Are 96 Percent of Managers and Executives Extraverted?

sidebarIn various places I have seen the startling assertion that 96 percent of managers and executives self-identify as extraverts (in other words, that self-identified extraverts almost fully dominate management positions). I do not believe it. I have worked in education, publishing, computer programming, and counseling; most of my bosses, from managers to executives, tended toward introversion, at least in my perception. So I wondered whether this figure accounted for all fields–and where it came from in the first place. I decided to find out. I reached the conclusion that the 96% figure needs major qualification.

I found the explicit claim–along with a cited source–in a Harvard Business Review article by Adam Grant, Francesca Gino, and David A. Hofmann. A sidebar in the article states: “Whereas just 50% of the general population is extroverted, 96% of managers and executives display extroverted personalities.” The source: Deniz S. Ones and Stephan Dilchert, “How Special Are Executives?” Industrial and Organizational Psychology, 2009. This paper examines determinants of executive success by analyzing the scores of a sample of 4,150 managers and executives on a personality test and a test of mental ability.

I was able to access this paper through the Columbia library database. Unfortunately it doesn’t give any information about the sample; instead, it directs the reader to an earlier paper: “see Dilchert & Ones, 2008, for sample description.”

I tracked down the second paper (Dilchert and Ones, “Personality and extrinsic career success: Predicting managerial salary at different organizational levels,” Zeitschrift für Personalpsychologie, vol. 7 [2008], 1–23.). I had to pay a fee for it, but I wasn’t going to stop now. Here’s what I found (on p. 6):

Participants were 4,150 individuals who completed a personality inventory as part of an assessment center. Of the total sample, 1,819 individuals were applicants to a managerial position and 151 individuals were considered for a promotion, and thus completed the personality inventory under selection conditions. In addition, there were 2,180 managerial job incumbents who completed the inventory for developmental purposes. All participants also completed a demographic form and provided information on their current employment status and employment history.

Wait–so the participants were taking the assessment for a job-related purpose: for employment, promotion, or development. The results were then used as data. The stakes were high, in other words (especially for those applying for employment and promotion), and their responses were initially not anonymous. This could well have influenced the responses.

Moreover, they all took it at a particular assessment center. This suggests to me that certain professions were not included: professors, academic administrators, principals, artistic directors, librarians, computer programmers (who manage their own teams), self-starting entrepreneurs, head physicians, and others.

In addition, the personality test was the Global Personality Inventory, which is geared specifically toward the workplace. Scores on this test may or may not correspond with scores on a Big Five inventory. The 2008 paper states:

The GPI is a thoroughly developed inventory backed by empirical evidence that supports its reliability and criterion-related validity for use in managerial assessment (see Schmit, Kihm, & Robie, 2000). Reliabilities for the Big Five facet scales typically have been reported to range between .65 to .88 (Schmit et al., 2000) and .58 to .88 (ePredix, 2001) in managerial samples (mean reliability across scales .73 and .75, respectively).

In neither of the papers do I see the figure of 96 percent; perhaps Grant, Gino, and Hofmann extrapolated it from the normative data and data on variability. Let us assume, though, that the figure accurately reflects the test results. It does not reflect the managerial and executive population as a whole, for three reasons:

  1. The test seems to have carried relatively high stakes (in comparison to a test administered purely for a study);
  2. The test was administered at an assessment center that may not be used by all professions and fields–thus the sample may be skewed;
  3. The instrument itself is designed specifically for the workplace; the extraversion score may not match scores on other personality tests. In particular, the questions may involve more context-specific details.

So, instead of saying that 96 percent of managers and executives display extraversion, I recommend saying, “On a Global Personality Inventory administered, at an assessment center, for employment, promotion, and professional development purposes, 96 percent of managers and executives gave responses suggestive of extraversion.”

One implication: The extent of the tilt toward extraversion in management may depend strongly on the field.

Another implication: It is important to look into claims of this sort.

Image credit: Adam Grant, Francesca Gino, and David A. Hofmann, The Hidden Advantages of Quiet Bosses,” Harvard Business Review, December 2010.

Update:  People continue to cite this misleading figure.

Your Personality, Your Noise

There are far too many unsupported and overhyped statements about what “science tells us” about introverts and extraverts. This distorts the dialogue and affects school and workplace policy. I take up this subject both because it overlaps with some of my interests and because it bears examination. “Science tells us” statements have popular appeal, a big market, and numerous high-profile outlets. They need vigorous questioning.

When it comes to introversion and extraversion, the findings are far less definite than pundits claim. Any blanket statement about introverts and extraverts needs unblanketing. (I see no need to call anyone an introvert or extravert in the first place, but that’s another matter.)

Here is an example. In an interview with the Harvard Business Review, Susan Cain says,

And just to give you kind of a concrete illustration of [how introverts and extraverts work differently], there’s this fascinating study that was done by the psychologist Russell Geen, where he gave math problem [sic] to introverts and extroverts to solve with varying levels of background noise. And he found that the introverts better [sic] when the noise was lower, and the extroverts did better when the noise was higher.

I can’t find that study anywhere, but the statement alone contains some problems, which I’ll lay out in just a moment. I did find Geen’s 1984 study “Preferred Stimulation Levels in Introverts and Extraverts” (Journal of Personality and Social Psychology, vol. 46, no. 6), and I found it precise and cautious in its wording.

This study consisted of two experiments. Subjects (all male) were selected from the upper and lower 25 percent of the Extraversion scale of the Eysenck Personality Inventory. They were undergraduates enrolled in psychology classes.

For the first experiment (which involved 30 extraverts and 30 introverts), Geen sought to determine whether extraverts and introverts, given the freedom to choose a noise level (other than zero), would differ in their choices, and whether such freedom of choice would equalize their arousal levels. There was a “choice” condition and two control conditions: one where the subject was given the noise level chosen by another (“yoked”) subject with the same personality type, and another where the noise level was selected by the experimenter. The subjects took this test one at a time. After selecting their noise level (or having it selected for them), they had to  wait two minutes. Then the projector was turned on, and they worked to complete a paired-associates learning task. The learning task ended when they completed two successive errorless trials. If, after twelve trials, they had not achieved this, the procedure was terminated.

The second experiment had more subjects (40 extraverts and 40 introverts) and a slightly different approach. This time, in addition to selecting their optimal noise level, introverts indicated the lowest acceptable noise level, and extraverts the highest. This established the four intensity levels used in the study. There was only one control condition: where the experimenter chose the noise level for the subject. (The “yoked” condition was eliminated because, in the first experiment, its results did not differ from those of the choice condition.) Otherwise this experiment followed the same procedures as the first.

I will focus on the results of the second experiment (because I have a bit more to say) and because, in Geen’s words, it provided “a replication and extension of the first.” With respect to pulse rate, something interesting came up: introverts and extraverts did not differ in pulse when they had chosen their noise level. When subjected to moderate noise, introverts were more aroused (i.e., had a higher pulse rate) than extraverts, but at the extreme noise levels (high and low), their rates did not differ from each other. This, right away, casts doubt on the blanket assertion that introverts are more sensitive to stimuli than extraverts. The statement needs careful qualification and questioning.

When it came to the task, introverts and extraverts performed equally well at their chosen noise levels. When subjected to superoptimal noise levels, both introverts and extraverts did significantly worse. When subjected to suboptimal noise levels, extraverts did significantly worse, but introverts did not (or they did worse, but not significantly).

This is interesting (and again, carefully thought out and presented), but I see a few caveats here. (And I haven’t forgotten about the Cain quote; I’ll come back to it in a minute.)

First, there are problems with selecting the top and bottom 25 percent on the extravert scale; your subjects are already at the extremes. How much of this applies to a full population is uncertain. That said, if you didn’t do that, you’d probably end up with so much noise (in the data) that you couldn’t draw any conclusions.

Second, I wonder to what degree the Eysenck Personality Inventory already relates to noise tolerance. That is, are these subjects defined as introverts/extraverts partly on the basis of their reported tolerance of noise? That would make the experiment somewhat redundant.

Third, I wonder to what extent the particular kind of noise influenced the results. (These were one-second bursts of white noise, with a mean of ten seconds between bursts.) I can imagine the high levels being particularly jarring (to anyone), and the low levels annoying (that pesky sound you can’t get rid of).

There are still more open questions; to begin to address them, I would need more statistical  knowledge and access to the raw data. To his credit, Geen does not draw rash conclusions from this study; at the end, he offers possible implications and describes the work that still needs to be done.

So I come back to the Cain quote, in particular: “And he found that the introverts better [sic] when the noise was lower, and the extroverts did better when the noise was higher.”

She could not have been talking about this study, because that was not the finding. There must be another  study that I haven’t located yet. Even so, a person making such a statement should specify the following:

  1. Which study is this? Give identifying information.
  2. How were the introverts and extraverts selected and defined?
  3. What kind of noise was used?
  4. What were the math problems, and what do you mean by “did better”?
  5. Was there any complexity/contradiction to the findings?
  6. Did the author bring up any caveats (and do you see any)?

That’s a lot, I know, and in an interview it might not be realistic. But that’s only scratching the surface; an expert should be able to do things I myself can’t do: for instance, explain the methods used for interpreting the raw data.

What I see instead (and not just from Cain by any means) is a tendency to oversimplify and exaggerate the results of studies, and to do so again and again.

Some of these studies are interesting and valuable. Others are bunk. Still others are works in progress. All of them have limitations, but when taken skeptically and cautiously, they can help us reach greater understanding.

The implications? I see no problem with the idea that different people work well at different noise levels. But reducing the matter to “introverts” and “extraverts” is unnecessary and unfounded. Much depends on the individual, the context, the particular task, and the type of noise. There are situations that call for quiet, situations that call for noise, and a range in between. While a workplace should probably establish basic quiet (so that the noise doesn’t get out of hand), people can learn how to handle both quiet and noise in reasonable degrees.

Note: I made a few minor revisions after posting the piece–and then made a few more a year later. Also, Science of Us (New York Magazine) has started challenging some of the pop-psychology assertions about introversion and extraversion. (Here’s another piece on the topic.)