In this episode, David and Drew dig into the potential link between psychology and workplace safety, specifically how personality tests could predict safety performance. They’ll review the research on the connection between personality traits and safety performance, examining how traits like extraversion, agreeableness, and conscientiousness can influence work behaviors. The episode also explores the potential impact of institutional logic on the link between personality and safety performance, weighing the benefits and drawbacks of using personality tests as predictive tools.
The paper reviewed in this episode is from the Journal of Applied Psychology entitled, “A meta-analysis of personality and workplace safety: Addressing unanswered questions” by Beus, J. M., Dhanani, L. Y., & McCord, M. A. (2015).
Discussion Points:
Quotes:
I have to admit, before I read this, I thought that the entire idea of personality testing for safety was total bunk. Coming out of it, I'm still not convinced, but it's much more mixed or nuanced than I was expecting. - Drew
If there was a systemic trend where some people were genuinely more accident prone, we would expect to see much sharper differences between the number of times one person had an accident and all people who didn't have accidents. - Drew
I think anything that lumps people into four or five categories downplays the uniqueness of each individual. - David
There are good professionals in HR, there's good science in HR, but there is a huge amount of pseudo-science around recruiting practices and every country has its own pseudoscience. - Drew
Resources:
The Safety of Work on LinkedIn
David: You are listening to The Safety of Work podcast, Episode 110. Today we're asking the question, can personality tests predict safety performance? Let's get started.
Hey everybody. My name's David Provan and I'm here with Drew Rae and we're from the Safety Science Innovation Lab at Griffith University in Australia. Welcome to the Safety of Work Podcast. In each episode, we ask an important question in relation to safety of work or the work of safety and we examine the evidence surrounding it.
I'm about to throw to you Drew for an intro to this episode and give us some background, but I guess it's up to me to apologize for a bit of a gap in our recording. It's probably been seven or eight weeks since we've released an episode and that's been entirely my doing from I guess a bit of work pressure and then some holidays. I was just mentioning to Drew that I took my first extended break in many years and I hope a lot of our listeners are finding time to do the same for themselves. Hopefully, Drew, we're back on track. At least hopefully I'm back on track and we can make this thing as regular as we hope it can be.
Drew: We are modeling good work life balance for our listeners, David. We're not taking a break. We're just illustrating the importance of recreational along with educational.
David: Yeah, we've done that recent episode on the four day work week, Drew, so I thought I'd take it to the extreme. Drew, this episode, this idea of personality test, my knowledge of personality test is my psychology degree in the 90s and then the ones that a number of organizations have put me through for different roles. Do you want to just talk a little bit about the history of the idea and maybe a little bit of its relatedness to safety?
Drew: Yeah, sure. Before we get too far into that, David, I wanted to ask you about what was your sort of assumption going into this? Because I have to admit, before I read this, I thought that the entire idea of personality testing for safety was total bunk. Coming out of it, I'm still not convinced, but it's much more mixed or nuanced than I was expecting.
David: I think I did for quite a bit. As you know, I'm into social theory during my PhD and trying to understand, particularly in relation to the behavior of safety professionals in organization, the relative influence of structural and agency type factors. Even in our literature review, we talked about institutional and kind of individual influences.
I guess I'm always of the bias that the organization, the institution, the structural factors matter more in organizational settings than the individual factors but that doesn't mean the individual factors aren't playing a role or aren't important in some way. The idea that we can solve all of our organizational problems by making sure that we give people personality tests, I tend to not agree with that.
Drew: The structure and agency is a good way of putting it because that's pretty much been the history of safety science and organizational safety. It’s this sort of oscillation between structural or system explanations for things and individual explanations for things. When we go for individual explanations, we have the idea of behaviorism, behavioral control, selecting individuals based on their personalities. Then at the structural end we focus on systems and environments and perhaps even sometimes broader measures like climate or culture.
It's not really that one of them is right and the other is wrong. It's obviously a mix. It's really sort of where we think the opportunities are and where we think the balance is as to which explanations are most helpful in managing safety.
The earliest swing towards this was the idea of accident proneness. This was sort of the idea that some people are intrinsically unsafe almost because they're clumsy. It was actually originally even talked about almost as a moral failing that there were these inferior people who were morally degenerate and accident prone and even some really suspect links to like race and class background as well.
That's one of the reasons why we moved away from those individual explanations. It was not that they weren't scientific so much as that they were just socially repulsive, some of these theories. We moved more into saying, okay, we put individuals into environments and we can control the environment so why are we blaming the individuals for how they react?
Then we had this sort of approach to human factors, which said sure, different pilots have different levels of skill, but a good cockpit makes every pilot better and a bad cockpit makes every pilot worse. So why don't we work on not putting the bomb drop lever right next to the landing lever rather than try to work out which human is never going to make that mistake?
David: Drew, you mentioned the social repulsion around these ideas and we're talking between, I guess, accident prone and human factors, the 1920s to the 1960s, this evolution of ideas. We know what was happening in the social scene in the developed world around that time. What does the actual evidence say? Let's talk a little bit about the background and maybe as a way of leading into the paper for today. How has this idea been solved by the literature and how's the idea evolved in the literature?
Drew: Very originally the idea of accident proneness, once people started to directly statistically investigate that, they realized that there was really no foundation to the idea that there are individuals who have accidents more than other people. Of course just from random variation, you'd expect some people just to accidentally end up with more accidents than other people. But if there was a systemic trend where some people were genuinely more accident prone, we would expect to see much sharper differences between the number of times one person had an accident and all people who didn't have accidents.
We sort of moved on to this more subtle idea of behaviorism where we say, okay, it's the behaviors that matter and we focus less on individuals and more on just broad patterns of behavioral control, which actually is kind of more system-based rather than individual, even though it's intervening at the level of individual behavior. It's basically treating all individuals as subject to the same forces and we are going to manipulate those forces.
With all things like not just behaviorism, it happens even with some of the new views of safety. When accidents still happen, people start to get frustrated and when they get frustrated, they start to look for, okay, what are the special factors that cause these accidents to still happen? We swing back to the idea of why is this behavioral intervention working better for some people than for other people? Maybe it is something about the individual. Maybe some people learn better or are more compliant, or are more responsive to our controls.
Unfortunately, that then starts to intersect with sort of safety and HR and hiring practices. I don't want to sort of bag on HR. There are good professionals in HR, there's good science and HR, but there is a huge amount of pseudoscience around recruiting practices.
Every country has its own weird pseudoscience. I think my sort of favorite is the way in France they use handwriting analysis as one of their recruiting tools. You look at that and you think, god, that's nonsense. Then you look at the US and these companies that use Myers-Briggs and you think, okay, maybe handwriting analysis is not so dumb after all.
David: Drew, you want to go on and list out all of the personality instruments that you don't like?
Drew: If you let me list out all the recruiting practices I don't like, my favorite is just that we've known for years that interviews are terrible at judging job performance. Anytime I feel like bagging on people using personality tests or people using handwriting analysis or people using blood type, yeah, there are companies that do blood type as part of it. I just think okay, everyone does interviews and interviews are bad. We are just generally bad at picking who will be good at a job. We would like to make it more scientific, which leads us prone to using things like personality tests.
David: The original question is can personality determine safety performance and I guess that's what we're really trying to ask. There's some underlying questions here. What we want to know a little bit is, is it really the case that employees based on their personality are more dangerous?
Two people in the same situation based only on their individual personality will behave quite differently, safer, and more dangerous. Then the second is can we actually identify these people in advance through personality testing? Those are really the two sorts of sub-questions or hypotheses if you like that we're actually trying to test in this episode.
Drew, should we introduce the paper?
Drew: Sure. I will point out that actually my way of choosing this paper was not at random. I did actually reach out to someone who has a fair bit of expertise in using psychometric testing and has a good handle on the psychometric literature, a previous colleague called Dr. Tristan Casey. This is basically not a paper chosen at random, but a paper representative of sort of what someone who is generally inclined towards that sort of thing and knows literature well. This is a good example of the state of the art.
The paper is in a very reputable journal, one we have multiple papers selected from before, The Journal of Applied Psychology. It's 2015, although things haven't particularly drastically changed since then. The title is A Meta-Analysis of Personality and Workplace Safety: Addressing Unanswered Questions. Their unanswered questions are pretty much the ones that we want to know the answers to. These are people looking at a whole bunch of other papers, putting them together to try to answer those questions that we care about.
David: Drew, I'll let you introduce the authors and then maybe I can talk a little bit about what the paper has set out to achieve.
Drew: Sure. The three authors. I don't know these people personally. I'm not familiar with their work. Lead author is Jeremy Beus, at the time of the paper, from Louisiana State University. They're now at Washington State University. Professor Beus primarily does safety climate, does lots of work with sort of detailed statistical modeling using surveys to predict workplace behavior.
Most of it's not actually specifically predicting safety behavior but it is sort of a strong focus on treating individual performance as what he cares about. If we screened out anyone who didn't care about individual performance, we would have no one who was competent to answer this sort of question. This is someone who is focused on measuring and trying to predict individual performance.
The other two authors have similar qualifications. The authors are Lindsay Dhanani and Mallory McCord. They both seem to sort of be most interested in workplace mistreatment, looking at how employees are treated. They've got the right sort of knowledge, background and statistical expertise to do this sort of meta-analysis.
David: Drew, the method of the paper is really set out in a way that does two things. First of all, the authors try to construct, I guess, a working and a plausible explanation for how personality might relate to safety. They've got to actually combine the existing personality literature and ideas and personality with its relationship to safety behavior and safety performance. They try to create a working model and set of hypotheses for that, then through trying to draw on potentially previous statistical connections between personality and accident rates.
It's not this idea that tries to actually look at all of the mechanisms that exist between personality and accident rates, as we might see in some other theory building papers. But they're just trying to actually bring these worlds together and then use that to go and conduct a meta-analysis where they can actually look for all of the existing studies in the published literature and see if there is a patent that holds up in the research between personality and accident performance.
Drew: David, should we get right into the sort of analysis and results of the paper?
David: Yeah. I am going to let you take the lead, Drew.
Drew: Okay. The first thing they say may be fairly self-evident, I think was kind of important, which was that they said that drawing an end-to-end link between personality and outcomes is both drawing sort of too long of a causal stretch and it ignores the fact that there's going to be some mechanism in between. It's not that you are an introvert, therefore things fall on your head. There's going to be some mechanism and they say, okay, the mechanism really has to be behavior. It's going to be your personality that influences your behavior. Your behavior influences whether you have accidents.
They said that was pretty obvious. We're going to get to that because they actually did test that. It turns out statistically not to be as true as they thought, which leaves a little bit of a gap, but that's their working explanation. By behavior, they basically mean a bunch of things lumped together. Behaviors that control, mitigate, any sort of physical threat or harm and behaviors that directly hurt by putting you into an unsafe situation.
They said, okay, let's reverse the safe ones, add them together, and we have this lump of behaviors that we can sort of define as less safe or more safe. A safe behavior would be wearing personal protective equipment. An unsafe behavior would be walking underneath a suspended load.
I think this is pretty non-controversial. It's pretty much where you have to start if you want to do this sort of research. Part of the difficulty, and this is kind of hidden behind this study, is that you do have to have reliable ways to measure the safe and unsafe behavior. In this study, they don't really go into how the different studies they're drawing on do that measurement.
Because you are lumping everything together, you have to assume that it's not very context dependent. One unsafe behavior has the same cause as another unsafe behavior. As an example, if you've got one behavior that's about drivers and you say speeding is unsafe and you've got another behavior about construction workers and you say wearing safety glasses is safe, you've going to sort of assume that those are roughly in the same thing with the same sort of personality causes in order to be able to add them together and to draw general conclusions.
I think there's a fair bit you can challenge there on sort of philosophical and methodological grounds. But also if you want to do this sort of combination of studies and you want to do this sort of analysis, you are just going to accept that those are the constraints we're working under. I think they're legitimate for this sort of work, but they do give you reasons if you ought to reject the conclusions, why you might reject the conclusions, even if they're statistically sound.
David: Yeah, Drew. I think as I was reading through the paper as well, I was thinking of this black and white categorization of behaviors as being safe and unsafe. I think some of our listeners might find challenges with that as you've mentioned. I think pragmatically for the research and selecting these 70 or so studies and trying to work out what they were measuring is necessary. At the same time it's also important that we understand that at certain times, unsafe behaviors might be calculated to offset a different type of safety risk and safe behaviors are purely driven by organizational factors.
I guess from our point of view, we understand there's a whole bunch of things going on here, but this is just the challenge of research where you're going to actually put a box around something and move forward.
Drew: The next bit is what I think is particularly interesting and was quite helpful for me in understanding where this work is coming from, which is they don't just say there's a direct link between personality and behavior, either they put in another step as well. This step is drawing on a theory called purposeful work behavior. The theory basically says that everyone at work has roughly the same set of goals, but our personalities affect the weighting that we give to those goals. Different people are inclined towards different goals. That seems to make sense to me and it makes sense that this is a real difference between people.
The goals they talk about are communion, getting along with and having meaningful contact with others, status, which is having power and influence and success, autonomy, which is about having sort of control over your own environment and having personal growth, and achievement, which is being competent, feeling competent, feeling that you've achieved stuff at work.
The idea is everyone feels each of those to some extent, but different personalities are going to predispose you to focus on some goals at the expense of others. As an example, if you have the personality factor that you are agreeable, agreeable individuals place higher value on communion with other people. That's sort of almost by definition of how they measure agreeableness. Do you like to hang out with other people? Do you get pleasure from your relationships with other people?
They say if you measure high on the personality scale of agreeable, you are more likely to do things at work that are instrumental in having good communion with other people in being socially accepted. It makes a fair bit of sense.
David: Drew, I think it's widely referred to as sort of a five-factor model of personality and we actually see how people move on these five factors if we can see how we think some of these goals relate to maybe safety behavior, then we can actually start to create these missing links in between personality and safety performance or at least propose the missing links and test them.
Drew: Just quickly before we move on, David, you are the psychologist. Do you want to sort of explain the five-factor model and its status within psychology research?
David: Look Drew, I haven't considered myself a sort of a psychologist in 20 years. There were actually far more jobs in safety when I graduated from my psychology degree. I guess there's lots of different models of personality and I think all models are wrong. As George Box would say, all models are wrong and some are useful.
I think this one is a little bit useful in a sense that we see things here that I think when a theory is useful, we see things that make sense in the world based on the patterns that we see around us because that's really what research is about. It's actually trying to find repeatable, observable patterns in the world.
This idea of extroversion, agreeableness, conscientiousness, neuroticism, which maybe has a bad rep but is about more about emotional stability, and then openness, we can actually start to see sort of these five scales here that we can maybe use as a lens over which to see people. I don't really like personality as a construct so much because I think anything that lumps people into four or five categories downplays the uniqueness of each individual, but I think in terms of the psychology literature, it's a fairly widely accepted theory.
Drew: I hear what you're saying, David, about not liking personality as a construct. To the extent that personality is a construct, there is pretty much a consensus within the psychology research community that the five-factor model is, if not perfect, the best we've currently got. You'll get people who want to tweak it and who want to break the individual factors down into sub-factors or to argue that some of those sub-factors are actually a sixth factor, but these are like arguing about modifying the five factor model.
Pretty much whenever I see research that uses something like Myers-Briggs and ignores the five-factor model, my hackles immediately go up because this is something that is picking a way of measuring personality, which is rejected by all the experts in measuring personality. Whenever you see someone using the five-factor model, it's also sometimes the acronym is OCEAN or CANOE, then you at least know they're sort of starting from the current psychological consensus.
David: Do we want to go on to talk about their hypotheses about how these relate to each other?
Drew: Yeah, let's sort of talk about how they sort of go from the personality to make predictions about what your goals will be at work. Shall I do it, David?
David: Yeah, get us started. I might have a few comments along the way.
Drew: Okay, so basically, they're telling almost like folk theories about how you would get from personality to a particular work behavior. There's a bit of an assumption built in. But remember, these are just hypotheses. They're not claiming this. They're setting up what's our sort of default for testing. I think that's okay.
For example, say if you are extroverted, extroversion is associated with status and prestige. That comes almost directly from the measurement of how extroverted this measurement is that extroverts do care about how others think about them. That's one of the questions they ask in the personality surveys. But then they sort of make the next step and they say, okay, if you care about status and prestige, that comes from competition with others. Because there's a trade off between safety and production, then if you are competitive, you are likely to cut corners. Therefore, we predict that if you are extroverted, ultimately you'll be behaving unsafely.
Notice that there's quite a few sort of jumps or steps in the middle there, but they're at least sort of trying to come up with this testable set of explanations that they can measure. Similarly, they say agreeableness is associated with getting along with others and cooperating. Breaking safety rules is socially dysfunctional. Therefore, they predict agreeableness is associated with safer behavior.
Now my concern is that you could tell us a folk story that goes the exact opposite way. You could say extroverted people like to get ahead. Following safety rules makes you look good to people high in the organization. So extroversion causes people to follow safety rules in order to look good and get ahead.
I could say agreeableness is associated with getting along with others. That means you don't want to rock the boat, you just want to do what everyone else is doing. If there's a norm about not following safety rules, you won't want to complain. You'll just go along with everyone else. Agreeableness causes people to break rules. I reckon you can sort of tell a story in any direction.
David: Yeah, and Drew, I think when I read this and listeners might go way back to, I think it's episode 33 or so on institutional logics, I was straight away like, there's actually a missing step here which is the 60 or 70 studies and all these people are the logic of the organization because that first one around safety and production is again depending on whether the organization sees success as being safety or production.
It might change a person with more of an extroverted personality how they would behave and the same with agreeableness. If you're in a culture where breaking the rules is a norm, then agreeableness might be more related to breaking rules than not. I think in this, the context and the logics of the organization or the structural factors may actually determine what these links are. But like you, I think if you create a story, then at the end of the day, the statistical testing is going to prove or disprove the hypothesis in some way. Then you can go about setting alternative hypotheses and test them.
Drew: The slight catch here is that if it really is the case that different institutional logics could reverse the effect of these factors, that's something that they don't and can't really test. They sort of test how much does the safety climate affect it and then does personality affect it on top? But they can't really test what if the different safety climate reverses the effect of something? That one, they can't really test.
In any case, here are the predictions they make and listeners, you can sort of listen along and see if you agree with what they assume that the predictions are. They predict that extraversion, neuroticism, and openness to experience are ultimately bad for safety. Using different stories, they say that you can think of extroversion as being very interested in the opinions of others, wanting to get ahead, and wanting high status.
You can look at neuroticism as being about sort of self-centeredness, that's not quite right, but that's sort of the folk version of it. Openness to experience has a number of sub-factors, but they're often related to risk taking and experiencing new things, even if those new things are dangerous. That's the story about why those things might be bad for safety.
Then they say that agreeableness, wanting to get along with others, is good for safety and conscientiousness, which is what it sounds like is good for safety. They also interestingly predict that all of these five factors are probably too broad labels to be very good at predictors and they suspect that better predictors will come from narrower sub-factors.
One of the key examples is they say openness to experience can cover a lot of things. One of the specific things it covers is risk taking and you would expect risk taking to be much more predictive than just general experiences.
David, have we sort of said enough about the theory? Should we get into the statistics and what they did?
David: Yeah. What they did and what they found. Happy for you to go again, Drew.
Drew: Okay. What they're doing is they're looking for studies that link one factor, at least from the five factor model, because some things just test introversion, extroversion, some things test the whole five. Then they compare that to outcomes, which include safety related behavior, safety climate, or actual safety outcomes.
Three different things. They're not going to lump those three things together. They're going to try to create a model that shows the different effects of these different things. In all, they found 69 different studies that do tests of personality leading to safety related behavior, safety climate, or safety outcomes.
This is my one big criticism of their work where they screened those 69 studies to rank them in terms of quality. In particular, at least as far as they report in the paper they didn't look at exactly how each study measured safety behavior. I went and had a look at some of these studies because I kind of suspected that some of them were about self-reported behavior and that was the case.
The other thing I noted was that almost none of the studies are about work. A large number of them, in fact most of them are about driver behavior and not specifically driver behavior at work, but young adult drivers and learner drivers or new drivers. One of the studies is about military fighter pilots and all through them, lots and lots of asking people how they behave rather than actually checking how they behave.
That's a problem because personality could affect how you talk about how you behave. For example, what if extroverted people are less honest or more honest about their safety behavior?
David: Practically, Drew, these are researchers at universities recruiting other students at universities for their study and giving them a personality test and then a driving behavior self-report survey and using those two as a battery to run some statistical correlations.
Drew: They weren't all that bad.
David: Okay.
Drew: But those ones are in the mix.
David: Okay. Got it. Alright. How did they combine these studies for the results?
Drew: I'm not going to go through the mathematics. As best I can tell, given all of the constraints, their math is fine. What they're looking at is fairly common in a meta-analysis. You look both at the sort of volume of people in each individual study, as well as the variability, how much each study sort of spreads out and is confident about its own conclusions. You use that to basically add the studies together, weighting them for their individual variation and quality.
Essentially, you end up with one big study that includes all the individual studies taking into account their statistical differences. That leads you then in that sort of virtual combined study, you can ask all of your research questions. Statisticians are going to hate the way I explained that, but hopefully that gives a rough sense for what they're doing.
David: I'm sure we've got less rather than more statisticians listening to the podcast, Drew.
Drew: In that case, that was perfect. That's exactly how it happened, David.
David: I'm sure everyone's interested in the results now across these sorts of hypotheses. Drew, take us through the conclusions.
Drew: All caveats aside, here are their specific findings. Extroversion weekly predicts more bad safety behavior. If you're an extrovert, you are dangerous, but not particularly dangerous. Just a little bit. If you're an introvert, feel good about yourself.
Neuroticism predicts bad safety behavior. Agreeableness and conscientiousness, both good for safety, but not by a lot, just by a little bit. Openness to experience doesn't seem to have much effect in other directions.
If we took those results as gospel, we would give you a personality test and basically give you a strike if you are extroverted, a strike if you are neurotic, and a mild tick if you are agreeable and conscientious.
Interestingly, that pretty much lines up with the research about non-safety work performance as well. Generally, if you do any sort of personality test, before hiring people, unless it is total bunk, and most of them are total bunk, if it's any good at all that's basically what it's measuring for. It's giving you a slight tick for introversion unless you are in a sales position, in which case it's giving you a slight tick for extroversion. It's giving you a tick for agreeableness and a tick for conscientiousness, which are believed to be things that make good employees.
That pretty much is what these findings line up with. But all of these are not big correlations and they are way smaller than the effect of the safety climate, which is the broader structural effect that affects everyone's behavior.
When we say smaller, it is on top of. It explains the variation using a safety climate. That explains a lot of variation between organizations. There is still some unexplained variation and personality can explain a bit extra on top, but that bit extra on top is smaller than what we've explained so far. It's not swamped by the safety climate. It's as well as. It's just much smaller.
David: Yeah, Drew, I think that makes some sense to me. I like the way that you've talked about the combination of the effect of climate and personality and the relative size of the effect of the personality results notwithstanding some of the challenges in this meta-analysis in terms of the study quality that makes it up.
Do you want to sort of share anything more about what we can conclude from this study before we talk about some practical takeaways?
Drew: Okay. There are a few things here that are both interesting and theoretically confusing because remember, they put a lot of work into setting up these how do we think this is working? How do we think personality is influencing behavior? How do we think behavior is influencing safety?
The first thing is statistically at least there's something going on here. There is some effect. Now I personally think that there's a high chance that this effect is all in the self-reporting. That personality has a big effect on how you talk about safety and how you answer survey questions. But even I'm acknowledging in saying that I think personality has an effect.
It is really quite plausible that there are other work situations other than answering surveys where personality does matter. This study does seem to offer some support to the fact that personality makes a difference. That is, remember not a difference to getting hurt but a difference to how you behave at work and how you behave at work in safety relevant ways. But the patterns here are a little bit weird.
Personality seems to have a bigger effect on the outcomes than it does on the behavior, which is kind of weird if personality influences outcomes by influencing behavior. This is literally introverted people behave a little bit differently from extroverted people, but are much less likely to have a thing fall on their head. We've got this unexplained, how are people protected by their personality except through their own behavior? And that's hard to understand.
What it really suggests is less that there is a clear explanation and more that there is a weirdness in how we are measuring things, causing this weird result. Weirdness in how we are measuring personality or measuring outcomes.
David: Potentially Drew, just in thinking and I'll probably ask this more of a question, is because of the way that they selected studies as having one of the five or more personality traits and connection to behavior, climate, or outcomes, is that there were some studies in here that were trying to link personality to outcomes.
Again, it might tie back into your thoughts on influencing reporting, maybe people are even more likely or less likely to report outcomes than they are to report behaviors even in the driving context. Yeah, I drive unsafely, but I don't have accidents. I just wonder where all that reporting came from and the validity of all of that reporting particularly about outcomes?
Drew: The authors do have their own explanation for this, which is that they say that when we measure safety climate, we probably accidentally suck up some measurement of individual personality when we try to measure a safe climate as well. Once you've taken the safety climate into account, you've accidentally sort of moved different personality factors forward or backwards so you then get your results for personality distorted. That is quite plausible, but it also suggests that personality is affecting how you answer your survey questions on safety climate. That feeds into reporting as well.
The next interesting thing is how does something like agreeableness work? Conscientiousness is fairly straightforward. Your conscientiousness influences things like how important you hold rules for. It makes a lot of sense that if you're conscientious, you tend to follow rules more. We've defined safety related behavior as following safety rules. Okay, we are not surprised by this.
Something like agreeableness is supposed to work because you care about the group, you care about fitting in with the people around you or you care about being socially accepted. But how come the studies that show agreeableness works that appears most often in the studies about solo drivers and it's a weaker effect when people are working in groups? That's weird.
Again, it suggests that actually, agreeableness is causing people to answer the questions differently rather than agreeableness causing them to change their behavior. Because for that to work, it would have to work more when you are part of a group than when you're alone. In fact, it works the opposite. It works more when you’re alone than when you're part of the group, which just makes no sense.
This doesn't sort of take away from the original idea that, okay, personality is having some effect here, but it's not having an effect in a straightforward way that we can easily explain. That should lead us to be confident that we know which personalities are going to be safer.
The final thing to note is that most of this work is about very specific situations so we shouldn't be extrapolating. Most of these studies are young drivers and sure, personality probably does influence how a young driver behaves. A young driver who is very high on risk taking, I would be fairly confident is more likely to speed than a young driver who is low on risk taking and is very very conscientious.
That shouldn't be surprising, but we shouldn't then assume that we get into a much more complex situation about how people respond to production pressure at work when their boss tells them to cut a corner. Should personality pick their behavior there? Maybe, maybe not, but I wouldn't use the young driver study to tell me what direction their personality would affect their behavior.
David: I guess, with the caveats that we put on this, as well as sort of pointing out the practical strengths of this research, clearly there is some effect here and maybe we don't understand it fully, but from a practitioner point of view or even for researchers, what would be the practical takeaways from this paper? I might add to some of yours as we go through.
Drew: The number one thing I would say is that I was surprised that this is not total bunk. This is not total bunk, but we definitely don't understand the impact of personality on safety nearly enough to use it as a tool to predict who will or won't make a safe employee.
I'm not a fan of safety and we have a way better understanding of safety climate and way better evidence for safety climate. If you're going to start anything, start by measuring the safety climate. If you're reluctant to measure the safety climate, definitely don't start investing in personality tests for your employees instead.
The second thing is this isn't total nonsense and I would not sneer at researchers wanting to get involved in this and wanting to try to make some advances and useful contributions here. There's enough going on, enough interesting things that more and better research could have value. In 10 years time, the answer to that first question may be different. In five or 10 years, maybe we do understand the impact of personality enough that we can start using it as a tool.
I doubt it's going to be an employee screening, but there are lots of other things we could do like measuring within the organization, looking at how we are going to train people, looking at whether our psychological safety programs are effective. There are lots of different ways that we could use personality to get some insights and to make some contributions.
David: Drew, I think as I sort of read through this paper and thought through this episode, I was the same. I had a whole bunch of more hypotheses that I thought were definitely useful in a sense that would give us some insights. We talk a little bit about safety being a transdisciplinary science with a whole bunch of parent fields, and psychology is definitely one of those foundation parent fields.
I was even thinking about the relationship between personality and employee voice and psychological safety is this group construct. What is the individual construct that even gives an employee more agency in their role? Because we know in complex systems, the law of requisite variety says that we actually want diverse opinions and maybe there are some things that we can look for within groups, not in every person, but at least within groups that could really help. There's a whole bunch here that I've left in.
Drew, I know we sort of segued a little bit from practical takeaways, but I always don't mind it when I leave a good paper with more questions than answers because I think sometimes in good research, you actually do come away with more questions than answers.
Drew: Yeah, none of the caveats I've put as to the conclusions of this work, I think all reflect badly on the researchers. I think they're pretty open about what they're doing. They're pretty open about where they're certain or uncertain, pretty clear about where they're speculating and where they think there's good evidence. In attempting to answer a couple of questions, they've raised a lot more questions, and that's what good research does.
David: Drew, the last takeaway there or I guess the last caution for people in this space?
Drew: So this is sort of a consumer warning. When we talk about the fact that we need more and better research here, I absolutely do not mean we need some proprietary personality tool. This advance is not going to be made by one company [inaudible 00:39:33] this is how to measure personality tools.
To make progress in something like personality, you need volume, you need replication, you need reliability and standardization across multiple research teams. We need people using our standardized consensus ways of measuring personality, and we need to use that across a wide volume of context and behaviors to start seeing patterns. Inventing brand new personality tools is not the type of advancement that's going to get us anywhere.
We can disagree, David, about whether personality is a construct or deserves to be a construct, but we've got broad agreement about if it exists, how do we measure it. What we need is people using those measurements to find out more about the relationship between personality and behavior in different situations in different contexts with different choices under different organizational influences. Just coming up with a new personality tool isn't going to make any sort of advance here.
David: Drew, the question we asked this week was, can personality tests predict safety performance?
Drew: Maybe. It depends. Sometimes, in some places not yet. I don't want to say no, but it's not yes yet either.
David: Very good. Thanks, Drew. That's it for this week. We hope you found this episode thought provoking and ultimately useful in shaping the safety of work in your own organization. Send any comments, questions, or ideas for future episodes to feedback@safetyofwork.com.