Factor analysis 1 Flashcards
What is differential psychology?
- seeks to understand the psychological dimensions that apply for everyone yet allow for or create differences between individuals
What are the two areas of ID?
- STRUCTURAL MODEL (how?)
- PROCESS MODEL (WHY, WHERE, WHEN)
Why are psychological tests important?
- allow for INDIRECT assessment of hidden/latent psychological attributes
what are some problems with sampling?
- representativeness
- biasedness
- response/non-response/dropout/volunteers
- instability
What are some assumptions and limitations of psychological measurement?
- it cannot measure the whole person
- it assesses a single psychological element
- personality exists and is real (concepts vs. constructs)
- personality can be measured in similar BUT not identical ways to physical properties
- personality possesses a relative stability, or predictive variability
What is a scale?
- a set of scores on a test (e.g. IQ scales)
What is a nominal (categorical) scale?
- an attribute is nominated an arbitrary numerical value e.g. dichotomous scales (yes/no), polytomous scales (I mainly use internet for school 1, i mainly use internet for information 2)
What is an ordinal scale?
- things are ranked differently based on rank order, and a numerical value is assigned to each rank
- distances between ranks are meaningless e.g 1st, 2nd, 3rd
- can be balanced (neutral scale pt in middle) or unbalanced (no neutral pt or pt not in middle)
What are some common types of ordinal scales?
- social- distance scales (I would willingly admit members of the following races)
- rating scales, more of a qu (how important is attractiveness 1- imp, 2- slightly)
- Likert scales - 5 pt balanced scales that assess degrees of agreement
- Likert- type scales: anything other than 5 points
What transform or standardise?
- often the raw scores are not sufficient for response comparisons
- results measuring the same attribute from different scales
- generation of population norms (normative scores)
- standardisation of scales
What is the use of population norms?
- allows for a percentile ranking in a group
- records population attributes
- compare an individual’s attributes to them
What are some cautions for using norms?
- sampling and sample size limitations
- sample type limitations
- distribution assumptions
- relative temporal instability.
What is content validity?
- the degree to which scores represent the content area (study domain) they are supposed to represent e.g. an exam
- scores should cover the whole or be unbiased
What are some issues with content validity?
- sampling bias, cluster bias
- systematic error
- ceiling/floor effects
- expert judges
What is criterion - related validity?
- degree to which a test correlates with one or more outcomes or parallel critera
TWO TYPES: Concurrent (present e.g. longer vs shorter) and predictive validity
What is construct validity?
- how well the operationalisation accurately reflects its construct
What is systematic error?
- biases that occur regularly
What is convergence validity?
- high levels of correlation between items that make up the same construct or related constructs
What is discriminant validity?
- low levels of correlation between two items that make up unrelated constructs e.g. neuroticism vs. psychoticism
What is external validity?
- consistency across samples, populations, settings, time-periods
e. g. ecological validity: degree to which a score reflects reality
What is internal validity?
- the degree of confidence on the nature of asymmetric (causal) relations between the measured constructs e.g. naturalistic (questionnaire designs) tend to have good EV but bad IV, experimental designs have good IV but bad EV
What did Norman Anderson say about validity?
- any validity type can be conceptually represented on a 2d space defined by the generality dimensions and focus of concern
What are generality dimensions?
- Internal (validity within a given sample)
- External (validity in other settings)
What are the focuses of concern?
- Outcome
- process
What is the distinct groups approach?
- assess construct validity by differentiating between the levels of a construct in dissimilar samples/groups e.g. levels of personality disorders btw clinically diagnosed and normals
What is the Multi trait/multi-method validation
- using multiple construct indicators via multiple methods ( can ask others to do this as well) e.g. to test tolerance: religious, racial, sexual assessment
What is reliability?
- the degree of consistency/stability of measurement scores across contexts or time
- absence of measurement fluctuations unaccounted by the measurements’ scope
- degree of RANDOM ERROR in observed variability of measurement scores
What is classical test theory?
- ppl have a true score T
- measurements and ppl have errors E (errors assumed to be random)
- observed score (X)
X= T + E
in variance2 = variances 2 T + variances 2 E
reliability index (r) = variance T/Variance x
What are some issues with CTT?
- true scores have: temporal score instability & chance fluctuations
- randomness of (psychological) error
- systematic error (biasedness & validity)
What is additivity?
- adding together items to get a more compound index (assuming that things are completely independent from one another)
- THINGS INTERACT
- problem of multiplicity
What are some assumptions of reliability estimates?
- additivity (problem of multiplicity)
- independence of item scores
- uncorrelated errors (errors from neuroticism need to be completely uncorrelated to psychoticism)
- coding (scoring) consistency
- random subject assignment
- equal dimensionality of scale-itmes: we want items of the same instrument to be measures on same scale.
What are some sources of measurement error on an individuals level? (Q-analysis)
- idiosyncratic (language, mood, fatigue, memory)
- generic (faking- desirability, flood/ceiling styles, acquiescence- agree with everything vs. nay-saying- disagree w everything; midpt/extreme response styles- floor/ceiling; random responses)
What are source of measurement error in items (r-analysis)?
- content-related(lack of clarity; leading/biased questions)
- format-related (range in content domain; no. of response categories e.g. yes/no)
- administration-related (learning/training; distracting settings; admin variations)
What is internal consistency reliability?
- the degree of consistency (homogeneity) in responses to scale items that measure the same construct
What is Cronbach’s alpha?
- (weighted) average scale-item intercorrelation
- takes values between 0 and 1 (0 being only error, 1 being perfect score)
- cut off criteria: btw 0.6-0.8
- the addition of relevant items in the measurement increase the value of the alpha
What is split-half reliability?
- estimated reliability based on the correlation of 2 equal parts of a measurement
What is test-retest (temporal) reliability?
- assesses the stability of test-scores over time
- normatively, same test is administered to the same ppl in atleast 2 different times
What is parallel forms reliability?
- similar to split half reliability but measures two equivalent forms of measurement.
What are some problems with test-retest reliability?
- dropouts
- optimal time interval
- temporal instability of constructs
What is inter-rater reliability?
- estimated reliability based on the correlation btw 2 or more independent judges’ ratings of an item or scale
(Cohen’s Kappa coefficient); similar to test-retest reliability
What is standard error of measurement?
- an index of the average random error of an individual’s or measurement’s observed score
- can be used to calculate the range of scores within which the true observed score lies (i.e. CONFIDENCE INTERVAL)