What personality types actually mean
April 30, 2010 10:23 AM   Subscribe

What do personality-type classifications actually assert about populations?

Let's say I have a system in which type-A people look at paintings and make decisions impulsively, while type-B people look at trees and make decisions carefully.

I'm skeptical of personality types in general, but I want to understand them better, so, what does this example system actually assert about people in general, rather than just about individuals?

(a) There are positive correlations between the traits of each type (that is, painting-looking is correlated with impulsiveness and tree-looking is correlated with carefulness) (2 correlations asserted)
(b) There are negative correlations between the traits of different types (4 correlations asserted)
(c) Both (a) and (b) (6 correlations asserted)
(d) I'm missing the whole point

Bonus: In the case of real personality-type classifications, are these correlations actually measured, and how? (And how many of them?)
posted by pengale to Science & Nature (18 answers total) 6 users marked this as a favorite
 
The only example I know of for a real-life example of how this can be helpful happened at a former workplace. The HR person told me that all five or six people who were the company's top managers had almost exactly the same Myers-Briggs results. Knowing this apparently helped them understand some weaknesses they had as a management team. Or something like that.
posted by bluedaisy at 10:41 AM on April 30, 2010


I'm skeptical of personality types in general, but I want to understand them better, so, what does this example system actually assert about people in general, rather than just about individuals?

Your question is somewhat ill-defined. People in general meaning the average person which may or may not include a person from the population of Type A or Type B? Personality types only really serve to make predictions about people who fall into those personality types, so unless you have some figure about how many people in the overall population are Type A and Type B, it's not very useful. Even then, you're dwindling your chances of making an accurate prediction because there is so much more variability.
posted by tybeet at 11:00 AM on April 30, 2010


I think this may have the answers you're looking for.
posted by valkyryn at 11:00 AM on April 30, 2010


d) - I think you're missing the point. It just means that type A likes looking at paintings AND makes decisions impulsively. The two don't necessairly correlate. It's usefulness comes into play when deciding what to get for type As birthday. A painting would be preferred over a tree. You can go to an art gallery and say 'pick out something' on that day, and they'll quickly and impulsively make a choice. For type B you may want to go with a plant. Because they like to make decisions carefully, don't put pressure on to make a choice immediately. Perhaps get a reference manual and a note saying - please pick a plant and I will purchase it for you.

This doesn't say that the painting person wouldn't like a plant, or a painting of a plant. It just says that the majority of the time they would choose a painting over a plant.

If you would like to extrapolate this out to something like.. myer briggs profiling.. the idea is that people generally have a preference for one trait over another. For some people this is 50/50 which makes the result effectively useless - you don't know which one they would prefer. For other people, if 90% of the time the prefer one over the other, then you have a useful piece of information.

What do personality-type classifications actually assert about populations?

Regarding populations, knowing that the majority of people tend towards one trait over another provides useful information. The most common Myer-Briggs classification of people are ESTJ (male) or ESFJ (female). These traits are often portrayed as the baseline of normal (please note - I do not agree that this should be the case) which has results ranging from media portrayls of characters to decisions on what to include in the DSM. It is also useful in small populations, such as the corporate culture of NASA. There was an interesting study that discussed a monoculture of types and how they may have led to some poor decision making - I'll post a link to the article when I find it.

I'm happy to discuss this in more detail if you have any questions.
posted by valoius at 11:02 AM on April 30, 2010


Bonus: In the case of real personality-type classifications, are these correlations actually measured, and how? (And how many of them?)

Personality type classifications are measured by self-report indices, which make different classifications depending on the way the personality types are defined. Two of the most common and considered to be most valid are the Big Five and the MMPI. Often, classifications from different scales will overlap because they are measuring essentially the same thing.
posted by tybeet at 11:04 AM on April 30, 2010


Put me down for (d). While your question ostensibly addresses personality types, your four scenarios have more to do with statistics and experimental design than with personality types. So it's hard for me to figure out what you really want to know.

If I just work off your title, though, here's my response: Any classification scheme (of personalities, in this case, but the same applies for anything) necessarily loses some data. The benefit of using a classification scheme, though, is that it reduces the variations to a more manageable number and allows you to infer some characteristics that are generally associated with the class even if they aren't specifically measured in any particular member of the class.

Depending on your purpose and the validity of the classification scheme, these inferences might be treated as strong assumptions (unless and until they are disproven) or they might be hypotheses that you would want to test out before you would actually use them as the basis for any decision.

For example, you can generally infer that an introvert (I) on the Myers-Briggs will be much more comfortable and cogent discussing a topic when they've had some advance notice. But I would only consider it a tenuous hypothesis that that same introvert would prefer to avoid large gatherings of people.
posted by DrGail at 11:41 AM on April 30, 2010


In my study, we use the NEO-FFI personality assessment, based off the "Big Five" factors that tybeet linked.

The questionnaire generates 5 subscores of personality; neuroticism, extroversion, openness, agreeability, and conscientiousness. We do not care much how any one of the subscores correlates with any of the other 4, either within or across subjects. That is something of interest to some personality researchers who are trying to question the validity of such constructs, but we assume that the subscores describe something useful to us for another purpose.

That purpose is generally how subscore X correlates with observed behavior Y- in my lab, that's drinking. We want to know if the extroverts are the big drinkers, if the conscientious people are better at controlling their drinking, etc. (not saying we have found any of those to be true) It's just one piece of info in the mountain of data we collect that might let us predict who is going to do behavior Y more often, based on giving the NEO and a bunch of other tests. That is the point of personality assessment for many researchers who are not primarily studying the idea of personality itself.

I share your skepticism about personality assessment in general, because the big tests are interested in a global picture of a person, which may not allow you to predict any one instance of their behavior, or even many instances of behavior in a particular context. Even if you get a huge correlation between subscore X and behavior Y in a lab setting, if you change the context of behavior Y slightly, that correlation may disappear! If I want to predict behavior Y, there are usually better, more specific ways of doing it. If I want a quick and dirty way of predicting behavior Y in a sample of 5000 people and I don't have time for more extensive testing, maybe a personality test will do the job, but that's really the only kind of situation in which I think they are at all worth using.
posted by slow graffiti at 12:01 PM on April 30, 2010


I view something like the MBTI as a strengths assessment in terms of personality. Apparently, a person can be pretty malleable on these tests up until the early 20's when an individual's results will become fairly consistent.

They are useful for understanding your personality and how you are best communicated to. Better yet, if you really understand the "types," you can figure out how to communicate with others as they need.

A broader example of this would be gender relations. Of course there are differences among individuals, but Deborah Tannen and Carol Gilligan were (completely independent of each other and in completely different ways) able to develop accurate generalizations in the communication styles of men and women. Carol Gilligan for example notes that women seem to emphasize empathy in their relationships, and migh quit an activity to avoid a conflict that might hurt a relationship with a o-worker. On the other hand, men sometimes have heated arguments about "the facts" of a situation, without any damage to a long-term relationship due to a "justice orientation."

his may not answer your exact question, but hopefully it offers a different perspective on the usefulness of personality tests.
posted by jander03 at 12:16 PM on April 30, 2010


I think I failed to make myself clear, and/or I'm just asking something that isn't usually asked about personality types. The answers so far have been insightful, however. Just concerning a different question ;) I was not asking about the utility of personality tests, nor about their application to individuals.

Here's the part I think I left out: there is typically no "other" category. Everyone who takes a personality test (again, the ones I know of) is given a type. Thus, a personality-type system makes a claim regarding what kinds of people there are in the world.

My reductionist example says there are painting-impulsives and tree-carefuls. It implicitly says that everyone is either one or the other. (If it were Meyers-Briggs there would be 4x2 types, but this is a reductionist example).

In other words, the implicit assertion of my system is that it is relatively unlikely for someone to be a tree-impulsive or a painting-careful. Yes?

In case the abstraction of this example is getting in the way: According to the Meyers-Briggs web site, thinkers prefer technical professions and want to make decisions fairly, while feelers prefer people professions and want to make decisions compassionately.

Does this constitute a claim that, in the general population, preferring technical professions is positively correlated with preferring fair decision-making or that preferring people professions is negatively correlated with preferring fair decision-making, or both, or neither?

The "how is it measured" part of my question did not refer to how personality tests are administered; it referred to whether anyone is going out and empirically determining whether preferring technical professions is correlated with preferring fair decision-making, etc., and so on for all/many other traits.
posted by pengale at 1:03 PM on April 30, 2010


I still don't really understand your question(s).

In other words, the implicit assertion of my system is that it is relatively unlikely for someone to be a tree-impulsive or a painting-careful. Yes?

You're dealing with four constructs here: impulsive, careful, tree-looking, and painting-looking. Unless you are to assume that tree-looking and painting-looking are dichotomous like impulsive-careful are often assumed to be and that the two spectrums (impulsiveness and looking) are orthogonal, then there's no reason someone could not be both impulsive and tree-looking or impulsive and painting-looking or careful and tree-looking or careful and painting-looking.

The "how is it measured" part of my question did not refer to how personality tests are administered; it referred to whether anyone is going out and empirically determining whether preferring technical professions is correlated with preferring fair decision-making, etc., and so on for all/many other traits.

I think what you're on to here is a question of construct validity. You're trying to determine if the descriptions of the personality types/categories are representative of how the measurements are being obtained?

To go back to your main question, whether it is 2, 4, or 6 would depend either 1) on assumption (as most early theories are), and as I explained above, or 2) by running a factor analysis, and seeing how many factors fall out of that. Most theories begin as 1) and then are validated by 2). The scales mentioned so far (MMPI and the Myers-Brigg) have been validated and the types measured are all, for the most part, independent of one another. Whether tree/painting-looking is independent of these is a whole different question.

I don't know if that answers it for you, but that's all I've got for now.
posted by tybeet at 1:37 PM on April 30, 2010


You're dealing with four constructs here: impulsive, careful, tree-looking, and painting-looking. Unless you are to assume that tree-looking and painting-looking are dichotomous like impulsive-careful are often assumed to be and that the two spectrums (impulsiveness and looking) are orthogonal, then there's no reason someone could not be both impulsive and tree-looking or impulsive and painting-looking or careful and tree-looking or careful and painting-looking.

Exactly. But my system doesn't have those categories. Just as the Meyers-Briggs doesn't seem to have a category for people who both strongly prefer people-related professions and strongly prefer fair/factual/etc decision-making.

Construct validity
sounds closest to what I'm wondering about, so I'm going to mark that as best answer and go read about statistics so I can grok this stuff better. Quote from socialresearchmethods.net on construct validity:

In 1959, an attempt was made to develop a method for assessing construct validity using what is called a multitrait-multimethod matrix, or MTMM for short. In order to argue that your measures had construct validity under the MTMM approach, you had to demonstrate that there was both convergent and discriminant validity in your measures. You demonstrated convergent validity when you showed that measures that are theoretically supposed to be highly interrelated are, in practice, highly interrelated. And, you showed discriminant validity when you demonstrated that measures that shouldn't be related to each other in fact were not. While the MTMM did provide a methodology for assessing construct validity, it was a difficult one to implement well, especially in applied social research contexts and, in fact, has seldom been formally attempted.

To put this in my own words: yes, a classification system with multiple traits does claim both positive and negative correlations exist. Not all of them have been empirically verified.

So, once more, to make sure I've got this: Assuming that the aspects of the Meyers-Briggs have, indeed, been empirically verified, then a random person who claims to prefer technical professions should be more likely than the general population to claim to prefer fair decision-making. Correct?
posted by pengale at 2:02 PM on April 30, 2010


So, once more, to make sure I've got this: Assuming that the aspects of the Meyers-Briggs have, indeed, been empirically verified, then a random person who claims to prefer technical professions should be more likely than the general population to claim to prefer fair decision-making. Correct?

On the Meyers Briggs scale (which is based on the Big Five personality theory), a random person scoring high on the Openness factor would prefer complex/technical professions. Assuming fair decision-making is technical (which I think you're trying to argue it is, and I would probably agree with), then yes, this person would be more likely (in terms of a statistical probability) than the average person (who would tend to score "average" on Openness) to prefer or appreciate fair decision-making. Is that what you're after? :)
posted by tybeet at 2:24 PM on April 30, 2010


Thanks. That's pretty much it, though I want to look more into how they verify the correlations, positive and negative, and just how strong they are. It sounds like factor analysis is going to be the important part. The fair-decision-making trait is listed on the Mayers-Briggs web site for T's.
posted by pengale at 2:33 PM on April 30, 2010


The fair-decision-making trait is listed on the Mayers-Briggs web site for T's.

I guess I missed that. It's been some time since I learned about those specifics. Good luck!
posted by tybeet at 2:35 PM on April 30, 2010


Assuming that the aspects of the Meyers-Briggs have, indeed, been empirically verified, then a random person who claims to prefer technical professions should be more likely than the general population to claim to prefer fair decision-making. Correct?

You're adding something else into the mix: what people claim vs how they score on tests. There is a difference.
posted by internet fraud detective squad, station number 9 at 2:40 PM on April 30, 2010


You're adding something else into the mix: what people claim vs how they score on tests. There is a difference.

True. But answers to personality-test questions are self-reported; the test administrator doesn't do participant-observation at your house. And some of the questions can be pretty direct. But not all, so "claim" was imprecise.
posted by pengale at 2:44 PM on April 30, 2010


Mostly they're shortcuts, I think. Giving a Myers-Briggs type is a lot easier than rattling off a list of traits.

Also, I am apparently the exact opposite of normal. Go me!
posted by dagnyscott at 6:00 AM on May 1, 2010



There was an interesting study that discussed a monoculture of types and how they may have led to some poor decision making - I'll post a link to the article when I find it.

valoius have you found a link for this yet? Or at least the title?
posted by bleary at 10:06 AM on May 1, 2010


« Older Looking for suggestions on volunteer gigs in...   |   Online Address Management Newer »
This thread is closed to new comments.