SUBSCRIBE E-NEWSLETTERS AWARDS COLUMNS MULTIMEDIA CONFERENCES ABOUT US RESEARCH
Reader Feedback

Thursday, October 18, 2007
Write To The Editor Reprints

RE: Peter Cappelli's article suggesting that personality tests may not be so pretty good in predicting success on the job. Looks like we've come to our senses . . . again.  

Joe Rousseau SPHR

Human Resources Director

Pines of Sarasota

* * * * *

The article that Peter Cappelli referred to in Personnel Psychology by Morgenson and the other editors was very interesting. They railed against the type of personality test in which somebody tells you their own views about their own personality -- a process that is contaminated by all sorts of biases that make such tests very inaccurate.

This kind of test is called a self-report personality test, and the editors wrote a scathing evaluation of them. However they fell short of condemning all personality tests.

They were much more positive about what are called "implicit" personality assessments. In fact they specifically mentioned a type of personality test that doesn't suffer from those biases and which works very well in predicting performance -- 200 to 300 percent better than self-reports. 

It's called a "Conditional Reasoning Test," and it was developed by an extremely well-respected psychologist and scientist down at Georgia Tech -- Dr. Larry James. 

I'm not by any means a fan of self-report personality tests. But before you write off all personality tests, its probably worth looking into Dr. James' approach to assessing personality.

It's kind of new; it's only been around 10 years or so, and most people in the field -- other than the cognoscenti like these editors -- don't know much about it yet. The approach works exceptionally well, so it's worth investigating. 

Charlie Brooks

Project Director: Organizational Performance

State Personnel Administration

Atlanta

* * * * *

. . . but what would the companies pumping out paper-based and computer-based personality tests do?

Seriously though, Peter Cappelli has quite properly brought to our attention once more the fundamental problem with personality testing and work behaviour -- the results simply don't predict behaviour! 

My own research at the turn of the century looked at the relationship between personality and managerial style and demonstrated that there is no relationship between the two, even though the personality test in question was trumpeted as one of the most widely used management selection tests in the world.

This highlights management's determination to find a "silver bullet" solution for to the selection problem -- how to make valid and reliable predictions about work related performance in the context of selection decision-making.

Thanks Peter -- you do a continuing service.

Professor David Lamond PhD FANZAM

Founding Dean, Kochi International Business School

Academic Advisor to Emerald Group Publishing Ltd.

Editor, Journal of Management History

* * * * *

Dear Peter, 

I have attended your classes in Miami (SHRM) in 2004. After having spent 30 years in HR across the many regions in Asia and Middle East, I still feel the best way to select is based on three factors -- which school/college, worked in a company known for its values and your GUT impulsive decision to have him as a part of your team.  

Antony Thomas

Regional Human Resources Manager, Middle East

Bovis Lend Lease

Dubai, United Arab Emirates  

* * * * *

Dear Editor:  

Two recent articles appearing in Human Resource Executive Online ("HRE") entitled "Assessing Personality" and "More on Selecting Employees" do a very poor job of providing your readers with an objective and balanced view of personality and intelligence testing. As a result thereof, I believe your readers would be interested in the following information.  

The articles indicate that there is a "new consensus that personality tests are a poor predictor of future job performance." As an industrial psychologist this so-called consensus is news to me, as well as the vast majority of my colleagues.

Rather than exclusively relying upon the analyses and opinions of only six industrial psychologists, a balanced article would have acknowledged that the findings of these six psychologists had been extensively criticized by many psychologists with equally as impressive credentials. 

Additionally, the field has a wealth of research documenting that personality testing -- especially assessments focusing on conscientiousness -- are extremely effective tools for predicting various forms of counterproductivity.  See "Comprehensive Meta-Analysis of Integrity Test Validities," Journal of Applied Psychology, 1993 and "The Truth About Integrity Tests: The Validity and Utility of Integrity Testing for the Hospitality Industry," Cornell Hospitality Reports, 2007.   

In the article "More on Selecting Employees" there is an acknowledgement of the utility of intelligence tests, but then there is also a misguided discussion about the how little performance variance such tests account for.

The article's approach of squaring the exhibited correlation coefficients  (commonly criticized as a means of minimizing the perceived impact of certain variables) led to very small numbers in terms of accounted-for employee performance variance -- certainly an approach and argument that could be utilized for any other selection measure as well.  

However, the article fails to recognize the simple fact that intelligence tests exhibit higher correlations with job performance than most of the selection tools used by employers (e.g., interviews, education, drug tests, background checks).

Newsletter Sign-Up:

Benefits
HR Technology
Talent Management
HR Leadership
Inside HR Tech
HRENow
Special Offers

Email Address



Privacy Policy

Further, as any well-trained industrial psychologist is well aware of, the correlation coefficient does not exclusively determine the ultimate utility of a test or other selection measure. Other factors, such as the selection ratio and the base rate of the employee behavior being focused on, impact the ultimate utility of the selection device.

Hence, a selection tool with a correlation coefficient of .20 can exhibit utility in certain circumstances. The Taylor Russell Tables that industrial psychologists commonly use provide significant insight on this matter.  

Finally, an excellent overall review of the validity of common selection measures can be found within an article from an edition of the 1998 Psychological Bulletin -- "The Validity and Utility of Selection Methods in Personnel Psychology: Practical and Theoretical Implications of 85 Years of Research Findings."  Therein, it is acknowledged that research shows that intelligence, integrity and conscientiousness tests have high utility and validity for predicting job performance.  

While certainly by no means exhaustive, these comments should be useful to your readers in gaining a much more objective understanding of the facts regarding personality and intelligence testing. I trust future articles in this domain will be a bit more objective, and hence more informative.    

David W. Arnold, Ph.D., J.D.  

General Counsel 

Wonderlic, Inc.  

Peter responds:

One of the great things about writing this column is to find out about all the different vendors who have a financial stake in the issues I discuss.

Here are just a couple of reminders about what I said in those two columns. The conclusions about personality tests not being useful are not mine but those of six prominent academic personnel psychologists who are also journal editors.

They reviewed all the studies David Arnold mentioned that support personality tests as well as others that he does not mention that do not support their use. And these experts have one important attribute that gives them some special credibility in this debate: None of them sell selection tests.

With respect to the column on intelligence, what I said was that intelligence tests are the item most undervalued by practitioners and that they are one of the easiest, cheapest and most reliable assessments to use.  

It is still the case, however, intelligence does not explain very much of the variation in job performance. 

David Arnold, whose firm is one of the major providers of intelligence tests, objects that I don't praise intelligence tests enough. He is certainly right in noting that even a test that doesn't predict much about job performance can under the right circumstances have a payoff as compared to just selecting randomly or using less valid tests

Under other circumstances, however, the payoff could be trivial. Readers will not be surprised to hear that there is no magic bullet for selecting employees.  

Copyright 2014© LRP Publications