Stress Test

A man applies for a job, which requires him to undergo three days of testing. But just before the testing period begins, he gets another good job offer. How does this affect his effort and performance?

If you’ve got your economics-blinders on, you’ll probably answer: Once he’s got an offer in hand, he won’t try as hard and therefore won’t do as well. But Michigan grad student Susan Godlonton decided to put that theory to the test. She was here in Rochester this week to tell us what she learned.

Godlonton did her research in Malawi, where her research funds went a long way. 278 job applicants were recruited for a three-day training program, with job offers contingent on their performance. At the beginning of the training period, about 20% of the applicants (randomly chosen) were offered an alternative job; another 20% were randomly told they’d not be getting the alternative offer. (Still others were told they had various probabilities of receiving the alternative offer, but let’s concentrate on the extremes.) These alternative offers were kept secret from the evaluators at the training session.

The result: After controlling for background characteristics such as performance on standardized tests, previous experience, age, and so forth, applicants with no outside offer perform considerably worse in the training sessions, as measured by written exams, and the quantity and quality of their verbal participation. Those with alternative offers are 11.3% more likely to make verrbal contributions, and their verbal contributions are better.

So what’s the deal here? Does the guaranteed job offer inspire applicants to work harder? Nope. Those with guaranteed jobs spend, on average, an extra 53 minutes a day watching television (according to their diaries), and correspondingly less time studying the training manuals. In other words, they’re expending less effort (as you’d predict if you were economist) but still doing better (as you might predict if you were a psychologist knowledgeable about the effects of stress).

This is, I think, fascinating stuff. There’s much more in Godlonton’s paper (this is Chapter One of three in her doctoral dissertation). Do you have an alternative explanation? And what does this mean for the way labor markets ought to work?

Share/Save

41 Responses to “Stress Test”


  1. 1 1 Bennett Haselton

    Is this an argument in favor of social safety nets — suggesting that, contrary to economic theory, people are more productive if they know that they won’t be in serious jeopardy even if they lose their jobs?

  2. 2 2 Sam

    It’s always nice to have options in life.

    There’s a risk associated with verbal contributions, and simply having an alternative mitigates this risk. The economics that could be invoked here I think is that people are more likely to engage in riskier behavior, where the consequences behind the risk are reduced or perceived to be reduced.

    I would further speculate that those on the particular margin exposed in the study tend to worry about how they’re perceived by others or are a tad socially anxious; and maybe for these people the perceived cost of ‘saying the wrong thing’ far outweighs the perceived benefits of ‘saying the right thing’ and on net keep quiet. When the costs (job loss) are lessened or removed, these types pipe up.

    @Bennett – I think your suggestion would only be the case if the job in questions payed something similar to that which people receive when on benefits, else why would anyone work. Something that’s not mentioned, but implied I think, is that the alternative job must be a good substitution for the job the participants are being tested for.

  3. 3 3 Brandon Berg

    Bennett:
    Not without understanding what’s going on. And I’m not really sure there’s any explanation which would make this a good argument for increasing welfare spending. Note that it caused them to put less effort into preparing (and presumably searching for other opportunities), but caused them to perform better in the interview. Interview performance is positional—there’s no net social gain if everyone performs better in interviews—and there’s no evidence that this improves actual job performance. Reduced effort in looking for work, with the only improvement being in interview performance, sounds like a net loss for society.

    And that’s assuming that the mechanism here is related to guaranteed income, rather than to the boost to self-perception resulting from the belief that they’d been deemed worthy of the alternative job.

  4. 4 4 Harold

    I think it interesting that they did significantly better despite working less.

    Interestingly, the economic blinkers were correct in one sense – those with a job offer did spend less time studying, as predicted. The result was different because effort is not directly linked with performance.

    Economics predicts that incentives line up with outcomes – the more motivated people will do better. If the incentive is misaligned, then markets cannot produce optimum outcomes. If this result is applicable over wider workplace performance, then the total negative effects could be huge.

  5. 5 5 Harold

    @3: “and there’s no evidence that this improves actual job performance” There is other research described in the paper that does exactly that. Performance was lower when the incentive was higher, although this was a more laboratory type environment. This (the Godlonton) paper avoids many criticisms of this type of study in that it was conducted on real job applicants who had real and normally encountered, rather than articficial incentives.

    “in my setting, the variation is over risk in securing real, meaningful employment equivalent to that for which subjects
    have chosen to apply through a competitive and arduous process.”

    Yerkes-Dodson showed in 1908 that performance has an inverse U correlation with stress, so we cannot say that this sort of thing is a surprise.

  6. 6 6 Jonathan Kariv

    I mean the idea of the interview being less stressful given the new job offer seems reasonable. Assuming the interviewees talk to each other those who didn’t get an alternative offer might be thinking “well I’m probably not getting this anyway”, and then get more stressed or less motivated.

    I’m also curious exactly what Steve means by diaries (re the extra 53 minutes of tv)? I certainly don’t record my tv-watching anywhere normally. Come to think of it those who are made the alternate offer might just be more productive during study time because of lower stress?

  7. 7 7 Ken B

    Confidence first, stress next. They didn’t know they were randomly chosen after all, so there is a confidence boost.

    And Bennett Haselton asks a good question. And not just for job hunting, but for prudent risk taking in general.

  8. 8 8 David R. Henderson

    Was there a real job for those 20% if they wanted it, or were they being lied to?

  9. 9 9 Steve Landsburg

    David R. Henderson: They were not being lied to.

  10. 10 10 Neil

    The candidates who Godlonton rates as most likely to get an outside offer do best in the stress tests. What a surprise.

  11. 11 11 Ken B

    @Neil 10: I am assuming the offers are random, not based on intervew performance. So it should be indpependent.

    This might tie to David R Henderson’s question though. If the job offers are real they are presumably not random, and vice versa. So I am unsure of the mechanics.

  12. 12 12 Matthew

    I think having a job offer calms you down. When you’re calm, you think better.

  13. 13 13 Ken B

    There’s another possibility here, slightly sinister in its implications, and so appealling to me. The interviewers do not know about the job offers, but can sense a difference in the subtle power and dominance relations. I am suggesting that the interviewers, whose (sic) evaluations we are relying on for the measure of the interviewees performance, might be affected by the shift in subtle dominance clues. It is quite common to think more highly of those who score highly in tests of these cues I believe.

  14. 14 14 Ken B

    Cues not clues in 13. I need to get a cue.

  15. 15 15 Luis

    I have always thought that the best time to look for a job is while you are employed. Besides the benefit of not having to take the first job offer you get, some leverage, etc., you just feel more confident.

    It seems to me interviewees worry too much about being liked during the interview process and do not act naturally. If your worst case scenario is having to go back to your current job you will act more confident than the guy without a job.

  16. 16 16 RichardR

    In the paper Godlonton suggests a handful of explanations for the better performance by applicants with an alternative guaranteed job offer but discounts all the explanations except stress. “Gift exchange is one possibility” but it is refuted by the evidence because they “exert less effort in studying”. The “nutritional-wage hypothesis might be a possibility” but again there is no evidence because Godlonton “does not observe differences in food expenditure”. “Stereotype might be the driving mechanism” but probably is not because “perceptions about their likelihood of being hired by the recruiter do not significantly differ across treatment groups”.

    With regards to stress as a causal explanation she admits that she “cannot identify the mechanism through which stress might act to impair performance”. She also claims that her results are “consistent with a stress response” however a stress response can “both increase and decrease performance” and also that “performance has an inverse u-shaped correlation with arousal (stress)”. Since stress can either increase or decrease performance any observation that she makes would be consistent with a stress response! Moreover she did not measure stress among the participants, “it would have been optimal to have biomarker indicators to measure stress directly but this was not feasible”. Therefore we do not know if the stress levels were different between the group with an alternative job offer and the group without an alternative job offer. Therefore there is no theoretical reason to believe that stress differences caused the difference in outcomes and moreover there is no evidence that stress levels actually differed.

  17. 17 17 Ken

    Bennett,

    No, this is not an argument for “social safety nets”. The so called “safety net” is merely the police state forcibly confiscating one group of people’s wealth and giving it to another group. Additionally, you’re ignoring the lost productivity from the group whose wealth is stolen because of the knowledge that their wealth is being transferred to strangers for the simple purpose of vote buying. And you’re ignoring the most important reason for the “safety net”, a lot of that wealth sticks to the fingers of grasping politicians and bureaucrats. If you don’t believe me, take a look at the incredible economic explosion of DC over the last 10-15 years, which has been going on completely unaffected despite years of recession and weak economic growth.

  18. 18 18 Mike H

    “No, this is not an argument for “social safety nets”. The so called “safety net” is merely the police state forcibly confiscating one group of people’s wealth and giving it to another group”

    “I thought that was the financial services industry?” said Billy Goat Gruff…

  19. 19 19 Harold

    @17. It is an argument for social safety nets. There are other counter arguments against, as you describe. What it means is that the negative consequences of providing a safety net do not make a conclusive case against them.

    On the paper, I think the results are conclusive – those offered a job did better in the interview process. This included both subjective assesment and testing. The author speculates that this is stress, but this is not proved.

  20. 20 20 Carter

    This is not the first study of the phenomenon. In the book “Drive” by Dan Pink he cites several such studies which completely turn on it’s head what we understand about what motivates us….and fear (of being jobless in this instance) does not motivate creativity as this study suggests and actually has the inverse affect. The same is true of using money as an incentive. The more you pay someone to solve a complex problem….the more difficulty they have solving the problem. What drives creativity is autonomy, mastery and purpose…not money.

    And to answer Steve’s question….it absolutely should change our understanding of how labor markets ought to work.

  21. 21 21 Nick J

    They knew they were participating in a study, and they probably knew that they were receiving this outside job offer based in some way on their performance in the test. (Yes, that wasn’t actually the case, but if I were taking this test I would assume that my actions were affecting how they treated me in some way.)

    If they take seriously the idea that individuals shirking when they receive the next best job offer is a big problem (I think it is) and that there may be social mechanisms to punish such behavior, then receiving a very good job offer (too good to be true?) may be a signal to the subject that they are being tested and that conditional on their performance in the test the job offer may be withdrawn + additional punishments imposed. Once they’ve taken the exam and the job starts to seem more secure they relax and start shirking a little bit more.

    What I would like to see is a study where the subject is fooled into thinking that their outside option is completely independent of the study they’re participating in.

  22. 22 22 nobody.really

    In the book “Drive” by Dan Pink he cites several such studies which completely turn on it’s head what we understand about what motivates us….and fear (of being jobless in this instance) does not motivate creativity as this study suggests and actually has the inverse affect.

    But I thought necessity was the mother of invention! Now you’re going to tell me that the mothers of invention was just a rock band?

  23. 23 23 nobody.really

    On a potentially related note, Zappos famously offers a cash bonus to trainees to quit. Few people take the cash. Allegedly this practice depresses quit rates. People postulate various reasons why a “sunk cost” would have this effect on people’s future job performance.

  24. 24 24 Carter

    Freedom is the mother of invention.

  25. 25 25 Phil King

    Here’s an alternative explanation -
    Those who received the job offer know what it takes to receive an offer, or at least believe they do. They have received a prior offer without testing and see no reason to up their effort. An economist would predict that someone would put in the least effort needed to have confidence in success. They also do not have to do anything differently.

    Those who did not receive the offer believe they have to change something. It’s not so much stress that negatively impacts them, but having to impress while putting on an act they are unsure of.

  26. 26 26 Harold

    “What I would like to see is a study where the subject is fooled into thinking that their outside option is completely independent of the study they’re participating in.”
    You are looking at one. The alternative offer was not dependent on their prerformance.

  27. 27 27 Ken Arromdee

    Note that it caused them to put less effort into preparing (and presumably searching for other opportunities), but caused them to perform better in the interview. Interview performance is positional—there’s no net social gain if everyone performs better in interviews—and there’s no evidence that this improves actual job performance.

    There’s a net social gain if every person puts less effort into preparing for interviews, unless preparing for interviews improves job performance too.

  28. 28 28 Nick J

    Harold: I know that was how the study was designed, but did the participants believe that the job offer was independent of their performance?

    Say I said to you, “I’m going to offer you a job. The job offer is certain. Now take this test.” Would you necessarily believe me that it’s certain? If I were in that situation I would assume that I was being tested on my ability to continue putting forward effort even when I had a job offer, and I would place about 50% probability that the job offer would be rescinded as punishment if I shirked and did poorly.

    If it’s true that participants believed this, then even a rational economic actor would do the following: Put forth extra effort during the test to signal that they’re not the sort of person who will shirk their duties just because they have an outside job offer, then, as it becomes increasingly clear that the job offer is certain and not dependent on performance, start to shirk more and more. Which I think is what happened.

  29. 29 29 James Knight

    Here’s an interesting question regarding man’s stress. A stressed man is on his way to the train station. He knows the train departs every 30 mins, but he doesn’t know the exact times. Will he improve his chances of catching the next train by running, or won’t it matter, given his unawareness?

  30. 30 30 iceman

    29 – If by “next” you mean the soonest train scheduled to arrive whether you’re there or not (i.e. your goal is to get to your destination at the earliest time), it seems obvious the earlier you get there the better so run? If you mean how long you’re likely to wait once you get there, seems random so it doesn’t matter so walk (unless you need to burn some extra calories)?

    Ken A and Phil King make interesting points. Both suggest there may be implications here for hiring techniques, but I don’t see sweeping conclusions for restructuring labor markets absent some evidence about differences in subsequent job performance. I am aware of previous work suggesting that some type of safety net can make people more comfortable taking risk (entrepreneurship, mobility etc.), but I see that being related to general uncertainty not situational “stress”. Of course there is also *much* evidence that setting the fallback position too high removes incentive to try to do better.

  31. 31 31 Harold

    “absent some evidence about differences in subsequent job performance.”
    There is plenty of evidence that stress and increased monetary rewards decreases some type of performance.
    “Tasks that involve only effort are likely to benefit from increased incentives… While for tasks that include a cognitive component, there seems to be a level of incentive beyond which further increases can have detrimental effects on performance.”

    http://www.bos.frb.org/economic/wp/wp2005/wp0511.pdf

    This does not suggest sweeping reforms to the labor market, but we should acknowledge that it does not work as econ 101 suggests it should. It also indicates that some reforms could at least in principle improve the labor market, since conventional incentives can have the opposite effect to that intended.

  32. 32 32 iceman

    Of course when talking of reform we still have to ask first whether it isn’t in employers’ own interests to adjust to these new insights, or we have some theory as to why they are unable to do so individually or collectively.

    BTW I’m not sure what exactly “labor 101″ consists of, since I recall it including efficiency wages and a story about an African tribe who when paid more simply showed up 4 days instead of 5.

  33. 33 33 Harold

    “or we have some theory as to why they are unable to do so individually or collectively.”

    This is the fascinating aspect. Why do societies continue with practices that are counter-productive? Very often we lack evidence that this is the case, but orchestra hiring gives one a clear example. Would an orchestra leader consistently hire musicians that were not the best? Why would that be? Yet when musicians audition behind a curtain, the chances of a woman being hired increases several fold.
    http://www.princeton.edu/pr/pwb/01/0212/7b.shtml

    They were a bit more open about their racism just after the second world war, although to be fair I think there was some justification for the following from the Vienna Philarmonic:
    “An applicant qualified himself as the best, and as the screen was raised, there stood a Japanese before the stunned jury. He was, however, not engaged, because his face did not fit with the ‘Pizzicato-Polka’ of the New Year’s Concert”.

    And even defining racism. From 1996:
    “What I have noticed that is interesting, is that the Vienna Philharmonic would also never take a Japanese or such. If they took one, this also would somehow by appearances put in question the noble character of Viennese culture. But this is not racist!”

    It is clearly possible for cultural bias to persist and end up with sub-optimal results. Whatever the reason, employers do not always adjust to these insights of their own accord.

  34. 34 34 iceman

    I thought we were talking about stress not discrimination

  35. 35 35 Harold

    Sorry – drifting off topic there a bit. I was replying to Iceman (32) “Of course when talking of reform we still have to ask first whether it isn’t in employers’ own interests to adjust to these new insights, or we have some theory as to why they are unable to do so individually or collectively”

    I was making the point that although it may be in the employers interests to make adaptions to stress, these adaptions may still not happen.

  36. 36 36 iceman

    Understood, I just don’t want to see us lumping all alleged cognitive dissonance together (the paper at hand focused on one particular issue). I was also trying to suggest that while these results are certainly interesting, absent some theory as to why employers would be either a) unaware of the phenomena or b) unable to adjust their hiring practices accordingly, perhaps the default position should be that this suggests they still don’t in fact believe any major changes are warranted. For example they may expect that most jobs will inherently involve some amount of encounters with stressful situations and therefore want employees who can perform the best under those conditions. This would not suggest one wants to remove stress from the interview process (rather the infamous “stress interview” is probably more common). To the extent employers can figure out which applicants really have other offers they might wish to factor this into their comparisons, but this may be difficult info to obtain / verify.

    One of the first things that turned me on to Economics was the idea tha that good economists don’t observe the world not conforming to some model and instinctively conclude the world is wrong, rather that their model likely could be improved.

  37. 37 37 Harold

    iceman: “they [employers] still don’t in fact believe any major changes are warranted”
    This does not mean that major changes are not warranted, just that employers believe they are not.

    Poor practice can persist for quite long periods, even when it is counter-productive.

    Scientific results are frequently ignored if they do not fit in with the existing paradigm. There are plenty of theories as to “why they are unable to [adjust] individually or collectively”.

    As I said, this does not necessarily suggest sweeping reforms, but it should be considered during hiring decisions.

  38. 38 38 iceman

    Also in the real world ‘already having an offer’ may indicate that someone performed well recently in a situation where they did *not* necessarily have an offer in hand.

    I agree these things should be considered — and the first question should be, do we have reason to believe they have not already been? All I’m suggesting is which way our *presumption* should lean. After all people have been hiring people for a very long time. I might think I’ve come up with a brilliant new way to run a restaurant, but if I observe that no restaurants have ever operated that way (and I have no hands-on expertise), that would probably give me pause to reconsider. Certainly we can get stuck in outdated paradigms, although the profit motive provides an incentive to break out of that which may not always exist in the scientific community? And I don’t immediately see an analogue to discrimination here (perhaps a bigot deliberately makes interviews more stressful than necessary to provide cover for an expected disparate impact on certain groups?).

  39. 39 39 Harold

    ” All I’m suggesting is which way our *presumption* should lean” We do research so we don’t have to presume. Not perfect, but improving all the time.

  40. 40 40 iceman

    1- It felt like we were jumping to the conclusion that there were significant conclusions to be drawn here, when it’s possible there aren’t
    2 – Research is often guided / influenced by our presumptions.
    3 – Sometimes we even have to decide where / how to allocate our research resources

  41. 41 41 TeeJaw

    Hemingway would have done better if he’d had no alternative job offer. The rest of perform better with less stress.

Comments are currently closed.