Chapter 7 Sixteen Personalities or Five Factors?
Before you read this chapter, you might want to go ahead and take one of the personality tests based on the Meyers-Briggs Type Indicator (MBTI) categories and/or the five-factor model of personality (also called the Big Five). There is only one “official” MBTI, which is a commercial product. However, there are several free alternatives online which use the same typology classification. There are also several variations of the Big Five.
Test yourself:
- MBTI-style at 16personalities.com or
- Big Five
- http://www.personal.psu.edu/~j5j/IPIP/
- https://bigfive-test.com/
- https://openpsychometrics.org/tests/IPIP-BFFM/
- General information about these test items: https://ipip.ori.org/
I will only minimially describe the MBTI and the Five Factor Model (FFM, or Big Five) here, in terms of the topics we have been discussing. There are many resources for learning more about these personality tests. Some are referenced under further reading.
MBTI
The MBTI will categorize people, based on their responses, dichotomously along each of four dimensions, also called “scales.” These are:
- Extraversion-Introversion (E-I)
- Sensation-Intuition (S-N)
- Thinking-Feeling (T-F)
- Judging-Perceiving (J-P)
Thus there are sixteen possible combinations, for example “INTP”. Each person is assigned to one of these sixteen personalities. Many online tests will provide you with a report to help interpret your classification. That is, the four dimensions are understood to come together in some holistic picture of your “type.”
Big Five
The term “Big Five” is a commonly used term for the five-factor model of personality. Based on responses to questionnaires, people are assigned a numerical score along five dimensions (also called scales or factors!)
- Neuroticism refers to the tendency to experience negative feelings.
- Extraversion is marked by pronounced engagement with the external world.
- Openness to Experience describes a dimension of cognitive style that distinguishes imaginative, creative people from down-to-earth, conventional people.
- Agreeableness reflects individual differences in concern with cooperation and social harmony. Agreeable individuals value getting along with others.
- Conscientiousness concerns the way in which we control, regulate, and direct our impulses.
Fun fact: both OCEAN and CANOE are mnemonic devices that can help you recall the names of the Big Five dimensions.
Since the results of a Big Five test, such as the IPIP-NEO, are five numbers, you don’t get assigned a personality “type” by these tests. Rather, you may be provided with an explanation of what it means to score high (or low) on, say, Extraversion. You may have noticed that extraversion (occasionally spelled “extroversion”) appears on both the MBTI and the Big Five.
Twenty Questions (about Extraversion)
Suppose, for whatever reason, we want to identify a person’s extraversion. We may want either (a) to classify them as extraverted or not (i.e., introverted), or (b) to quantify a degree of extraversion, say on a scale of 0-100. Why not then just pose the question in the following way? In the first case:
- Choose the one that describes you: Extraverted | Introverted
or, in the second case,
- Locate yourself on the scale: Extraversion 0 . . . . 50 . . . . 100
Personality tests, such as those we’ve discussed above, do not ask questions like these. Rather, they include many different questions, sometimes twenty or even more, about things like going to parties, making friends, and drawing attention to oneself. Why ask twenty questions instead of just one?
Recall from the great toilet paper debate that it was not necessary to ask twenty questions to know whether you were an over-hanger or an under-hanger. However, when we discussed digitidiness, we suspected that two different questions may have both been getting at the same latent factor. The situation here, in the real-life domain of personality testing, is similar.
Golda asks: So the idea is to design questions that are smarter than the user? Do we not trust people’s abilities to self-categorize? Maybe these tests would be more effective if they allowed people to draw their own conclusions about themselves?
Sidney says: Developers of these questionnaires acknowledge that it is not as easy for people to self-categorize (what do you really mean by introverted? is that a bad thing?) as it is to describe how they feel at parties. Furthermore, category labels may seem socially desirable to the respondent, leading them to make a certain choice for “appearances.” So, to some extent yes, psychologists may not always trust people to self-categorize. But I think they would still defend everyone’s right to draw their own conclusions from the results of a personality test.
Psychologists believe that extraversion is an underlying factor invented the idea of extraversion to explain patterns of behavior, including patterns of responses to questions about how people feel in various situations. Such as enjoyment or lack thereof in being the center of attention. The use of indirect evidence such as questionnaire responses to make inferences about psychological traits is the main task of psychological measurement (also called psychometrics). The main challenge of psychological measurement, perhaps even the reason for its existence as a method and field of study, is that human beings are noisy.
Put another way, you cannot expect a deterministic relationships between a person’s response or behavior (how a person feels or acts) in one situation and how they respond or behave in another. Even someone we may want to call an extraverted person is not always extraverted. And an extraverted person might not always answer questions about their feelings in the same way. It is also hard to directly observe extraversion or to ask people to simply self-report it. Extraversion manifests itself differently at different times and in different contexts. Whether this noise is due to some mysterious internal process, like a coin flip in your brain, or due to many unnaccountable external factors, like whether you slept poorly that day, we can’t really say. What we can say is that human noisiness manifests itself as measurement error when we try to measure things like extraversion using questions and other observations.
Variability in human behavior is one of the factors that may contribute to measurement error. If the same person’s behavior varies randomly over time, then our observations at any given time are subject to this variation. But we could also imagine having measurement error sources that come from ambiguously worded questions, which would still lead to errors even if behavior were stable. Say someone is asked if they engage in risky health behaviors. Even if their behavior is very consistent (for argument’s sake), different people make have different perceptions of what counts as risky. So in that case measurement error results from (variation in) the interpretation of the question rather than from variation in risk-behavior over time.
The word error makes it sounds like there is a right answer, and that (personality) tests get it wrong. This is, indeed, one view (called true score theory in psychometrics). However, you don’t have to believe that there is a right answer. For example, you can believe that human beings have some amount of inherent unpredictability. But just because human behavior is not perfectly predictable, just because observations are not deterministically related, we can still say that—in spite of measurement error—some patterns do remain. The responses to different questions about extraversion, for example, are associated with one another. People who say that they feel comfortable at parties are also likely to say that they make friends easily.
But what’s the point?
Trying to describe people in terms of kinds or numerical scales is complicated. Why do we even bother? It’s tempting to say that we just want to understand ourselves better, and that is certainly a reasonable answer. Sometimes, though, we want to predict how someone will act in the future, perhaps in a situation that differs from one that they have faced in the past. In that case, we can’t exactly use the past to predict the future, unless we do so by making inferences about underlying traits from past behaviors and then predicting how someone with those particular traits would act in a new context. This purpose drives some uses of tests based on the MBTI and the Big Five, for example by employers or career counselors. However, although the MBTI is often used for these purposes, one should exercise caution in doing so (Pittenger 1993). You should certainly not assume that all personality tests do an equally good job of providing information for the desired inferences.
According to the standards of the American Psychological Association (American Educational Research Association and American Psychological Association and National Council on Measurement in Education and Joint Committee on Standards for Educational and Psychological Testing 1999), whenever psychological tests are used for some specific purpose (e.g., employment, admission to a school or hospital, or even in court) there must be a valid argument for the intended purpose of the test scores. This validation argument will usually involve many facets, including how consistent the results of the test are, whether it is a fair test for all groups of people, whether test scores really are associated with relevant outcomes in the domain of use, and so on. These arguments, and challenges to them, are all part of what we mean by validity.
Recap
So far, in pursuit of our question “How many kinds of people are there?”, we have not been overly concerned with specific proportions or probabilities. Our discussion has been rather metaphysical. We have tried to understand how differences that we observe among people can be expressed in terms of kinds (categories) or quantities (numbers), which are themselves different kinds of data.
We have looked at dichotomous questions of both frivolous (toilet paper orientation or peanut butter preference) and less frivolous (attainment of Bachelors degree, employment status) kinds. Similarly, we’ve considered numerical variables from poopiness and crappiness to neuroticism and conscientiousness. We’ve seen that categorical questions can include more categories than two (e.g., peanut butter preference, if we include “hate all” and “don’t care”, but also ethnicity or neighborhood-type). We saw that numerical variables can sometimes be discretized and thus converted into categories, such as when income is classified as high or low, when age is converted to “Generation Z”, or when a score on a personality test assigns you the value “Perceiving” as opposed to “Judging.”
We saw that categorical questions can naturally classify people into types, and that when we combine different kinds of questions, the number of types/kinds/buckets can increase. But we also observed that the number of observed types does not necessarily increase to its mathematical maximum because answers to different questions can be associated. In other words, the number of factors or dimensions can be smaller than the number of questions. (But it can also be equal to the number of questions, in simple cases).
We also saw that numerical variables can be used to classify people. This can happen when we use cut-scores as thresholds and assign people above a threshold a value like “pass” or “abnormal” or “introverted.” The distribution of a numerical variable can be observed using a histogram, and sometimes a histogram is observed to have two bumps, which suggests a mixture of two different groups. But we also saw by plotting points on a coordinate plane of poopiness and crappiness that numerical data can cluster into groups, even when we don’t see tell-tale signs of those groups along any single variable. The notion of clustering is all about identifying groups of data points that might designate types. According to some researchers, types can actually be recovered in the five-factor (Big Five) model of personality, which until recently was strongly thought of as not having types.
You’ve probably figured out by now that this course is less about answering the Big Questions than about understanding how they might in principle be answered. What does one have to understand to even formulate an answer? Different people may indeed come to different answers about how many kinds of people there are. Hopefully you feel better equipped to reason rigorously and to discuss more precisely the evidence used to support any particular claim.