Lets Talk Distributions

[edcoaching]

...I guess it gets back to the fact that Myers wasn't trying to diagnose or design an instrument that would verify whether the types existed in the first place. She, Briggs and Jung had built their theoretical model to explain the very real differences they saw in people. The MBTI was designed to sort people into the Jungian categories. We don't look for instrument scores or results to verify the theory because you're right--she built it to see what she presupposed it would show. This just isn't how the academic world views testing so they can't wrap their brains around the fact that her instrument meets its goals and works well for its given purposes.

The validity for the MBTI comes from the high percentage of people who agree with their results, but that's only a small part of it. More important are the studies that show the effectiveness of using the theory.

• You don't have to use any instrument to use the theory.
• As far as I know, all of the tools out there are self-reporting instruments, even the NEO-PI, so there's more error variance than on, say, the MMPI
• Validity of the theory arises from the body of studies that show that people behave in patterns predicted by the theory
• Probably more important, when you help people apply the theory to teaming, communicating, coaching, parenting, career search, etc., they become more effective.

Yes, you can use the five-factor models to help people discover essentially the same elements of their personality (the correlations with MBTI scales and NEo-PI scales for Extroversion, Agreeableness, Openness and [I forget the term for the one that correlates with S-N]. But because the five-factor models do get at neuroticism (although some have relabeled the term) they aren't very useful in group settings. People have enough problems getting along without using a tool that labels abnormalities.

So in fact a lot of type professionals are rather independent of any one instrument. Note that Roger Pearman of Certification training in coaching tools, leadership skiils assessment, personal and organizational assessment | Qualifying.org just switched to training people to use the Golden Type Indicator from Pearson. It's the theory we love to use, not any one instrument.

 

[edcoaching]

And again, on the very practical side of using these tools with people, take the Sloan/Big 5-style test on this site and view the language used to describe you.

Big Five Personality Test

Can you imagine handing that stuff out to a team where there's already trust issues or conflict? Or where a manager is trying to control and has mega favorites???

There's a reason that so many type descriptions have lengthy descriptions of strengths first, with carefully worded developmental needs following, an equal number for each type

 

[mlittrell]

myers briggs is a model for something we just dont understand, nothing more, nothing less

 

[ptgatsby]

I see sorting as one form of measurement. It has to do with what possible values the actual situation would create. Probability theory, plus some basic assumptions followed with mathematical reasoning yields which model we use.

Right, and I'm asking - why this model for MBTI/personality theory (what assumptions are you making, and are they fair/etc.) Imagine having to defend it in a thesis, after your observations, your tests, your interviews and the analysis of root factors all show that behavior, attitude, self-selection and so forth are all normally distributed.

More than this, how do you take normally distributed data - data that shows people's preferences - turn it into a strict bimodal model in which you show people's preferences are strongly opposing (ie: you assume strength contrary to data), claim that is who they are, then hedge your bets by adding noise (ie: "has a tendency to", "may") to soften the bimodal fit...?

I don't oppose doing it - but I don't understand how the data supports the model. You can assume it, sure, and would if you gain some understanding. What value do you see?

FWIW, I agree with Edcoaching and the tool usage, the limits of using models in a "friendly" academic way and the effectiveness. I do disagree with the sorting mechanism.

The question, simply put, is: Do you think you are better able to understand people by calling them "F", or by having someone state their strength relative to population (40% F).

Do you understand me better when I say I'm an "I", or when I say I'm in the top 5% of Is. What about if I say I'm in the top 70% for S, 50% for T, 20% for P. Or would you be better off just knowing I'm an ISTP? Is it reasonable justification to call me an ISTP, stripping away those percentages, and then say that "most ISTPs will be thrill seekers", when even a quick glance shows that because of strong I tendencies and weaker S tendencies, this would not likely represent me at all (example for highlighting the issue, not a real assumed trait.) Adding "most ISTPs" is just adding fuzzyness to hide the real root of the fuzzyness - assumption of strength (the bimodal assumption).

 

[Jack Flak]

Strength of preference is a useless concept. How are you measuring that? Not a test, because they do nothing of the sort.

 

[ptgatsby]

Strength of preference is a useless concept. How are you measuring that? Not a test, because they do nothing of the sort.

All tests do, as they contain multiple expressions of underlying traits, giving you the probability that the underlying condition responsible for the trait falls within certain ranges. This remains true if a test only offers binary choices or asks for self-reported strength.

This assumes that the questions are accurately grouped into an underlying cause.

 

[Eric B]

Yes, you can use the five-factor models to help people discover essentially the same elements of their personality (the correlations with MBTI scales and NEo-PI scales for Extroversion, Agreeableness, Openness and [I forget the term for the one that correlates with S-N]. But because the five-factor models do get at neuroticism (although some have relabeled the term) they aren't very useful in group settings. People have enough problems getting along without using a tool that labels abnormalities.

Actually, Openness (i.e. "to new experiences") is the one correlated with S/N.
Conscientiousness is correlated with J/P. But more accurately, Conscientiousness would probably fit Keirsey's Cooperative/Pragmatic better, while Agreeableness (correlated with T/F) would be Role-Informative/Directive. (Both factors do alternate between T/F and J/P). Perhaps Keirsey's theory wasn't big or respected enough for the factor analysts to use those scales. But the concepts are pretty much the same.

Then, there's also the Type Differentiation Indicator, which is a version of MBTI that adds "Comfort/Discomfort" to match Neuroticism. It seems to have more of a specialty use, like for mental hospitals, or something like that. Its subscales were supposedly originally part of Myers' research, but were dropped from being seen as too negative.

 

[edcoaching]

Actually, Openness (i.e. "to new experiences") is the one correlated with S/N.
Conscientiousness is correlated with J/P. But more accurately, Conscientiousness would probably fit Keirsey's Cooperative/Pragmatic better, while Agreeableness (correlated with T/F) would be Role-Informative/Directive. (Both factors do alternate between T/F and J/P). Perhaps Keirsey's theory wasn't big or respected enough for the factor analysts to use those scales. But the concepts are pretty much the same.

Knew I shoulda pulled out the manual...
As for Kiersey, he didn't release any reliability/validity data on his instrument until I believe last spring. It was 2007 or 2008 at any rate. I don't remember how it held up to the other instruments

Then, there's also the Type Differentiation Indicator, which is a version of MBTI that adds "Comfort/Discomfort" to match Neuroticism. It seems to have more of a specialty use, like for mental hospitals, or something like that. Its subscales were supposedly originally part of Myers' research, but were dropped from being seen as too negative.

Yeah...believe me, for widespread distribution talking about problems just isn't a good idea--especially in self-reporting instruments...

 

[ygolo]

Right, and I'm asking - why this model for MBTI/personality theory (what assumptions are you making, and are they fair/etc.) Imagine having to defend it in a thesis, after your observations, your tests, your interviews and the analysis of root factors all show that behavior, attitude, self-selection and so forth are all normally distributed.

More than this, how do you take normally distributed data - data that shows people's preferences - turn it into a strict bimodal model in which you show people's preferences are strongly opposing (ie: you assume strength contrary to data), claim that is who they are, then hedge your bets by adding noise (ie: "has a tendency to", "may") to soften the bimodal fit...?

I don't oppose doing it - but I don't understand how the data supports the model. You can assume it, sure, and would if you gain some understanding. What value do you see?

FWIW, I agree with Edcoaching and the tool usage, the limits of using models in a "friendly" academic way and the effectiveness. I do disagree with the sorting mechanism.

The question, simply put, is: Do you think you are better able to understand people by calling them "F", or by having someone state their strength relative to population (40% F).

Do you understand me better when I say I'm an "I", or when I say I'm in the top 5% of Is. What about if I say I'm in the top 70% for S, 50% for T, 20% for P. Or would you be better off just knowing I'm an ISTP? Is it reasonable justification to call me an ISTP, stripping away those percentages, and then say that "most ISTPs will be thrill seekers", when even a quick glance shows that because of strong I tendencies and weaker S tendencies, this would not likely represent me at all (example for highlighting the issue, not a real assumed trait.) Adding "most ISTPs" is just adding fuzzyness to hide the real root of the fuzzyness - assumption of strength (the bimodal assumption).

Actually, questions like this are why I brought up the topic. I have the same sort of questions. I am really rather open to answers in general.

I have the same doubts about Myers-Briggs Theory as I do about FFM.

What the assumptions are in Myers-Briggs Theory are easier to find, but there is also a lot of misinformation. Many FFM supporters say they make no assumptions. That is nonsensical to me. It is simply not possible.

I am not saying that you are one of the FFM people. I certainly don't want to make it personal.

I would like to discover the logic behind the "truth" of either model.

I may not have made myself clear with the "object detection" example. I was simply illustrating that "bad measuring equipment" could be the source of the Gaussian, and non modality--in fact, Gaussians are what "bad measuring equipment" tends to produce.

Take pure randomness, break it up into groups, and average the groups, you will get a Gaussian. This is mathematical fact. In fact, which Gaussian you get is based purely on how you define the groups--again Mathematically provable.

Generally speaking you cannot "support" inferences by looking at "data" alone. Data simply describes or summarizes what is seen. You could be seeing what you are seeing for any number of reasons, including broken equipment.

I realize that many FFM advocates will say it is simply describing the data seen in personality surveys. I have no issues with that in particular. But you could really just be measuring they survey and not the people taking the survey.

I don't think I need to remind you off all they polls taken about political and economic beliefs. Quite often, what is being measured is the reaction to survey questions, more than anything else.

Survey questions produce Gaussians in general (unless you find a particular question that creates a repeatable anomalous response).

The anomalous response is the phenomenon we we are after when sorting, but are bad when we want honest surveys of the population.

"Rate how much you agree with the following statement from 0-100, 'I enjoy killing babies.' 0 is completely disagree, 100 is completely agree."

This, I suspect, will be bimodal (or multi-modal), in it's response, with one mode strongly clustered near 0, and another one somewhere else. Those people who answered somewhere else are a different kind of survey responder. We won't know why. But they are different in some way (at least at the moment they responded).

In a nation divided between those who support intelligent design and those who don't, you will also likely see a bimodal distributions for statements like, "I support the teaching of intelligent design along with ." If this was a more ordinary issue, you would get a more Gaussian distribution.

If you are aiming to sort the people taking the test, your very questions on the survey will be different from surveys for central tendency.

Note also, the reason I brought up the Central Limit theorem is that when your survey is based of some sum or average of scores on individual questions like these, you will make the resulting distribution more Gaussian because that is simply a mathematical fact--Summing creates Gaussian distortion.

If you're intention is to create a sorting mechanism, questions that don't evoke a bi-modal response are actually noise.

This is the way science always works, you set-up your equipment to find the phenomenon you are looking for. The equipment is meant to amplify the signal you are looking for, and filter the other stuff because the other stuff is noise.

The flip side of-course is if you believe Myers-Briggs Theory is nonsense, and that MBTI researchers are like ESP researchers in that they keep looking for something that doesn't exist. That is a different issue.

I find it hard to find research on establishing the "truth" of any personality model.

 

[Eric B]

Knew I shoulda pulled out the manual...
As for Kiersey, he didn't release any reliability/validity data on his instrument until I believe last spring. It was 2007 or 2008 at any rate. I don't remember how it held up to the other instruments

So a factor analysis can't be done until the person releases this information?
That would make sense. Wonder if anyone will try to redo the analysis with Keirsey's additional factors, now.

 

[edcoaching]

So a factor analysis can't be done until the person releases this information?
That would make sense. Wonder if anyone will try to redo the analysis with Keirsey's additional factors, now.

It's more that Kiersey's instrument was really nothing more than a parlor game as long as there was no reliability/validity data.

 

[ptgatsby]

If you're intention is to create a sorting mechanism, questions that don't evoke a bi-modal response are actually noise.

That is a very good point. I hadn't quite thought of it that way before, so it made me pause.

I'm not a FFM person, really, so I won't defend it. It does have less assumptions than MBTI theory, I will say that, but the methodology used from the outset makes it prone to certain biases. I do use it every bit as much as I use it as MBTI. To me, both are a form of language. Something used to describe things, where 'things' are just mutally agreed upon concepts.

I do like FFM for this purpose, simply because it is much easier to use in putting people into a framework. I don't do coaching and never will - I'm not a people person I also don't care about helping them with the tool, or avoiding offending them.

I do believe in using the best tool possible to understand something. I think the main difference between us is that I'm excessively pragmatic - I actually don't care about the validity of the tests themselves. I'll use MBTI theory to explain conceptual theories - and yet, I strongly disagree with MBTI theory. I'll use the MBTI test results to explain something else, even though I don't believe it should be used as a sorter. I'll use FFM when I can relate it to other research. I've been asked why I sometimes use them and sometimes attack them. It just depends on if I think there is a better tool available. Otherwise, I use what I can.

However, your statement here brings up a very good point about the validity of testing, and I would say strongly supports Ed's viewpoint on applying them. I'll have to think on it for a while... it doesn't change much for my use, but it does change my perception on reported scores.

 

[edcoaching]

Actually, questions like this are why I brought up the topic. I have the same sort of questions. I am really rather open to answers in general.

I have the same doubts about Myers-Briggs Theory as I do about FFM.

What the assumptions are in Myers-Briggs Theory are easier to find, but there is also a lot of misinformation. Many FFM supporters say they make no assumptions. That is nonsensical to me. It is simply not possible...

If you're intention is to create a sorting mechanism, questions that don't evoke a bi-modal response are actually noise.

This is the way science always works, you set-up your equipment to find the phenomenon you are looking for. The equipment is meant to amplify the signal you are looking for, and filter the other stuff because the other stuff is noise.

The flip side of-course is if you believe Myers-Briggs Theory is nonsense, and that MBTI researchers are like ESP researchers in that they keep looking for something that doesn't exist. That is a different issue.

I find it hard to find research on establishing the "truth" of any personality model.

Okay, I've got an MBA in finance and only took basic stats so...I'll admit I can't answer all the questions you're posing!!!

But...It's simply a fact that Myers believed that Jung's theory accurately described how normal people take in information and make decisions. She did not design the instrument to see whether reality bore out the theory. Rather, she wanted to make the theory accessible to people so they could improve their lives, careers, relationships, etc.

Coincientally I'm reviewing Gordon Lawrence's (People Types and Tiger Stripes, retired UF Gainesville professor, worked with Myers....) newest manuscript, prepublication. In it he cites two key studies comparing five-factor and MBTI, both reported in Measures of the five factor model and psychological type: A major convergence of research and theory, James Newman, ed., 1996, CAPT. Remember the NEO-PI was developed based on trait theory, where I is considered lack of E. MBTI is type theory and is either/or. You have a preference and develop skills as opposed to behaviors being driven by traits...

At any rate, to compare the two instruments MBTI scores were made continuous and the high correlations were found. (Extraversion/EI: .70, Openness/SN, .70, Agreeableness/TF, .45, Conscientiousness/JP, .47--all at the p<.005 level of significance...) Mitchell investigated whether type then, really described traits. To do this he looked at 1,568 randomly chosen Step II results where the sbscores are on a continuum to see if they were explainable as traites, i.e., no type influence was exerted on teh scores. The resource above gives the full results but in summary, Mitchell wrote "The analyses presented here yield clear evidence of Jung's theory, his typology...There were clear differences between the trait and typological predictions, sufficiently clear, in fact, that there is no middle ground between them. There are few times in teh behavioral sciences, and especially personality research, when one encounters such dramatic, clear-cut differences bwtween two opposing viewpoints in as many different empirical tests aw were presented here. The results, in every case, support the MBTI version of the Jungian typology and refute a trait interpretation of its dimensions..." And so on.

If you search the bibliography at CAPT: Training, Books, Research for MBTI, Archetypes, Leadership, Psychological Type. you can find the study under Newman--apparently they presented at one of the type conferences, so you could dismiss it as biased or try to find a copy. They make copies at cost...

 

[edcoaching]

That is a very good point. I hadn't quite thought of it that way before, so it made me pause.

I'm not a FFM person, really, so I won't defend it. It does have less assumptions than MBTI theory, I will say that, but the methodology used from the outset makes it prone to certain biases. I do use it every bit as much as I use it as MBTI. To me, both are a form of language. Something used to describe things, where 'things' are just mutally agreed upon concepts.

I do like FFM for this purpose, simply because it is much easier to use in putting people into a framework. I don't do coaching and never will - I'm not a people person I also don't care about helping them with the tool, or avoiding offending them.

I do believe in using the best tool possible to understand something. I think the main difference between us is that I'm excessively pragmatic - I actually don't care about the validity of the tests themselves. I'll use MBTI theory to explain conceptual theories - and yet, I strongly disagree with MBTI theory. I'll use the MBTI test results to explain something else, even though I don't believe it should be used as a sorter. I'll use FFM when I can relate it to other research. I've been asked why I sometimes use them and sometimes attack them. It just depends on if I think there is a better tool available. Otherwise, I use what I can.

However, your statement here brings up a very good point about the validity of testing, and I would say strongly supports Ed's viewpoint on applying them. I'll have to think on it for a while... it doesn't change much for my use, but it does change my perception on reported scores.

Just curious...is it the binary (or did we agree bimodal is appropriate here) distribution that you disagree with? Or that Jungs theory of preferences for Perception and Judgment accurately describe how minds are organized, albeit trait or type?

 

[ptgatsby]

Just curious...is it the binary (or did we agree bimodal is appropriate here) distribution that you disagree with? Or that Jungs theory of preferences for Perception and Judgment accurately describe how minds are organized, albeit trait or type?

The issue I have is the relationship between MBTI and MB theory. I don't believe they support one another.

However, if you are asking my viewpoint, I disagree that the bimodal (and most notable, the binary) distribution is the correct model to assume for people. I realise where ygolo is coming from, so it may be feasible that it is correct on some dimensions. Even if that was the case, absent of the ability to find clumping around the traits, I don't believe that people are best served by sorting. So I have a lot of levels of disagreement.

But as far as it goes, in this thread, I'm arguing that MBTI data does not support the sorting theory (that is, MBTI <> MB's functional theories).

I remember HR at work going through various instruments and methods, and MBTI was rejected for much the same purpose. Most of the company ended up using the predictive index instead, and only with our manager, to aid in understanding the work relationship we had with them. FWIW, I thought this was a much better approach (but obviously not suitable for group dynamics, whereas I do think MBTI is suitable, accurate or not, because it fosters overcompensation towards understanding others.)

 

[edcoaching]

The issue I have is the relationship between MBTI and MB theory. I don't believe they support one another.

However, if you are asking my viewpoint, I disagree that the bimodal (and most notable, the binary) distribution is the correct model to assume for people. I realise where ygolo is coming from, so it may be feasible that it is correct on some dimensions. Even if that was the case, absent of the ability to find clumping around the traits, I don't believe that people are best served by sorting. So I have a lot of levels of disagreement.

But as far as it goes, in this thread, I'm arguing that MBTI data does not support the sorting theory (that is, MBTI <> MB's functional theories).

I remember HR at work going through various instruments and methods, and MBTI was rejected for much the same purpose. Most of the company ended up using the predictive index instead, and only with our manager, to aid in understanding the work relationship we had with them. FWIW, I thought this was a much better approach (but obviously not suitable for group dynamics, whereas I do think MBTI is suitable, accurate or not, because it fosters overcompensation towards understanding others.)

And the MBTI is only accurate for about 75% of the people who take it, which is why the best-fit process is part of the ethics of using it...It is so superficially used in so many workplaces that I'd rather people used other tools...

And if it's used as you are/aren't vs. you prefer __ and develop skills with the other preference it definitely doesn't serve anyone well.

 

[ygolo]

If you search the bibliography at CAPT: Training, Books, Research for MBTI, Archetypes, Leadership, Psychological Type. you can find the study under Newman--apparently they presented at one of the type conferences, so you could dismiss it as biased or try to find a copy. They make copies at cost...

Thanks edcoaching. I wound the paper in the list from searching. To request a copy do we call them?

I think my interest in type and its application have been renewed.

Its funny, I did get a sense that FFM was based on some form of trait theory, but I didn't see it explicitly mentioned in the web descriptions I read so far. It would be interesting to learn the details of the trait theory.

I doesn't surprise me that persistent personality traits and persistent archetypes are at odds with each other despite the correlations. One could make analogy to phenotypes and genotypes in biology.

 

[INTJMom]

Gee. It's too bad you guys have been talking way over my head for most of this thread.
I'd really like to understand the conclusions you have come to.

 

[edcoaching]

Thanks edcoaching. I wound the paper in the list from searching. To request a copy do we call them?

I think my interest in type and its application have been renewed.

Its funny, I did get a sense that FFM was based on some form of trait theory, but I didn't see it explicitly mentioned in the web descriptions I read so far. It would be interesting to learn the details of the trait theory.

I doesn't surprise me that persistent personality traits and persistent archetypes are at odds with each other despite the correlations. One could make analogy to phenotypes and genotypes in biology.

Yup. 1-800-777-capt.

 

[ygolo]

Gee. It's too bad you guys have been talking way over my head for most of this thread.
I'd really like to understand the conclusions you have come to.

We didn't come to any major conclsions as far as I could tell.

The basic question comes down to:
1) Are people all essentially the same with varying degrees of a number of traits, or
2) Are people fundamentally different from each other in that they come from distinct psychological archetypes
?

The last paper edcoaching mentioned purports to both:
1) show that the archetype assumption and trait assumption are measurably different in emperical terms.
2) show that the data supports the achetype assumption.

We could dismiss it as biased or take in what the argument is (and then dismiss it as biassed, if we want).

Now, we are all human, and we share humanity, that is not what the question is about.

This sort of distinction is among the few times, I have felt the familiar feeling of "science" happening in personality theory.