Thursday, April 15, 2010

The perils of testing incumbents

Many of us plan or administer promotional tests on a daily basis. It's a normal way for the organization to figure out if people are ready for the next step--as journey person, lead, supervisor, manager, etc.

What isn't so usual is testing incumbents to determine if they can keep their jobs (drug testing aside). Amtrak, who is poised to take control of Metrolink, Southern California's commuter rail service, ran into fierce opposition recently when they announced that they will require train crews to take and pass two personality tests traditionally used only for screening applicants.

Why now? Amtrak is trying to prevent a repeat of the Chatsworth crash, where a Metrolink engineer crashed head on into a freight train, resulting in 25 dead (including the engineer) and 135 injured. Records revealed the engineer had a history of sending and receiving text messages hundred of times while operating trains (violating safety rules), including seconds before he ran a red led and crashed. Amtrak officials hope to identify "psychological issues", particularly those that manifest themselves during times of stress.

The tests in question are the Applicant Personality Inventory (API) and the Hogan Personality Inventory (HPI). The tests have been used since 2004 (API) and 2002 (HPI) to select among applicants for engineer and conductor positions, and the pass rate for the API is around 80%.

Union leaders have aggressively resisted the idea, claiming that the tests aren't valid or "relevant" measures of a trained and experienced employee's ability to safely operate trains. Instead, they claim the tests are a "witch hunt."

There are several things I find interesting about this situation:

1) Union leaders don't believe the tests are valid for incumbents, but have no problem with using them to select among potential hires. Assuming the personality aspects tested for by these measures are relatively stable, why would they be okay with testing applicants but not members? Methinks the issue is less validity and more membership.

2) If the vast majority of applicants, who are presumably a more heterogeneous group than incumbents, pass these exams, why would the union think that any incumbents would "fail" them? If it is indeed a witch hunt, what evidence do they believe management is relying upon to pre-judge certain individuals?

3) On a related note, given the high pass rates why would Metrolink/Amtrak think that any incumbents will "fail" these exams? (One wonders if the pass point would remain the same) This could be a situation where much political capital is expended with very little utility as a result of the assessments.

Incumbent testing is always a risky business. Even in the best of situations, those that fail or are passed over may harbor feelings of resentment, even anger. HR professionals must treat these selection situations with particular care and plan for how to communicate about the exam, before and after results are announced and used. But as this case demonstrates, when the tests in question are used to determine continued employment in one's current position--and the tests are personality inventories--tempers may flare particularly high.

Hat tip to my friend Warren Bobrow for this story.


Reid said...

I happened to meet an HR specialist who worked for Amtrak and had knowledge of situation. He indicated that they were requiring everyone to re-apply for their positions which increases the defensibility of the testing program somewhat. However, there obviously are performance and other data available about the incumbents' level of qualification in addition to their assessment results which I would imagine should also be considered.

Anonymous said...

If a small group of quantitative "eggheads" got together and estimated the probability that testing incumbents in this manner would prevent *any* transit accidents, I bet the probability would be quite small, regardless of whether they are conservative or liberal in their estimates of the various factors (e.g., probability of an accident given testing of only new applicants, probability that operator personality would be a key cause of an accident, probability of the test correctly identifying a true problem employee, probability of such a person eventually getting removed from the job). If I had any authority on this situation, I'd want this type of analysis to inform the decision. Like you said, it involves political capital, an investment in resources (esp. everyone reapplying for jobs!), and has other implications like org/job satisfaction. That is not a knock on personality testing, they just need to keep it in perspective - and learn a bit about reliability and validity.