Psychometric Theory & Factor Analysis Flashcards
1
Q
TRAIT IDENTIFICATION/MEASUREMENT
A
- can be used in combo:
LEXICAL APPROACH - all relevant dimensions of personality exist in natural language
THEORETICAL APPROACH - start w/theory which guides selection of terms/question formation
PSYCHOMETRIC THEORY - statistical approach; use factor analysis to guide selection of terms/questions
2
Q
THE ESSENTIAL TRAIT APPROACH
A
- attempts to reduce large trait numbers to a few traits essential to understanding personality
- goal = find smallest trait number by which individual difs in personality can be adequately described
- underpinned by psychometric theory/factor analysis
3
Q
PSYCHOMETRIC THEORY
A
- psychometrics refers to theory/methods of psychological measurement
- psychometrics includes IQ testing/measurement of personality traits/vocational testing
- impact of individual testing and in wider society = substantial so it is important to have high professional standards for the development/administration/tests of interpretation
4
Q
FACTOR ANALYSIS
A
- lies at psychometrics heart
- statistical method that allows a lot of data to be reduced to few important factors
- developed by Spearman at beginning of 20th century for the analysis of psychometric data in abilities field
- statistical technique used by Eysench/Cattell; eventually results in Big 5
- many disagreements between dif theorists = result of difs in factor analysis use; important to have understanding of technique
- cornerstone of factor analysis = correlation concept
5
Q
FACTOR ANALYSIS: BRIEF SUMMARY PROCESS I
A
- measure large number of people in various ways using numerous items/qs
- correlate scores on each measure w/scores on every other measure (correlation matrix)
- determine how many factors (traits) need to be hypothesised to account for various clusters of inter-correlations
- when items show high correlation w/one another they are thought to measure same ability/characteristic (factor/trait)
6
Q
FACTOR ANALYSIS: CONCEPTUAL MODEL
A
- FA uses correlations among many items to identify factors
7
Q
FACTOR ANALYSIS: BRIEF SUMMARY PROCESS II
A
- subjectively decide the meaning of each factor/label it
- label should reflect meaning of items that cluster together to make the factor
- standardise personality measure; test measure on hundreds of people representative of population you’d want to measure
- analyse responses/develop norms; all future scores assessed against these norms
8
Q
EYSENCK’S HIERARCHICAL (PEN) MODEL OF PERSONALITY
A
- Eysenck personality can be described using 3 dimensions:
1. His theory identifies 3 second-order factors (super-traits) derived from inter-correlations between source traits.
2. Super-traits are broad traits that subsume large surface number/narrow traits.
3. Eysenck’s personality model based on 3 super-traits; often referred to by acronym PEN.
9
Q
PEN
A
PSYCHOTICISM
- tendency towards psychotic/sociopathic beh
EXTRAVERSION VS INTROVERSION
- high extroversion = low cortisol arousal levels
NEUROTICISIM VS EMOTIONAL STABILITY
- high neuroticism levels = excessive activity of autonomous nervous system
10
Q
HIERARCHICAL STRUCTURE OF EXTRAVERSION-INTROVERSION (E)
A
LEVEL 1
- super-trait
LEVEL 2
- narrow/subordinate traits (ie. sociable/lively/active/assertive/dominant/carefree/surgent/venturesome/sensation-seeking)
11
Q
PERSONALITY ASSESSMENT: RELIABILITY
A
- all tests should have norms BUT still doesn’t guarantee a reliable/valid test
- before using any standardised test we need to examine reliability/validity
- test = good reliability w/consistent measuring
- reliability = 2 distinct meanings:
1. TEST-RETEST RELAIBILITY - stability over time/repeatability
2. INTERNAL CONSISTENCY - whether all items are measuring same thing
12
Q
TEST RE-TEST
A
- test-retest reliability = correlates scores from large pps set who take test on at least 2 occasions
- .8 correlation = minimum for test-retest
13
Q
TEST-RETEST: AFFECTING FACTORS
A
- subjects characteristics (ie. ill/tired/upset)
- characteristics in test (ie. poor test instruction/complex responses/subjective scoring/guessing)
14
Q
TEST-RETEST: DISTORTING FACTORS
A
- time gap (at least 3 months)
- difficulty levels of items
- subject sampling (representative of the relevant population)
- samples size (at least 100)
15
Q
INTERNAL CONSISTENCY
A
- internal consistency = all items must measure same thing/capture breadth of concept
- best internal consistency index = Cronbach’s Alpha
- reliability = NEVER below .7
- very high alphas = “bloated specifics” (.9+); test too narrow/specific to be valid
- low alphas = items compromising scale probably measuring dif concepts
- very high/low internal consistency reduces test validity; either items = too specific to capture concept breath/items so diverse they aren’t measuring anything coherent
- valid test = ^ internal consistency
