PART 3
>All objective scientific evaluations conclude that no psychometric personality test has ever achieved a correlation between predicted and actual occupational performance greater than 0.3. Professor Paul Kline was a professor of psychometrics at the University of Exeter until his death in 1999.<
I have read several of his books and have discussed his books and ideas with one of his proof readers, also a PhD psychologist. You are either misleading on purpose or you don’t know any better. Here are the correlation coefficients I have seen reported.
0.00 for a Flip of a Coin
0.14 for an Interview Only
0.26 after adding Background Checks
0.38 after adding Personality Testing
0.54 after adding Abilities Testing
0.66 after adding Interest Testing
0.75 after adding Job Match Testing
plus a few random ones...
0.18 for Job Experience (years)
0.38 for Unstructured Interviews
0.41 for Integrity Tests
0.51 for General Mental Ability Tests
0.51 for Structured Interviews
0.54 for Work Sample Tests
The secret is to use more than one method.
>In a magazine article written shortly before his death Kline says -
“A huge analysis of the predictive power of tests in occupational psychology by Schmidt & Hunter (1998) indicated that while intelligence was a relatively good predictor of occupational performance, with correlations of about 0.6, personality tests were of far more limited value, with correlations rarely exceeding 0.25. This lack of predictive power again suggests that all is not well with psychometric measurement.”<
Assessment users know that when they use both assessments the correlation is above 0.6, see my list above.
>For those unfamiliar with correlations, 0.6 is 36% and 0.25 is 6.25%.<
Do you know why? I’m educated in both engineering and business and I find your rudimentary presentation on correlations deceptive in that it does not go far enough. Copying and quoting does not substitute for understanding.
>This statement from a professor of psychometrics is clearly at odds with the wildly extravagant and exaggerated claims of many in the HR industry.<
Actually it isn’t at odds but then I again understand correlations and how to use them in business and engineering.
>Many disbelieve me when I quote Kline and I am frequently accused of making this quote up so for the doubters, the quote comes from an article in…<
Quoting an expert does not make you an expert in the subject matter quoted.
>We now each have a decision to make. Which evidence about the reliability of psychometric testing is more credible? The evidence offered by those selling psychometric tests or the evidence from the pen of a professor of psychometrics.<
What is the correlation if we interview, perform background checks, administer a personality test, assess for abilities, assess for occupational interests, and then perform a job match? The answer is found in the list above.
>From my own impartial position the decision is easy.<
You are not impartial just because you claim impartiality. And worse than that even an impartial person can be wrong. Impartiality does indicate correctness.
>For those deeply involved in HR, the decision is one that will cause considerable anxiety and cognitive dissonance for many years to come.<
Where is the evidence?
>… Psychometrics may appear a magnificent edifice to some but it has very weak foundations.<
Psychometrics is a tool that when used appropriately adds value.
>Good luck to all of you with this thorny problem.<
It is a thorny problem only if a wrong assessment is used or the right assessment is misused.
Bob Gately
>All objective scientific evaluations conclude that no psychometric personality test has ever achieved a correlation between predicted and actual occupational performance greater than 0.3. Professor Paul Kline was a professor of psychometrics at the University of Exeter until his death in 1999.<
I have read several of his books and have discussed his books and ideas with one of his proof readers, also a PhD psychologist. You are either misleading on purpose or you don’t know any better. Here are the correlation coefficients I have seen reported.
0.00 for a Flip of a Coin
0.14 for an Interview Only
0.26 after adding Background Checks
0.38 after adding Personality Testing
0.54 after adding Abilities Testing
0.66 after adding Interest Testing
0.75 after adding Job Match Testing
plus a few random ones...
0.18 for Job Experience (years)
0.38 for Unstructured Interviews
0.41 for Integrity Tests
0.51 for General Mental Ability Tests
0.51 for Structured Interviews
0.54 for Work Sample Tests
The secret is to use more than one method.
>In a magazine article written shortly before his death Kline says -
“A huge analysis of the predictive power of tests in occupational psychology by Schmidt & Hunter (1998) indicated that while intelligence was a relatively good predictor of occupational performance, with correlations of about 0.6, personality tests were of far more limited value, with correlations rarely exceeding 0.25. This lack of predictive power again suggests that all is not well with psychometric measurement.”<
Assessment users know that when they use both assessments the correlation is above 0.6, see my list above.
>For those unfamiliar with correlations, 0.6 is 36% and 0.25 is 6.25%.<
Do you know why? I’m educated in both engineering and business and I find your rudimentary presentation on correlations deceptive in that it does not go far enough. Copying and quoting does not substitute for understanding.
>This statement from a professor of psychometrics is clearly at odds with the wildly extravagant and exaggerated claims of many in the HR industry.<
Actually it isn’t at odds but then I again understand correlations and how to use them in business and engineering.
>Many disbelieve me when I quote Kline and I am frequently accused of making this quote up so for the doubters, the quote comes from an article in…<
Quoting an expert does not make you an expert in the subject matter quoted.
>We now each have a decision to make. Which evidence about the reliability of psychometric testing is more credible? The evidence offered by those selling psychometric tests or the evidence from the pen of a professor of psychometrics.<
What is the correlation if we interview, perform background checks, administer a personality test, assess for abilities, assess for occupational interests, and then perform a job match? The answer is found in the list above.
>From my own impartial position the decision is easy.<
You are not impartial just because you claim impartiality. And worse than that even an impartial person can be wrong. Impartiality does indicate correctness.
>For those deeply involved in HR, the decision is one that will cause considerable anxiety and cognitive dissonance for many years to come.<
Where is the evidence?
>… Psychometrics may appear a magnificent edifice to some but it has very weak foundations.<
Psychometrics is a tool that when used appropriately adds value.
>Good luck to all of you with this thorny problem.<
It is a thorny problem only if a wrong assessment is used or the right assessment is misused.
Bob Gately