?” I’m increasingly impressed by the wide range of first-hand education/job experiences; clearly the world is full of puzzles few economists have ever conceived, much less addressed.
Several respondents doubted the usefulness of applicants’ test scores. But this contradicts a large literature finding that IQ-type tests are among the very best predictors of job performance for a very wide range of occupations.
A few highlights from the comments:
1. When I graduated in 1998, a lot of high-test hiring companies like finance and consulting did ask for SAT scores.
4. HR departments hammer home that you’re not supposed to make decisions based on information you’re not supposed to know.
5. I have seen something similar happen, where applicants who were
disabled would volunteer information about their condition that
suggested they would not need any more sick time or “accommodations”
than a non-disabled person… Anecdotally it tended to
benefit the discloser.
I recently served as the hiring manager on a government job for which
most of the applicants were recent graduates. I received a few hundred
applications, of which 5-10 percent provided SAT or GRE scores. Those
scores typically put the applicant in the top couple percent of test
takers and were frequently from Ivy League schools.
In spite of recognizing their accomplishment, I took the scores as a
negative signal in the vein of Prof. Barnett. Did they have nothing
relevant to the job for which they were applying that they could be
using to fill that space?
As an entry-level actuary out of college I provided my (what I
thought were impressive) test scores, and I was hired (not as a result,
but not not as a result?)
I didn’t think it was socially awkward at the time, but that was more
my naivete than an astute observation. I subsequently took the scores
off of my resume.
Bob Knaus really made me think:
It worked for me, in 1996. As a high-school dropout with good tech
experience but no credentials, SAT scores of 750 verbal and 735 math got
me a position as a management consultant. Saved me all the time and
expense of a college education.
The decisions to show/not show scores is based entirely on the social
context for this. In industries like strategy/management consulting, it
is not uncommon (even for many partners) to show their GMAT/LSAT/SAT
scores. When you’re selling generalist “smarts,” you’re expected to use
anything to your advantage.
When you’re selling technical skills, you show OTHER qualifications –
certifications (look at how popular PMP certs are in consulting… I
have yet to find someone who was transformed into a better manager
because of it…).
I conducted a natural experiment a few years back on this. I was looking
for a job, competing against ivy league types. Initially, my call back
rate was about 5 percent. Once I put my GRE score on my resume, my call
back rate jumped to 80 percent (for targeted applications). I wouldn’t
argue that this works for all jobs, and now my experience trumps test
scores, but in the investment industry, it makes a difference.
First, I don’t believe you are allowed to rely on such information
whether it is volunteered or asked for. I once mentioned my immigration
status and another time I told my interviewer that I was married. (both
are protected categories) It was in passing but the interviewer’s
reaction was the same. They were obviously shaken and made a point to
tell me that it was not relevant to the process. Of course, a small
organization could make those decisions informally and take such
information into consideration. But an organization of any significant
size will need explicit hiring policies.
The question employers face isn’t “what the law says” but “what you can get away with.” I can believe it’s harder for large organizations to unofficially break the law. But even large organizations ultimately have human being(s) – not “hiring policies” make the final decision. On “obviously shaken” interviewers, I’ve heard virtually the opposite: Even lawyers applying for positions in labor law say that interviewers habitually ask illegal questions. Few humans outside of North Korea manage to scrupulously avoid forbidden small talk.
GMU’s own Dan Klein:
Suppose Sally is submitting a resume to employers. Suppose she has
really high SAT scores, and that she has skipped going to college.
Could Sally overcome the impropriety of including the SAT scores by saying:
“As I have not followed the usual path of going to college, I here provide my SAT scores …”?
“In equilibrium, then, we’d expect workers to do employers’ dirty
work for them. Since firms risk lawsuits if they ask for test scores,
applicants would volunteer their test scores without being asked.* When
workers don’t deliver their scores, would employers quietly assume the
worst and hire accordingly.”
Does this slight edit change the signaling?
In equilibrium, then, we’d expect workers to do employers’ dirty work for them. Since firms risk lawsuits if they ask for Facebook passwords, applicants would volunteer their Facebook passwords without being asked.* When workers don’t deliver their Facebook passwords, would employers quietly assume the worst and hire accordingly.
Interesting turnaround, but is it relevant? Are you saying that applicants don’t share their test scores because they feel it’s a massive invasion of their privacy? Doesn’t sound plausible to me.
Why is it “just not done”? I think because it breaches some sort of
signaling manners. You’re supposed to signal how intelligent you are,
but not quite so directly and blatantly, which may come across as
announcing that one is better looking than 85% of other people.
My immediate response to this seems to be no one’s else’s response: why
would you believe them? Schools get official ETS score reports.
Employers would get someone volunteering their test scores with no way
to confirm this.
A partial solution would be to attach a copy of your test results. Yes, it could be forged, but that’s a lot harder than baldly lying.
If I were a signaling skeptic, I’d look at all these results and say, “The signaling model is meaningless. Almost anything can signal anything, apparently.” But as Thaler says, I’d rather be vaguely right than precisely wrong. How on earth would a naive human capital story even begin to explain the first-hand experiences readers shared in the comments?