Comments by "EebstertheGreat" (@EebstertheGreat) on "UsefulCharts" channel.

  1. 97
  2.  @ronnymaccaroni  "And most people doesn't really fall in-between. Usually you are clearly on one side of the scales." No they don't. It is an empirical fact that most people fall near the middle of the raw score in the scale. This has been known at least since the 1960s. It's hard to find good sources with full text available online, but "Bipolarity in Jungian Type Theory and the Myers--Briggs Type Indicator" is one such article (full text on Research Gate). You can check the citations and cited bys for some other good sources. The general conclusion of all the abstracts I was able to find on the subject was that none of the MBTI subscales are bimodal, while some claim the E/I subscale to be weakly bimodal (with later papers alleging this to be an artifact of early forced-choice procedures). For instance, consider this paragraph from the introduction: "The forced-choice format of the MBTI is appropriate only if the MBTI dimensions are truly bipolar. However, the assumption of bipolarity of the attitude and function pairs has been challenged. Metzner, Burney, and Mahlberg (1981) argued that the oppositionality of the functions cannot be assumed on theoretical grounds and has not been adequately tested empirically. They and Jarrett (1979) pointed out that even if neither the perception nor judgment functions can be used simultaneously, it does not necessarily follow that they cannot be equally well developed in the same individual. For example, it is conceivable that an individual could be equally capable of and prone to approach problems from a logical (thinking) or subjective (feeling) perspective." They performed an experiment where people gave responses not in the yes-no format but on a 1-5 ("Likert") scale, reflecting the possibility that some people are near the middle on any given item. Check out this paragraph from the conclusion. "The results provide very little support for the bipolarity of the Jungian pairs of functions and attitudes. When subjects were free to state each of their preferences independently, a total of 26.3% scored high on both aspects of at least one of the three MBTI dimensions tested. For each dimension, the number of subjects who scored in the top third on both attributes was not significantly different than would be expected if high scores on one attribute are unrelated to high scores on its supposed opposite. In addition, when preferences were measured in a Likert format, superior (most preferred) and inferior (least preferred) functions were not dimensional opposites for 28% of the sample. Furthermore, correlations between Likert scores for attributes on the same MBTI dimension were low and, in the case of Thinking and Feeling scale scores, statistically significant in the positive direction. Scores for the Introversion and Extraversion scales yielded the only significant negative correlation, and the size of this correlation was quite modest; the proportion of shared variance between the Introversion and the Extraversion scales was only 13%." In other words, if you are more strongly Thinking, then you are also more strongly Feeling, which is totally contrary to the idea that these are diametrically opposed (in fact they are mutually reinforcing). If you are more strongly Sensing, that has no bearing on how strongly you are Perceiving, and the same for Sensing/Intuition. The only minor support was that highly Extroverted people were likely to be slightly less Introverted, but only slightly. This "type theory" is based on a fundamentally bad assumption that, as you say, "most people doesn't really fall in-between." In fact, most people do. Other studies compare raw scores to a normal distribution and find a good fit. At a minimum, it is clearly a centrally-peaked distribution, meaning "falling in-between" is the most common result. You are precisely wrong in your final claim. The subscales on their own may be useful (or they may not), but the typology is definitely useless, because of this basic misunderstanding of human personality variation. It's just like how a "size" scale could be meaningful but dividing people into large and small "types" cannot be. But it's actually worse than that, because some of the scales that are supposed to oppose each other are not. What if you found out that people who scored extra high on your "small" scale also scored extra high on your "large" scale. That would show the scale was invalid! So it is with the MBTI so-called dichotomies.
    20
  3. 11
  4. 9
  5. 7
  6. 7
  7.  @km_studios  But being right- or left-handed is an actual binary. The overwhelming majority of people are one or the other. Of course there are still varying degrees, with many people being broadly competent with both hands, but the preference is still not subtle. If you found some objective measure of handedness and plotted a histogram, there would be one big peak on the right with a smaller and somewhat less prominent peak on the left, with only a small number of people near the middle. The personality scales (both the Big 5 and the MBTI) are exactly the opposite. For any given scale, a significant majority of people are near the middle, with only minorities at the ends. This means that most people are not right- or left-handed, but rather ambidextrous with a slight or even no preference. These people are all much more like each other than they are like either rare extreme. However, they get split right down the middle and lumped with the extremes, making the result basically meaningless for the majority of people. Like, everyone is either taller or shorter than average, in a technical sense. But that doesn't mean there is a "tall height type" and a "short height type," and everyone must be placed into one or the other. A better categorization would have a "middle height type" containing most people, but even that would be silly, because there are no types, only a continuum. There are no "height types" and there are no "personality types." EDIT: I should point out that the "preference" thing was specifically for handedness. There is no evidence types correspond to any particular kind of preference. They just refer to whether you are above or below the median on some arbitrary scale. It could well be that 80% of people have the same "preference" on that scale (whatever that may mean), but MBTI attempts to split them down the middle anyway, putting lots of people on the wrong side. More to the point, each of these categories is just a factor, not really a thing to do or agree with. It is an aggregate of many small behaviors and opinions that are known to be correlated with each other. It's as if I had a "size" scale that looked at a bunch of answers to questions to guess roughly how big you are, noting that various properties like height, weight, armspan, waistline, and shoe size are correlated. This scale would have validity, but it wouldn't make sense to describe being on one side or the other of the scale as a "preference."
    6
  8. 5
  9. 4
  10. 4
  11. 3
  12. 3
  13. 3
  14. 2
  15. 2
  16. 2
  17. 1
  18. 1
  19. I don't want to harp on the typology too much, and I understand why Matt chose to use the MBTI here. He is effectively using it as a quick and dirty proxy for personality scales with higher validity and reliability. The concern of reliability here is not significant, since he is gathering large datasets in a manner that is fundamentally unreliable anyway. The low validity is the most serious problem with this approach, especially the E/i scale which mixes neuroticism and introversion, as he mentioned. There is no way around this problem, MBTI is just not the ideal scale to use here. But since work had already been done correlating religiosity and Big 5 personality traits, there was no reason to retread old ground. (And for what it's worth, more utility studies of MBTI are needed anyway, whether for atheism or anything else.) A more basic problem with MBTI is one that Matt mentioned but maybe didn't explain in much detail. Instead of scoring people on separate scales, the MBTI places people firmly in one extreme or the other. For example, the whole premise of dividing people into "introverts" and "extroverts" presumes that most people are clearly one or the other, with only a minority somewhere in the middle. The reality is the opposite, with most people scoring near the middle and only minorities scoring near either extreme. In other words, most people are neither introverts nor extroverts but somewhere in the middle, and even people who are barely on one side or the other are being done a disservice by being told they are entirely in one "type" instead of being told the truth that they are basically average in that respect. How would this system deal with your True Neutral DnD character? It couldn't. Such a thing doesn't exist according to Myers–Briggs. Nevertheless, while this is devastating for the test's intended function (determining individuals' personality "types," given that such "types" evidently don't exist), it is not such a big problem for this sort of population-level analysis. If all we are looking for is trends anyway, then those trends should still show up even after this artificial threshold is applied, though the experiment will lose some statistical power. A final problem with MBTI, also not relevant to this study, is in the pseudoscientific description of what the types "mean," and of the "functional stack." As Matt said, this relies on a modified version of an old Jungian understanding of psychology that was largely conjectural with little to no supporting evidence. Not even all MBTI enthusiasts still think this has any validity.
    1
  20. 1
  21. 1
  22. 1
  23. 1
  24. 1
  25. 1
  26. 1