indiv dif 1b Flashcards
latent variables
variables that cannot be observed. e..g psychological variables
measured via psychometrics
how do psychmetrics measure variables that cannot be observed
infer/estimate based on behaviour and self reported experience
what do correlations tell us ab psychometric data
enables us to determine which terms are related to each other. And so represent a common ‘latent factor’
Factor analysis
when datasets are too big for correlations.
A multivariate data reduction technique
instead of showing how each items correlates with every other item,
Creates latent variables. E.g. labels them factor 1 and Factor 2. Shows how each variable correlates with the underlying factor variable
usually. one group would correlate highly with one factor, other group with other factor.
we would look at the similarities and develop a name for the factor = ‘factor loadings’.
Researchers need to decide how many factors to extract.
what are factors advising this
For independent factors, items should load strongly on only one factor (wiht low cross loadings)
e.g. toffee, potato, pasta flapjack
separating into carbs incorporates all
separating into starchy carbs and sweet carbs, flapjack may correlate with both ( = cross loading) and so it would be better to sort into just carbs
how did Raymond Cattel reduce the 4500 words on personality into 171 terms
what did he then do with the 171 words
- group synonyms
- pairs of antonyms (scale)
- selected exemplar for each (to name the scale)
He then got 100 people to rate one/two people on all 171 terms .
This data was used to conduct a form of factor analysis. He examined patterns of correlations between pairs of words, identifying clusters of terms. Generated 60 clusters.
He found terms from psychiatric literature.
He reduced to about 45 terms based on own judgement and guidance from the literature.
he then perfromed factor analysis to identify 16 ‘source traits’ / personality factors
negatively critiqued on
- subjectivity and arbitrariness in analysis
- 16 = too many
- Are correlations between factors (not independent)
- Better accounted for by fewer, independent factors (e.g. an analysis on same data found extraversion, anxiety, self-control-Independence/accommodation & openness )
what were Cattel’s 16 factors rewarded for
- methodological, data driven,
- promotes factor analysis use
- revised 16PF still used
- Formed the basis of the big 5 models
what is Costa and McCrae’s big 5 model of personality
5 super traits: (NEOAC)
Neuroticism
Extraversion
Openness to experience
Agreeableness
Conscientiousness
each super trait has it’s own 6 facets (primary traits)
was there any overlap between the big 5 perosnality model and eyesencks PEN model
They both contained neuroticism and Extraversion scales
Which correlated with eachother
But none of the other traits correlated with each other
what were Jack Block’s problems with the big 5 model
- Factors derived from factor analysis depend on which variables are entered AND on the number of factors selected
- Single words insufficient to describe all aspects of personality (sentences and paragraphs needed) e.g. pecking order personality (dismissive to a rank below etc) ALSO Single words out of context can mean diff things. e.g. aggressive
- Overreliance on lay persons (especially undergrads to collect data) result in important omissions (of descriptive words. e.g. capricious, censorious)
- Simple correlations do not capture full complexity of relationships. e.g. intelligence cor witty. But the correlations are not symmetrical (som1 who is intelligent isn’t necessarily witty). Not captured in factor analysis.
- Arbitrary choice of 6 factors
- Some important factors not represented e.g. narcissism
Evidence for OCEAN model
- Cross cultural validity
- Emerging evidence for biological basis
- Most widely used SO a common currency for personality research
