Traits

Mini Review

• A trait is a difference between people in what they feel, do, and think that holds across situations and over time.
• Trait words group into a small number of broad factors, because the things those words name are correlated.
• The Big Five is the set of five broad traits that consistently emerge from factor analysis of personality scores.

Why Personality Traits Cannot Be Observed Directly

• Nobody has ever “seen” extraversion or agreeableness the way we can see the sky or the person sitting next to you.
• Instead, we see someone talking to strangers and speaking up in class. We infer the construct of extraversion from that behavior.
• Construct: A psychological characteristic we want to measure that is not directly observable.
  • Ex: Intelligence.
  • Ex: Stress.
  • Ex: Self-esteem.

Turning a Personality Trait Into a Score

• Operational definition: Defines a construct in terms of precisely how it will be measured in a specific research study.
• The researcher decides what will count as evidence, before collecting any.
  • Ex: Extraversion is defined as the total score on the eight extraversion items of a questionnaire.
  • Ex: Extraversion is defined as the number of words a person speaks during a ten-minute conversation.
• Those two definitions are both reasonable, but they will not necessarily place the person in the exact same spot on the spectrum of the trait.

Self-Report Measures

• Self-report measure: Measures in which participants report on their own thoughts, feelings, or behaviors.
• The person answers questions about themselves, usually by rating how much a statement fits them.
  • Ex: “I start conversations with people I do not know.” Rate from 1 to 5.
• This is the most common source of personality data, because it is fast, cheap, and works on large numbers of people.
• The person being rated has seen themselves in more situations than anyone else has.

The Limits of Self-Report Measures

• People have limited understanding of their own personality.
• Social desirability: The tendency to answer in the way that makes a person look good rather than in the way that is most accurate.
  • Ex: A job applicant rates himself higher on conscientiousness than he actually is.
• People answer relative to whoever they are comparing themselves against.
• People might understand the question differently than the way the researcher meant it.
  • Ex: One person reads “I am a nervous person” as a question about feeling jittery, and another reads it as a question about worrying.

Informant Report Measures

• Informant report measure: Measures in which someone who knows the participant reports on the participant’s thoughts, feelings, or behavior.
• A roommate, a parent, a coworker, or a friend rates the person on the same items the person rated themselves on.
• Informants see the behavior from the outside, so they catch things the person does without noticing.
  • Ex: A woman does not realize she interrupts people, but her coworkers see it every day.

The Limits of Informant Report Measures

• An informant only sees the person in the settings they share, which is part of the person’s life rather than all of it.
  • Ex: A coworker giving an informant report only sees the person in their work environment.
• Informants cannot observe what a person feels or thinks, so they are guessing on those items.
• Informants have their own reasons to answer favorably, especially about family.
• Two informants who know the same person often disagree, which raises the question of which one to believe.

Behavioral Measures

• Behavioral measure: Measures that record participants’ observable actions or performance on structured tasks or in natural settings.
• Somebody counts what the person actually did, so nobody has to remember or judge anything.
  • Ex: The number of times a person starts a conversation during a group task.
  • Ex: The number of days a student turns work in late across one semester.

The Limits of Behavioral Measures

• One recorded session is one occasion, and one occasion cannot tell you a trait.
• Getting enough occasions to average is slow and expensive, which is why these studies use small samples.
• People behave differently when they know they are being watched.
• The same action can come from different traits, so the count still has to be interpreted.
  • Ex: A student who says nothing in a group might be quiet, bored, or afraid of being wrong.

Digital Record Measures

• Digital record measure: Measures that score a construct from records a person generated without being asked to report anything.
• The record already exists, so nobody is answering questions and nobody is being watched on purpose.
  • Ex: The words a person uses in their social media posts.
  • Ex: How often a phone reports that its owner left home.
• These give thousands of occasions per person, which no other source can do.

The Limits of Digital Record Measures

• Only the recorded parts of a life end up in the record, and people differ in which parts they record.
  • Ex: Someone who posts nothing looks the same as someone with no social life.
• What a person puts online is partly a performance, so it carries the same problem as social desirability.
• The link between a digital signal and a trait is found by pattern-matching, so it can hold in one sample and fail in another.
• Collecting this data without asking raises questions the other three sources do not.

The Reliability of Personality Scores

• Score reliability: The consistency of scores (operationalized constructs).
• Reliability is about the score, not about the person and not about the questionnaire.
  • Ex: A person takes the same questionnaire twice in one month and gets close to the same score both times.
  • Ex: Two observers watching the same conversation count a similar number of interruptions.
• A score that jumps around for no reason cannot support any claim about a trait, because traits hold over time and situations.

The Validity of Personality Scores

• Score validity: The accuracy of scores (operationalized constructs) for a particular use.
• Validity is about whether the inference you draw from the score is accurate.
  • Ex: Scores meant to measure extraversion should line up with how outgoing the people who know them say they are.
• A measure is not valid or invalid by itself. The same scores can support one inference accurately and another one poorly.

Reliable but Wrong

• A score can come out the same every time and still measure the wrong thing.
  • Ex: A scale that says you weigh exactly 5.2 lbs every time you step on it is giving you very reliable scores. But those scores are also very invalid!
• Reliability tells you how consistent the score is. Validity tells you how accurate it is. We want both. How do we get that?

Using More Than One Type of Measure

• Each of the four types of measures has different strengths and weaknesses. Thus, using more than one type and seeing if the scores tell the same story is a good way to evaluate their validity.
• When scores from self-report, informant-report, behavior, and digital records agree, we are more confident that we’ve accurately measured personality.
• However, using multiple measure types takes a lot more time and resources. Thus, personality research often relies on self-report measures alone.

In Sum: Measuring Personality Traits

• A trait is a construct, so it is inferred rather than observed.
• An operational definition specifies how a construct will be measured.
• There are four types of measures we use: self-report, informants, behavior, and digital records.
• Reliability asks whether a score is consistent. Validity asks whether a score is accurate for a given use.
• Let’s apply what we know about traits so far…

Activity: Sorting Statements Into Factors

• Each group has 29 statements from the personality questionnaire you scored me on, one per card.
• Sort the cards into piles, one pile per Big Five factor.
• Within each pile, separate the statements that describe the high end of the factor from the statements that describe the low end.
• Ten minutes. Be ready to say which cards were the hardest to place.

Discussion: What Did You Notice?

• Which statements were hardest to place, and which two factors did they fall between?
• Did every group end up with the same number of piles?
• Which statements describe the low end of a factor, and how could you tell?

Activity: Measuring Dr. Smith’s Personality

• I am the one person everyone in this room knows in at least one context (this class).
• So let’s rate my personality and compare!
• 23 of the statements you just sorted, scored from very inaccurate (1) to very accurate (5).
  • Poll: Informant Report

From Sorting the Statements to Scoring a Person

• Sorting the statements is what the questionnaire’s authors did with a factor analysis of thousands of people’s scores.
• Twenty-three of these statements are the ones you scored me on, and that I also scored myself on.
• The next slides compare the two sets of scores.

Two Operational Definitions of One Person’s Traits

• The construct: my standing on four Big Five factors.
• Each operational definition turns that construct into scores from 1 (very inaccurate) to 5 (very accurate) on the same 23 statements.
  • Definition 1: my own scores. This is a self-report measure.
  • Definition 2: the class average of your scores (the 16 of you that did it). This is an informant report measure.

Reliability: Did You Agree With Each Other?

• Reliability: How much did the 16 of you agree in your ratings? One way to check is to randomly split the class in half and see how well the two halves correlate with each other, which is what the figure above does. The closer to the line, the more agreement between the two halves.
• The two halves of the class agree with each other at r = .85, a strong positive correlation. Overall, the class was fairly consistent.
• But how much did you agree with me?

Did You Agree With Me? (On Average)

• The line is basically horizontal, which means there was essentially no correlation/relationship between the class average and my own score across the 23 statements!
• Why?

Social Desirability: How Favorably the Class Scored Me

• You gave a 4 or 5 on 64% of your scores. I gave a 4 or 5 on only 35% of mine.
• Social desirability is the tendency to answer in the way that makes a person look good rather than in the way that is most accurate. We defined it for self-report, and the same tendency applies to informants. I told you this was anonymous, but maybe a part of you still felt the need to rate me higher. ## Did You Agree With Me? (per Factor) {.sm}

• Each factor score is the average of that factor’s statements, with the low-end statements flipped (aka, reverse-scored). (No neuroticism statements were included, so there are four factors here rather than five.)
• My average score and the class average agree on extraversion and openness but differ by more than a point on conscientiousness and agreeableness.
• Notice how spread out your ratings are for conscientiousness and agreeableness. You scored those items less consistently (reliably) than the other factor items.

In Sum: Why Did We Disagree?

• Person AND Situation: you scored the version of me that exists in a classroom. My self-report averages across every situation I am in, which is what a trait is. How far apart the two land depends on how different the classroom is from the rest of my life.
• Interpretation: perhaps we did not interpret the items the same way.
• Does that mean that your scores are invalid?
  • It depends on what you want to do with them! Validity is the accuracy of an inference drawn from a score, so the question is which inference: what I am like in this classroom, or what I am like in general.

What Could Make Either Set of Scores More or Less Valid?

• Which set of scores is likely more valid about how I am as a teacher, and which is likely more valid about me in general?
• Which set would a University find more valuable before hiring me?
• Which set would a counselor find more valuable if I came to them for therapy?
• Which statements would my mom score differently from all of you? Which would she score differently from me?
• Did you score me favorably because I am your professor?
• Did you default to 3 when you had no idea?
• What would raise your confidence in either set?