Personality testing part I -- Michael rethinks use of personality tests for hiring

(FAKE INKBLOT FROM FAKE RORSCHACH TEST)

Michael has previously written in this blawg about personality testing and employee selection (see here and here)

At the time of the earlier of his posts, he and I had some interesting conversation on the subject, and I drafted a lengthy post that I never posted.

Now Michael has written about personality testing again, this time in Electronic Recruiting Exchange.

Click over for the link to his article and my nutshell version of it. Some more of my own thoughts and some other pertinent links on the subject will follow in later part(s).


Michael's article is entitled "The Cult of Personality: Rethinking the Use of Personality Tests for Hiring."

Michael starts with a reference to the book "The Cult of Personality," by Annie Murphy Paul, which he describes as "present[ing] a scathing criticism of the use of personality testing in a variety of settings, including the workplace." [Amazon editorial reviews here, and some very interesting customer reviews here]

[Strange coincidence: the next item on my Amazon search for this book was this title: "Soviet Leaders: From the Cult of Personality to Collective Rule." Beware that cult of personality!]

According to Michael's read of this book, some of the folks in the personality test business had -- well -- kind of weird personalities: "The man who developed the Rorschach Inkblot Test, Hermann Rorschach, may have died of heartbreak from the failure of his test to gain wide acceptance. . . . Starke Hathaway, the creator of the MMPI, sometimes wore shoes that didn't match and frequently came to classes with grease stains on his clothing."

Beyond such trivia, Michael's article addresses four of the author's assumptions about the use of personality tests for hiring and selecting employees.

Assumption #1: Personality tests do not predict job performance.

After explaining "criterion-related validity" studies, Michael says that although "for many years, academic researchers were skeptical of the predictive validity of personality tests in the workplace," researchers have now concluded that personality tests do have "some, albeit limited, predictive validity in the workplace." Specifically, scores on "conscientiousness" have been found "modestly correlated with job performance." Also, paper-and-pencil honesty or integrity test have been found "correlated with a broad array of workplace behaviors, including counterproductive activities and general job performance." Michael concludes:

Hence, there is solid scientific evidence that some personality dimensions do in fact predict job performance to some extent. Nonetheless, combining invalid scales with valid scales (e.g., conscientiousness) to predict job performance may actually reduce the predictive power of the valid tests. The bottom line here is, as with many things in life, personality testing brings value in selection if used appropriately, responsibly, and in an informed manner.

What does the comment about combining valid scales with invalid mean? I suppose that if you rely on test scores on factors such as "openness to new experience," or "extraversion" -- which might seem relevant to particular jobs but have not been proven valid predictors of performance -- you may be eliminating very conscientious people whose strength on that factor would more than compensate for any weakness their scores on the others might suggest (perhaps improperly).

Assumption #2: There are better ways to select people for jobs than to use personality tests.

Discussing alternatives to personality testing for the hiring process, Michael finds no magic bullet:

Cognitive ability tests, for example, might be used, but they too have weaknesses. . . [T]hey too tend to be far from perfect predictors of job performance . . . A second weakness is that they tend to produce disparate impact, which makes companies . . . susceptible to discrimination charges. Other tests, such as assessment centers, are expensive to develop and use, and also are far from being perfect predictors of future job performance.

A second potential alternative is to rely more heavily on interviews. But interviews are far from perfect predictors and they may be fraught with biases and subjectivity. . . . Interviews have one more limitation: while a relatively inexpensive test might be administered to large numbers of applicants, few companies are willing to interview hundreds of candidates to weed out the poor ones. Thus for positions where there are many applicants, a personality test might be an effective first step . . .

Assumption #3: A large number of honest people fail integrity tests.

Michael says:

Such tests do . . . show a modest degree of predictive validity, using criteria such as performance ratings and even theft on the job.

The more contemporary "integrity tests" of today measure a broader scope of behaviors. These are often referred to as "tests of conscientiousness." While they indirectly predict dishonesty, these tests evaluate "work values" such as showing up to work on time, following rules, taking responsibility for getting work completed, and risk-taking on the job.

Assumption #4: Job applicants have few legal rights when it comes to personality tests.

Michael disagrees to some extent, pointing out three areas of possible rights: 1) protection from racial, gender, and other related types of discrimination, which may allow challenges to use of personality tests if they have a disparate impact on a protected group; 2) privacy laws, which may allow challenges based on invasiveness of questions or perhaps entire tests; and 3) the Americans with Disabilities Act (ADA), which may provide some protection for job applicants from the use of certain tests such as the MMPI that are designed to disclose mental illness.

Michael concludes:

From an applicant perspective, these and other laws may be seen as insufficient protection from the arbitrary use of personality tests. From a company's perspective, on the other hand, such laws may be seen as a potential deterrent in using personality tests.

I agree with the applicant perspective -- in many cases it will be extremely hard to prevail in a lawsuit based on such theories. But I also agree with the company perspective, and would also strongly caution against ignoring the possibility of such legal challenges. Even a losing employee lawsuit (in terms of factual and legal merit) is often a winning lawsuit to the extent the employee's goal is forcing a settlement or at least forcing the employer to pay for extensive -- and expensive -- legal defense.

I think it is fair to say that some employers may be using personality tests without sufficient consideration of either their effectiveness or their legality. One issue I hope to think through and write on further in the future concerns disparate impact. (I've not forgotten that months ago I started a series on disparate impact -- the post on Griggs has been one of my most popular -- this is just a thorny subject and one I haven't studied in a while, so it's taking a while to get back to it.)

Maybe Michael can answer my questions (but he'll probably just say I'm the lawyer!)

Suppose a particualr test is guaranteed by the vendor not to have a disparate impact -- and impressive studies support this claim. Then suppose the employer's overall hiring/promotion rates, and/or the resulting composition of the workforce, are challenged as statistically reflecting suspicious disparities.

Under what circumstances will this require the employer to justify the validity of the test(s) used in the hiring/selection process? If a personality test thus comes under scrutiny, will its lack of disparate impact clear it, or must it then be shown to be valid since it is part of an overall process having a disparate impact? If validity must be proven, I'm skeptical courts will be happy with Michaels's "modestly correlated with job performance."

Having said that, I'm 100% with Michael on these Conclusion and Suggestions:

The uncritical use of personality tests in the workplace is foolish at best, and at worst, may be costly to companies. If you are either currently using personality tests, or considering the use of these tests for hiring and selection decisions, here are some suggestions:

1. Make sure the personality test is properly validated.

2. Review the test for possible legal problems, including potential disparate impact, possible privacy violations, ADA concerns, and other legal violations.

3. Review all test items so you are sure they are not offensive or leave a negative impression with job candidates.

4. Make sure you clearly understand how to use the results of the personality test. Contrary to the achievement tests that most of us are commonly exposed to, higher scores are not always better scores. Some jobs may require employees who score low on the attributes measured by these tests.

Finally, personality tests are not an automatic panacea for hiring problems. Learn what strengths they offer and make sure that you aren't missing out on other, more effective testing options.