Scientific evidence for the MBTI assumptions

[Nails]

Hello all,

First of all, a disclaimer: I am new to MBTI and to this forum, I have been browsing through threads and did not immediately find anything on the issue, I am concerned with. So, if you guys have already discussed the issue below elsewhere or I am making the wrong assumptions about MBTI, please excuse me.

Anyway, the MBTI assumption of people being either E or I, N or S etc. seems to be valid for most of my entourage. Blackwater is very much ENTP, Economica and I are very much INTJ (even though I completely understand why people could mistake her for being E), Eco's boyfriend and Vortex are really ENFPs etc.

In general, the binarism of the four dimensions seems to be valid, most people are either one thing or the other, and there is very little doubt about what they are. Psychological mechanisms such as self-amplification (people do and improve at what they are already doing) can explain the binarism. The duality is very nice, we can then efficiently describe personalities and understand their social interaction between the various types.

But then, from time to time, I meet people that I have a really hard time classifying. Vortex has a best friend, we are pretty sure about this friend being F but the other dimensions escape us. I thought she was ENFP, Eco is fairly convinced she is ISFJ, Vortex believes her friend is xNFx but even though Vortex has known her friend since high school, she can't say for sure about the two remaining dimensions.

My concern is now, what does this mean for the MBTI binarism? If the binarism does indeed hold, with everybody being one thing or the other, then typology misclassifications may occur due to lack of knowledge or misunderstandings, but in general a large number of MBTI analysts should agree on the personality type of a specific individual.

So, what I would like to know, is this really what is happening? If a large number of analysts examines one individual, will, let's say 95 % of them or more, agree that this person is indeed e.g. ENFP? Or could it be that 45 % would say Sensing, while the remaining 55 % would pick N?

Do we have any statistics on this issue? I have seen a lot of data on scientists being mainly INxJs, policemen being all S etc, but I have not seen any statistics describing the classification of one specific individual by a large number of MBTI analysts. Please inform me, if you are aware of such statistics on the certitude of the MBTI classification of individuals.

I am sure that everyone (who knows him obviously) will agree that Eco's boyfriend is ENFP, but I could very well imagine that we would all totally disagree on Vortex' old friend. What does this mean? Could it be that most people are one thing or the other, but that some are in between? In that case, would it be more appropriate to describe people using percentages, e.g. 65 % N, 35 % S etc.?

Any thoughts or ideas on the subject are most appreciated.

BR

Edit: We are sure about the type of Vortex, it is her friend that is uncertain.

 

[ptgatsby]

Do we have any statistics on this issue? I have seen a lot of data on scientists being mainly INxJs, policemen being all S etc, but I have not seen any statistics describing the classification of one specific individual by a large number of MBTI analysts. Please inform me, if you are aware of such statistics on the certitude of the MBTI classification of individuals.

Someone mentioned this to me before and I did a bit of reading on it, which I wasn't aware of. However, I have generally concluded that he was correct - the instrument is kurotically distributed. The binomial push really is a forced method after the instrument (whereas the instrument uses binary choices, which makes the argument that much more complicated). Course all this says is that the instrument is much more similar to other inventory responses tests.

In any case, the analysts base it on the instrument results and the certainty measurement in the test. Reliability on the instrument is relatively solid, so analysts are generally as reliable as the reliability of the instrument (with some deviation).

However, while I don't know of any statistics on the ability of people to read other types accurately... I do know that relatively speaking, and I must emphasise this, a stranger not even meeting the person can predict the person's type just as well by looking at his personal living space for 15-20 minutes as a long term friends (many years of knowing eacf other). The least accurate would be the ability to judge "at a distance" (ie: stranger meeting the person, observing or reading interviews/etc).

I don't know if that answers any of your questions, but if not, I might be able to answer a more specific question (rather than interpret the question "what does it mean for MBTI binarism" )

 

[Introverts_Unite!]

The statistical evidence for the Myers-Briggs Type Indicator and associated research over the years -
(but note this is NOT for other copycat indicators like Kiersey etc) -
can be found in the Manual, available through CPP - Consulting Psychologists Press, the publisher of the MBTI.
The Manual is pretty dry (not to mention expensive) but it's important if you want to seriously tackle the theoretical basis.
I became a qualified administrator for Form J in 1996 (prior to Form M). I love the stuff but it's expensive to keep up with all the seminars unless you can get a good tax writeoff, so I haven't done much lately.

the following description was copied/pasted from the publisher's website:

CPP Item Detail - MBTI Manual (A)

MBTI

 

[Maverick]

Someone mentioned this to me before

That must have been me if I remember correctly.

In response to the OP, there's quite a bit of litterature on this already so you should take the time to google it up. Basically, there are no sound statistical reasons for asserting that most people have a preference.

There is little scientific evidence to many of the MBTI's assumptions.

 

[MerkW]

Hello all,

First of all, a disclaimer: I am new to MBTI and to this forum, I have been browsing through threads and did not immediately find anything on the issue, I am concerned with. So, if you guys have already discussed the issue below elsewhere or I am making the wrong assumptions about MBTI, please excuse me.

Anyway, the MBTI assumption of people being either E or I, N or S etc. seems to be valid for most of my entourage. Blackwater is very much ENTP, Economica and I are very much INTJ (even though I completely understand why people could mistake her for being E), Eco's boyfriend and Vortex are really ENFPs etc.

In general, the binarism of the four dimensions seems to be valid, most people are either one thing or the other, and there is very little doubt about what they are. Psychological mechanisms such as self-amplification (people do and improve at what they are already doing) can explain the binarism. The duality is very nice, we can then efficiently describe personalities and understand their social interaction between the various types.

But then, from time to time, I meet people that I have a really hard time classifying. Vortex has a best friend, we are pretty sure about her being F but the other dimensions escape us. I thought she was ENFP, Eco is fairly convinced she is ISFJ, Vortex believes she is xNFx but even though they have known each other since high school, she can't say for sure about the two remaining dimensions.

My concern is now, what does this mean for the MBTI binarism? If the binarism does indeed hold, with everybody being one thing or the other, then typology misclassifications may occur due to lack of knowledge or misunderstandings, but in general a large number of MBTI analysts should agree on the personality type of a specific individual.

So, what I would like to know, is this really what is happening? If a large number of analysts examines one individual, will, let's say 95 % of them or more, agree that this person is indeed e.g. ENFP? Or could it be that 45 % would say Sensing, while the remaining 55 % would pick N?

Do we have any statistics on this issue? I have seen a lot of data on scientists being mainly INxJs, policemen being all S etc, but I have not seen any statistics describing the classification of one specific individual by a large number of MBTI analysts. Please inform me, if you are aware of such statistics on the certitude of the MBTI classification of individuals.

I am sure that everyone (who knows him obviously) will agree that Eco's boyfriend is ENFP, but I could very well imagine that we would all totally disagree on Vortex old friend. What does this mean? Could it be that most people are one thing or the other, but that some are in between? In that case, would it be more appropriate to describe people using percentages, e.g. 65 % N, 35 % S etc.?

Any thoughts or ideas on the subject are most appreciated.

BR

Well, MBTI isn't really a "Yes/No," "0/1," Binary system that you seem to think it is. There are 8 "cognitive functions" that make up each type. There are judging functions and perceiving functions, which can then be further broken down into the attitude of the function (i.e. extroversion/introversion), like so;

1. Judging Functions: 2. Perceiving Functions
- Extroverted Feeling (Fe) - Extroverted Sensing (Se)
- Introverted Feeling (Fi) - Introverted Sensing (Si)
- Extroverted Thinking (Te) - Extroverted Intuition (Ne)
- Introverted Thinking (Ti) - Introverted Intuition (Ni)

So for example, if you have trouble with typing Vortex accurately on the E/I, S/N, and J/P dichotomies, you should try looking at it from a functional perspective. I am too lazy to go off on a rant about Jungian functions, so I will give you some useful websites. Here is one website: Understanding the Eight Jungian Cognitive Processes / Eight Functions Attitudes
Here is another: Jungian Function Theory
And a last one: The Lenore Thomson Exegesis Wiki

When you read through some of these pages you will find that an ISFJ and an ENFP are radically different (almost opposite, in fact), and thus should be easy to tell apart. If it still seems difficult, perhaps Vortex is neither an ISFJ nor an ENFP. INFJ? INFP? Etc. After becoming more familiar with Jung's cognitive processes, it becomes much easier to type people.

 

[ptgatsby]

That must have been me if I remember correctly.

Regarding the distributions of the instrument, it was Fridayeyes that PMed me about it... but I think this has come up a lot between us and others. I'm slightly confused about what the OP was asking though, so I kind of hit it with a broad brush.

This is a common problem... is it the instrument or the interpretation that is the issue? I think the assumption made in the OP is about how the instrument is given (ie: the degree of interpretation), which may be the confusion.

Course, CAPT is not very open about outsiders building their own response interpretations (including using lawyers), so the whole argument is quite involved.

 

[Economica]

But then, from time to time, I meet people that I have a really hard time classifying. Vortex has a best friend, we are pretty sure about this friend being F but the other dimensions escape us. I thought she was ENFP, Eco is fairly convinced she is ISFJ, Vortex believes her friend is xNFx but even though Vortex has known her friend since high school, she can't say for sure about the two remaining dimensions.

I know this is totally tangential to the main point D) but for the record, I don't have an opinion ('fairly convinced') about said friend's type (except for the IF) though I may have given that impression. I remember pointing out that she herself identified as S when she first heard about the dimensions and that our definitely ISFJ friend thinks of her as 'more like me than like all you Ns', but that could easily be true of an INFP as well (with dominant Fi instead of dominant Ni/Ne like most of the Ns the ISFJ was referring to).

That said, the problem of some people escaping classification remains... Though I have to wonder how much is due to our relative typing inexperience. For instance, I don't know any female INFP well enough to compare with Vortex's friend.

Well, MBTI isn't really a "Yes/No," "0/1," Binary system that you seem to think it is. There are 8 "cognitive functions" that make up each type. (...) After becoming more familiar with Jung's cognitive processes, it becomes much easier to type people.

I agree that Nails needs to acquaint himself with function theory (and temperament theory), but I know all the theory and the problem remains. Are you saying you have never been unable to type someone? If so, may I ask how many people you have typed (where the accuracy of the typing has been confirmed by the person and preferably also by those who know the person well)?

I sure hope my failures are due to typing inexperience...

 

[MerkW]

I agree that Nails needs to acquaint himself with function theory (and temperament theory), but I know all the theory and the problem remains. Are you saying you have never been unable to type someone? If so, may I ask how many people you have typed (where the accuracy of the typing has been confirmed by the person and preferably also by those who know the person well)?
I sure hope my failures are due to typing inexperience...

I never doubted thy knowledge of the system. I was merely responding to Nails. And, yes, of course there are many occasions where I am unable to type someone, yet these occasions are usually with people whom I do not know very well. With people I know, I usually am quite sure that the person is either one of two possible types.

 

[Economica]

I never doubted thy knowledge of the system. I was merely responding to Nails.

Oh, I know.

And, yes, of course there are many occasions where I am unable to type someone, yet these occasions are usually with people whom I do not know very well. With people I know, I usually am quite sure that the person is either one of two possible types.

1. One of two possible types? That's not good enough!

2. You may be quite sure, but that's not good enough either. The first year I knew about MBTI there wasn't anyone I didn't think I had typed correctly. :rolli: :blushing: What I asked was: "How many people have you typed where the accuracy of the typing has been confirmed by the person and preferably also by those who know the person well?"

 

[Nails]

Thank you for the responses, so far. I am not quite sure what ptgatsby means by "kurotical" distribution, even though I think we are on the right track.

Let us imagine the following experiment and two possible outcomes: We ask a 100 senior members of MBTI central to type a random high school class with 30 students. 3000 interviews later, all the students have been analyzed a 100 times each and we can study the certitude of the typology. To keep the math simple, we will just study the N/S dimension, not the entire typology.

Outcome I:

There is 100 % agreement for every student in the class. This would mean that 1) the typology does indeed hold and 2) the MBTI seniors really do know their stuff.

Outcome II:

There is > 90 % agreement for 10 students. There is 75 % agreement for 10 others and only 50 % agreement for the remaining 10. (50 % agreement means that a student is typed as 50 % N, 50 % S).

Here we are in trouble, and we must try to determine the origin of the deviations in the results. Possible consequences for MBTI that I can imagine are the following:

a) Humans have no preference for the four dimensions, N does not contradict S in a personality and the MBTI assumptions do not hold.

b) People do have preferences, but these should be analyzed in a more relative manner. We could say that person A is more N than B but more S than C. People should be typed by comparison and percentages would be more suitable for describing preferences.

c) People have preferences, they are indeed binary, but the MBTI central seniors were having a hangover the day the interviews were conducted from too much and the day before.

I guess a) and b) are what ptgatsby means by a problem with the "instrument" and c) is with the "interpretation"?

My feeling, inspired by my own difficulty of classifying people, is that Outcome II of the experiment is indeed very possible, and I would be most interested in more information to find out whether a), b), c) or something else(?) is the explanation for the result.

So people, what are your thoughts on this?

And do you have any idea whether such an experiment has ever been carried out? Statistics on this would be most interesting.

 

[Maverick]

Sorry, but that is the most ridiculous experimental plan I haver ever read, and I have read many.

 

[Economica]

Sorry, but that is the most ridiculous experimental plan I haver ever read, and I have read many.

Real constructive there, Maverick.

 

[ptgatsby]

So people, what are your thoughts on this?

Well, just to be clear - there are a lot of personality dimensions... that has been tested. MBTI has performed factor analysis to confirm their four dimensions - so I think arguing about that aspect needs to be very robust (more than the ability to read preferences at a distance).

In general, the consensus is that personality is a gradient - most instruments in use now are normally distributed (for all intents and purposes, kurtotic distribution is similar (here's a good graph). Either way, MBTI forces bimodal distribution (well, binary, but since it's not always clear, even within MBTI, bimodal is fairly reasonable to call it)...

The reality of this isn't disputed - MBTI calls it the degree of expression of the underlying function... but behaviourally it is certainly a gradient.

However, the methodology you mention really speaks as to the ability of people to type more than anything else.

A better test would be to look at brain stimulation (ie: pdf warning thank you proteanmix) or such to see if there really is a point of threshold that would define a binary preference (in this link, this would relate to a certain chemical balance where Es would look for stimulation - not saturated chemical balance while Is would avoid stimulation - over saturated chemical balance.)

 

[Nails]

Sorry, but that is the most ridiculous experimental plan I haver ever read, and I have read many.

You clearly do not posses a scientific background... Let me explain more thoroughly:

Whenever an experiment in science is performed, the most basic requirement for the experiment to be reliable is an estimation of the uncertainty of the experimental measurement. This estimation is performed by repeating the measurement a large number of times. The average result is calculated and the average deviation from the average result is the uncertainty.

So, if you measure the speed of light 10 times and obtain 10 very different results you can conclude that 1) your measurement setup is not working, or 2) the speed of light is not constant. If 100 MBTI seniors type one high school student, and half of them says N and the other half says S, then you can conclude that 1) the MBTI seniors are poor at typing people or 2) the MBTI assumption of binary preferences does not hold.

If the result of the high school experiment is 50-50, then we must figure out whether the MBTI seniors were indeed having a hangover, or whether there is a problem with the binary preference.

You could then ask "OK, but I already believe in the MBTI assumptions, why would I need to make this boring experiment?" Well, you don't, but that is what differentiates science from religion. If you accept the MBTI assumptions without question, then your subsequent logical conclusions may be flawed simply because the foundation of your system is not rock solid.

Now, don't misunderstand me, I am not comparing MBTI to religion... but a hardcore sceptic might. And without scientific evidence, you won't be able to convince the sceptic that MBTI is actually real and useful.

Well, just to be clear - there are a lot of personality dimensions... that has been tested. MBTI has performed factor analysis to confirm their four dimensions - so I think arguing about that aspect needs to be very robust (more than the ability to read preferences at a distance).

I think we are on the right track, but I am still a bit .

What is this factor analysis you mention? Is it something that supports the binary preference of the four dimensions?

In general, the consensus is that personality is a gradient - most instruments in use now are normally distributed (for all intents and purposes, kurtotic distribution is similar (here's a good graph). Either way, MBTI forces bimodal distribution (well, binary, but since it's not always clear, even within MBTI, bimodal is fairly reasonable to call it)...

The reality of this isn't disputed - MBTI calls it the degree of expression of the underlying function... but behaviourally it is certainly a gradient.

So, are you saying that the binarity is a simplification and that people really should be typed in a more relative manner, (e.g. with percentages)?

I am trying to get a clear yes/no answer.

However, the methodology you mention really speaks as to the ability of people to type more than anything else.

A better test would be to look at brain stimulation (ie: pdf warning thank you proteanmix) or such to see if there really is a point of threshold that would define a binary preference (in this link, this would relate to a certain chemical balance where Es would look for stimulation - not saturated chemical balance while Is would avoid stimulation - over saturated chemical balance.)

You are concerned with the practical issues of the experiment... you are sooo ISTP.

Obviously, I am making the assumption that the MBTI seniors weren't the night before the experiment and that it is somehow possible to employ a reliable typing technique. Whether it is based on interviews, behavioral monitoring or biochemical measurements is currently less important. What is important is only that such a reliable typing technique exists. If it doesn't, then we are entering a philosophical discussion of the metaphysical aspect of MBTI.

BR

 

[ptgatsby]

What is this factor analysis you mention? Is it something that supports the binary preference of the four dimensions?

Factor analysis in general or for MBTI... well, I guess it doesn't matter.

Support for binary preferences wasn't found through factor analysis, but the existance of the four traits was (although there are plenty of attacks to be mounted on factor analysis, it does seem to be valid). Although not as robust as FFM or something similar (due to the existance of pre-defined traits - it was to confirm, not deduce), what happens is a gigantic amount of traits are measured, then blindly seperated into "clumps" of traits... those end up being the four traits.

I gotta run, and won't be on much for the next week or so... but I'll come back to this to give a better explanation just didn't want to leave it hanging.

So, are you saying that the binarity is a simplification and that people really should be typed in a more relative manner, (e.g. with percentages)?

I am trying to get a clear yes/no answer.

IMO, yes, it should be a gradient. The point here was that almost every instrument does this - type simply takes it one step further after the fact.

Obviously, I am making the assumption that the MBTI seniors weren't the night before the experiment and that it is somehow possible to employ a reliable typing technique. Whether it is based on interviews, behavioral monitoring or biochemical measurements is currently less important. What is important is only that such a reliable typing technique exists. If it doesn't, then we are entering a philosophical discussion of the metaphysical aspect of MBTI.

It's more than that. Even strict behavioural measurements need to be analysed in some way to determine groupings - factor analysis is the statistical way of deriving common themes, however direct measuring is vastly more effective at giving a baseline for measurement... factor analysis is very much susceptible to "making the numbers work". This becomes increasingly true the more factors one tries to measure (typically there are about 30 or so personality factors to measure, even before we get into tail end cases or aptitude measurements).

There is no way, currently, to accurately measure the cause of behaviour - we have derived some number of traits that act independently (mostly) from each other, but that's about it. For example, someone who is E (say - positive emotion based) does not influence how Neurotic you are (say - negative emotion based) (this is from the PDF I mentiond earlier). We are starting to measure things in more absolute terms, but it's just a start.

 

[Blackwater]

To my knowledge, any MBTI survey ends in a conversation between the Psychologist and the Testee.

As such, the proof would very much be in the process itself. Just like you'd test a clairvoyant's mettle by asking another clairvoyant the same question(s) and compare the answers, I very much like Nails' idea of sending the same person to 10 different MBTI practitions and then see what comes out. To my knowledge, this has not yet been done.

 

[Nails]

I gotta run, and won't be on much for the next week or so... but I'll come back to this to give a better explanation

I'm still not quite sure what factor analysis is, so I am looking forward to that.

IMO, yes, it should be a gradient. The point here was that almost every instrument does this - type simply takes it one step further after the fact.

Okay, I see. Then the next question that pops up, is how should we treat personalities traits that are near the 50 % border?

Would it be more appropriate to categorize an ENTJ with 51 % N as ENTJ or as an "ExTJ", where the x indicates neutrality?

I am raising this question as I could imagine that two ExTJs with 49 % and 51 % N would be more alike than two ExTJs with 51 % and 100 % N. For these personalities I think the ExTJ label would be more suitable.

Also, do we have any idea of the size of the fraction of the population having personality traits that are "neutral", say in the 40-60 % range?

BR

 

[red13]

Let us imagine the following experiment and two possible outcomes: We ask a 100 senior members of MBTI central to type a random high school class with 30 students. 3000 interviews later, all the students have been analyzed a 100 times each and we can study the certitude of the typology. To keep the math simple, we will just study the N/S dimension, not the entire typology.

I think there are quite a few issues here.

Even if all the practitioners agree on type this is no proof that type is significant. They may, for example, be typing based on some physical feature e.g. Hair colour.

In the literature I’ve read they talk about a person having three types, their tested type, their agreed type and their real type. The tested type typically comes from a multiple-choice test that can be scored. Their agreed type comes from deeper analysis and understanding of type by the target subject. Finally the real type can never be discovered with certainty. I would suggest that in this thread we are talking about agreed type i.e. exam plus study.

I’ve always understood the four types to be dichotomies and it to be meaningless to say that you are for example 49% Introverted. All this means is that you haven’t managed to discover your natural type.

Your example of N/S is interesting as I think it’s the easiest to ascertain on a personal level. I’m 100% certain that I’m an N, the other three letters are all less certain. Although I don’t think I could spot an N by observation, I’m sure after 5 minutes or so of probing and discussion I could make a pretty accurate assessment of the S/N dimension.

Finally though, I think the most important experiment would be to demonstrate that type could be used to make reliable predictions. My particular interest is people working in pairs. I would hope to be able to show that certain combinations of type perform significantly better than other combinations. It would not surprise me to dicover that all 4 dimensions may not be significant e.g. maybe NT + SF proves to be the best (this is a random example I have no reason to believe this to be a particularly good combination)

 

[Nails]

Your example of N/S is interesting as I think it is the easiest to ascertain on a personal level. I’m 100% certain that I’m an N, the other three letters are all less certain. Although I don’t think I could spot an N by observation, I’m sure after 5 minutes or so of probing and discussion I could make a pretty accurate assessment of the S/N dimension.

You first state that there is a difficulty in typing people correctly, but you also indicate that you can make accurate assesments of at least the N/S dimension...

Perhaps, to clarify, do you believe that the dichotomy is appropriate for MBTI and that any discrepancy in typing results of a particular individual is caused by poor typing techniques? To be more precise, if 50 MBTI analysts claim a high school student to be N and 50 others say S, do you think that 50 of them are "wrong"?

Finally though, I think the most important experiment would be to demonstrate that type could be used to make reliable predictions.

I think it would be most useful to be able to make such predictions, and would eventually like to pursue that issue further.

However, these predictions rely on the fact that we can type everyone accurately, and I am still not convinced about that.

BR

 

[red13]

You first state that there is a difficulty in typing people correctly, but you also indicate that you can make accurate assesments of at least the N/S dimension...

I wasn’t trying to say that it’s difficult to type someone. My point was that even though there needs to be a high level of repeatability in the typing, it would be meaningless, from a scientific point of view, unless type could be used in some predictive way.

Perhaps, to clarify, do you believe that the dichotomy is appropriate for MBTI and that any discrepancy in typing results of a particular individual is caused by poor typing techniques? To be more precise, if 50 MBTI analysts claim a high school student to be N and 50 others say S, do you think that 50 of them are "wrong"?

From the reading I’ve done types are dichotomies and therefore if 50 people say S and the other 50 say N then I think 50 have got it wrong. However, although tests and practitioners can give indications to type preferences they are probably only useful if the subject agrees with the typing. Ultimately I think type has to be discovered by the person their self. From my experience of testing and typing individuals there are some people who have been impossible to type. So I think we need to have a category of unTyped and only use those people that are comfortable with their type. Maybe some people are impossible to type, that would not necessarily reduce the value of MBTI for those that can be typed.

I think it would be most useful to be able to make such predictions, and would eventually like to pursue that issue further.

However, these predictions rely on the fact that we can type everyone accurately, and I am still not convinced about that.

After vigorous examination of my type over the last seven years it has never changed from INTJ, although my level of confidence in it may have varied. I’m very certain now that my type as defined by MBTI is INTJ, and I think that many other people are also very confident of their type, although possibly not a majority of people. That doesn’t really matter though, the important question is; can we make behavioural predictions for this group of people that can be typed?