human personality (1)

What is the subject of personality science?

Personality science is the study of naturally occurring individual differences in recurring psychological tendencies

It focuses on how people consistently differ from one another across time and situations, how these differences form broader patterns (traits), what causes them, and what consequences they have

How is personality science distinct from other psychology fields?

Focuses on variation between people, not similarities

Treats variability as the main subject, not error

Studies everyday, naturally occurring differences, not just lab effects

Looks at broad patterns across many variables, not isolated processes

How is personality science similar to other psychology fields?

Still aims to discover general laws of human psychology

Uses scientific methods (measurement, correlation, data)

Studies similar content (thoughts, feelings, behaviour), just at a different level

Uses variability to infer universals (e.g. if traits correlate with outcomes)

Why do personality differences exist?

Evolutionary advantage: variation allows selection of better traits which leads to adaptation

Innovation: different ways of thinking/behaving generate new solutions

Learning and adaptation: people learn from others differences

Functional diversity: different roles/jobs require different traits

Social benefits: promotes tolerance and better societal functioning

What are ABCDs?

Stable tendencies (probabilities) to feel, behave, think, and desire in certain ways

Abstract, generalised across time and situations

What are abcds?

Specific momentary states (e.g. crying today, feeling anxious now)W

what is the A in ABCD?

Affects (feelings/emotions)

e.g. feeling anxious, happy, sad

What is the B in ABCD?

Behaviours (actions)

e.g. talking a lot, helping others, arguing

What is the C in ABCD?

Cognitions (thoughts)

e.g. worrying, planning, believing things

What is the D in ABCD?

Desires (wants/motivation)

e.g. wanting success, wanting approval, avoiding risk

What properties must ABCDs have?

Consistency over time
Consistency across situations

How are ABCDs studied?

Repeated measurements (e.g., questionnaires)

Compare individuals across:

• time (test-restest correlations)

• situations (cross-situational correlations)

Use correlation as main tool

Evidence of ABCDs

moderately consistent over time

• Mottus 2017 - .53 over five years

• Mottus 2019 - .37 over 15 year interval

consistent across situations

• Leikas 2012 - test if people behave like the same person in 4 different situations

  • ~ .35 for specific behaviours (talking a lot)

  • ~ .43 for broader tendencies (enjoyment)

  • ~ .20 for self-reports

  • across different situations people show patterns (correlations), which are traits

Expected correlation sizes (ABCDs)

Typical: .10-.30

Strong in psychology: ~.50

Near 1.0 = unrealistic

Small correlations are meaningful in personality science because behaviour is influenced by many factors

Two strategies for selecting ABCDs

Deductive (theory driven)

Inductive (data driven)

Deductive ABCD strategy

Theory-driven

start with theory (e.g. biological models), group behaviours based on assumed causes

Strengths and weaknesses of deductive ABCD strategy

S

• gives a clear explanation (cause) of personality traits

• provides a structured way to organise ABCDs

W

• theories often lack strong evidence

• personality traits are complex and influenced by many caues

Inductive ABCD strategy

Data-driven

Start with many candidate ABCDs, use data (correlations) to find patterns

What is the drop or combine procedure?

A method of simplifying many ABCDs

Dropping highly similar items (redundant)

Combining moderately correlated items into aggregates

What do you gain in drop-or-combine?

Reduces complexity

Increases reliability

Reveals broader traits

What do you lose in drop-or-combine?

Lose specific detail

Blurs unique aspects of individual ABCDs

May oversimplify personality

What is factor analysis (PCA)?

A statistical method that identifies patterns of correlations among variables (ABCDs)

Groups them into fewer underlying factors

Main steps of factor analysis

Collect many ABCD measurements

Compute correlations between them

Extract factors (find underlying dimensions explaining shared variance)

Rotate factors (makes structure easier to interpret)

Interpret factors (label traits)

Outputs of factor analysis

Factors/components (e.g. extraversion)

Eigen values - how much variance each factor explains

Factor loadings (how strongly items relate to factors)

What is typically meant by items in personality trait science?

Adjectives that characterise how people typically tend to feel

Descriptive statements about tendencies to think, feel, act, or behave in a particular way

What is the lexical hypothesis?

States that important individual differences in personality become encoded in language

• if something matters people develop words to describe it

How is the lexical hypothesis used?

Extract personality-descriptive words from dictionaries

Measure them in people (ratings)

Use factor analysis / drop-or-combine

Identify broad trait dimensions

Why is a taxonomy of traits useful?

(like the big five)

Organises hundreds of ABCDs into a manageable system

Provides a common language across studies

Allows generalisation of findings (efficiency)

Enables prediction of life outcomes (health, work, relationships)

Serves as a framework across psychology and other fields

What are the big five?

The big five are five broad personality trait domains

• Openness

• Conscientiousness

• Extraversion

• Agreeableness

• Neuroticism

They are aggregates of many ABCDs

adjectives of emotional stability / neuroticism

nervous, self-critical, dependent

adjectives of extraversion

merry, talkative, carefree

adjectives of agreeableness

trustful, generous, cooperative

adjectives of conscientiousness

persistent, orderly, dependable

adjectives of intellect/openness to experience

insightful, creative, perceptive

Why are informant reports important?

Provide independent validation beyond self-reports

Reduce bias (self-perception vs reputation)

Increase scientific confidence

what is the robustness of the big 5 across informant questionnaires and the self

when two informants rate someone they both know well, their ratings agree with each other to nearly the same degree

when you rate yourself high on extraversion, it is likely that in about 70-75% of cases that your friend and relatives also rate you high on this domain

what is the robustness of self informant personality traits

correlation between two measurements are above .80

Why are cross-cultural studies useful?

Test whether traits are universal vs culture-specific

Check robustness across languages and societies

Cross cultural studies findings (big five)

Many lexical studies in languages other than English do not reveal the big five, without guidance

Lexical studies sometimes vary across languages

Overall: broad support, but not perfect universality

What is the HEXACO model?

Six-factor personality model

• Honesty-Humility

• Emotionality

• Extraversion

• Agreeableness

• Conscientiousness

• Openness

What is implicit personality theory?

refers to peoples beliefs about which traits go together

• people infer traits even with little information

• can produce big five-like patterns even for strangers

Why is implicit personality theory important?

Helps explain why trait structures are robust

but may reflect bias + real observation combined

What are the two meanings of traits?

Descriptive - traits summarise patterns of ABCDs

Causal/explanator - traits are underlying mechanisms (glue)

both describe patterns and explain them

What is the duality principle?

Traits have two roles simultaneously:

• summary of behaviour (ABCD patterns)

• Underlying cause of behaviour

What is the personality trait hierarchy?

Domains

Facets

Nuances

What evidence supports something being a trait?

A valid trait should show:

• stability over time

• cross-rater agreement

• consistency across situations

• heritability (genetic influence)

• predictive validity (life outcomes)

What is the bandwidth-fidelity dilemma?

Trade-off between:

• High bandwidth (broad traits)

  • Capture more behaviour

  • less precise

• High fidelity

  • more precise

  • less generalisable

What are broad traits better for?

prediction

generalisation

What are narrow traits better for

precision

specific explanations

How do traits provide an integrative framework?

Traits allow us to

• map diverse psychological variables into one system

• compare findings across fields

• organise behaviours, attitudes, and outcomes

instead of studying 100 behaviours, study their relation to extraversion

Why do trait findings generalise across psychology?

Traits are broad aggregates of many ABCDs

They capture shared variance across behaviours

Many psychological phenomena are linked to multiple ABCDs

Therefore

• findings at trait level apply widely across behaviours, outcomes, and contexts

steps to create a personality test

define a domain

write plenty of items for each facet

create instructions and rating scales

collect as much data as possible

retest after about two weeks

also collect informant ratings to validate scales

what is CFA

confirmatory factor analysis

• computer evaluates model with exactly factors we want, defined by out selected items

• fit such a model to the data and ask how well it fits data

• gives factor loadings

What is the ideal factor structure loadings?

A good personality scale should show a simple, clean factor structure where:

• each item loads highly on its intended factor (facet/domain)

• loads low on all other factors

This means

• clear separation between traits

• minimal cross-loadings

Items should strongly reflect one facet only

What is reliability?

reliability is the consistency or dependability of measurement

i.e., whether we would obtain the same score upon repeated measurement under similar conditions

what is the typical retest reliability for the big five

big five > .80

facets ~ .70 to .85

items ~ .50 to .80, typically above .65

What is internal consistency?

The extent to which items in a scale correlate with each other

based on average inter-item correlation, and number of items

more correlated items = less random error

limitations of internal consistency

underestimates reliability

boring for test takers as many questions need to be the same

prevents sampling nuances broadly

What is the principle of aggregation?

Aggregating multiple items cancels out random error, and increases reliability

if a test has reliability .80, what does this mean for a persons score

80% reflects true trait level

20% is measurement error

what is the standard error of means (SEM)

estimates how much a score typically carries due to error

tells you precision of the score

SD x square root of (1 - reliability)

what does SEM mean in practice (e.g. 4.5)

a score may vary by about ± 4-5 points

Why avoid overly similar items?

Inflates internal consistency artificially

Reduces content validity (poor coverage)

Creates redundancy (same nuance repeated)

How to improve reliability?

Increase number of items (aggregation)

Improve item clarity

Use retest measurement

Combine multiple raters

Avoid ambiguity

What is validity?

What the test actually measures

correctness

Types of validity

Content validity

Face validity

Construct validity

Convergent validity

Discriminant validity

Content validity

Did we cover the whole trait?

Whether your scale includes all important parts (facets/ABCDs) of a trait

Face validity

Does it look like it measures the trait?

intuitive, surface-level meaning of items

high face validity is good in personality science

Construct validity

Does the test behave like the trait should?

Supported by convergent and discriminant validity

Convergent validity

Does it correlate with similar things?

High correlation = good

Discriminant validity

Does it NOT correlate with different things?

Low correlation = good

construct validity of the big five scales

convergent correlations between ~ .50 and .70, reliability > .80

discriminant correlations between ~ 0 and .50

what would criterion validity be

relationships with meaningful outcomes/correlates of the construct

should be able to predict future and explain current

what is consensual validity

agreement with other raters of the same construct

a personality judgement is more likely to be accurate when multiple independent observers agree

two things to remember when interpreting scores

direction arbitrary

• low neuroticism, high emotional stability = the same thing

single score has no meaning

• needs to be compared to something

How to improve validity?

Broadly sample ABCDs (content)

Use multiple methods (self + informant)

Validate against outcomes

Avoid bias in items

What are the units of trait variation?

Trait variation is measured using:

• Raw scores (sum/average of items)

• Standard scores:

  • z-scores (subtract mean, divide by SD) (mean = 0, SD = 1)

  • t-scores (mean = 50, SD = 10)

  • IQ-type scores (mean = 100, SD = 15)

Standard scores show how far someone is from the average.

z scores mean and SD

mean = 0

SD = 1

t scores mean and SD

mean = 50

SD = 10

IQ-type scores mean and SD

mean = 100

SD = 15

How are personality trait scores typically distributed?

Trait scores are usually normally distributed (bell shaped)

• most people are near the average, fewer people at extremes

How should a personality trait score be interpreted meaningfully?

Must be interpreted relative to a norm group - score has no meaning on its own

• convert to standard scores (z or T), interpret as distance from mean

What are the main biases in self-ratings?

Social desirability (adjust answers to appear more socially acceptable)

Acquiescence (agreeing with items regardless of content)

Extreme responding (using the ends of the scale a lot (e.g. strongly agree)

Reference group effect (people rate themselves in comparison to peers)

How can we reduce the main biases in self-ratings?

Use balanced items (positive and negative)

Write neutral items

Statistically control bias

Use informant reports

what is the purpose of lie scales and what bias does it reduce

special items to detect unrealistically positive responding

include statements most people would not honestly endure (I have never told a lie)

if someone agrees with many of these it suggests social desirability bias/faking good

what are response profile comparisons

comparing a persons response to a socially desirable (ideal) profile

high similarity suggests social desirability bias

How much of personality scale variance is true vs bias vs error

~ 40-55% true (shared across raters)
~ 35-40% method/rater bias

remaining random error

personality scores are true trait + bias + error

What are the advantages of self-report scales?

cheap and efficient

access to internal states (thoughts, feelings)

easy to interpret

What are the disadvantages of self-report scales?

Susceptible to bias

Limited self-awareness

Method-specific variance

Why is it useful to combine self- and informant reports ?

Reduces random error (aggregation)

Cancels some biases

Improves validity

Allows triangulation of the true unit

to separate real trait stability from bias and measurement error

What are the most common personality scales?

NEO-PI-R

Big Five Inventory

HEXACO

What is IPIP

Open source pool of 1000+ items

Used to build many personality scales

Free alternative to proprietary tests

how do we estimate real stability

divide observed cross-time agreement by cross rater agreement

what are alternative methods to personality tests

experimental tasks

implicit measures (IAT)

digital trace data

experience sampling

what is an example of experimental task

marshmallow test

• delayed gratification - walter michael

• people who delayed taking do better in life (apparently)

hand in ice water for ego strength - roy baumeister

go NoGo task for measuring self control/impulsivity

these are more objective but low validity, limited scope

what is implicit association test

if a trait is part of peoples self concept, they quickly connect with it

more automatic associations

low reliability, unclear meaning

• .60 for IAT

what is digital trace data

tracing personality from online digital data

• facebook likes, words used in posts, credit card records

advantages

• objective, non-invasive measurement, large data sets available

disadvantages

• ethics, trait coverage, face validity

based on 70 likes a computer can know you as well as a friend, based on 300 liked can know you as well as a partner

what is experience sampling/momentary assessment/ambulatory assessment?

measuring peoples thoughts, feelings, and behaviours in real time, repeatedly, in their everyday life

• participants get prompts on their phone, several times a day, over days or weeks

mor accurate, aggregation improves reliability, better representation of real life

burdensome, expensive/time-consuming, missing data, still self-report

what is time-series data

measure someone many times over time