First Half Flashcards
What is a latent construct?
Unobservable attribute
What is a psychological test
A systematic procedure for comparing behaviour of 2 or more people (cronbachs)
What are Cronbachs 3 components?
Test involves behavioural samples
Behavioural samples must be collected in systematic way
Purpose to compare differences
What is psychometrics?
Science of evaluating the theoretical attributes of tests
What is arbitrary zero?
The relative zero, hypothetical indicator to quantify an attribute.
What is a nominal scale?
No meaningfulnorder
Often frequencies or %
Assigned numerical label
What are Ordinal scales
Groups/categories in a meaningfulnorder
Ranks people but amount of attribute
Frequencies or percentages
Which scales use categorical data?
Nominal and ordinal
What are Interval scale?
No true zero
eg, celcius
What is a ratio scale?
Data with a true zero
Which scales use continuous data?
Interval and ratio
What are interindividual differences?
Between-person differences
What are intraindividual differences?
Within-person differences
What is explicit research?
Direct aim
Directly intend to explore source and meaning of differences
What is implicit research?
Attempt to learn more about individual
How to work out sd?
Squareroot of variance
Where is the mode in a positive skew of distribution?
Below mean and median
Where is mode in negative skew of distribution?
Above mean and median
What is covariance?
a measure of the relationship between two random variables
How does a scatterplot display stong association
Dots closer together
How to work out covariance
Identify deviation from mean multiplied by each persons deviation
What indicates direction and magnitude of association?
Correlation coefficient
What does >0 correlation signify?
Positive associations
What does <0 correlation signify?
Negative associations
Tests for magnitude of correlation?
Pearsons for normal distributed data
Spearmans for non normal distributed
What does a r value between .10 and .30 mean?
Small magnitude correlation (low consistency)
What does r value .30 to .50 mean?
Medium magnitude (some consistency)
What does r value between .50 to 1.0 mean?
Large magnitude (strong consistency)
What is a z score?
Standardised scores
Mean converted to 0 sd to 1
What does a positive z score mean?
Enables to identify degree score is above the mean
What does a negative z score mean
The score is below the mean
What is reliability?
The consistency and replicability of a measure
What is classical test theory?
A measurement theorythat defines conceptual basis of reliability
Outlines procedures for estimating reliability
What is a percentile rank?
Indicates number of people below a specific score
Eg/ 90th percentile = higher than 90%
What is reliability coefficient (R)
It indicates degree of reliability
If the reliability coefficient (R) = 0 is there reliability?
No
If the reliability coefficient (R) = 1 is there reliability?
Yes, observed at true scores perfectly aligned
4 models of reliability?
Parralel model
Tau equivalent model
Essentially tau equivallent
Cogeneric model
What is the cogeneric model?
Least restrictive model of reliability. Same construct different scales
Different degrees of precision and amounts of error
What is the tau equivalent model?
Same construct and same units
Different amounts of error.
What are the 2 parallel forms test?
Alternative forms reliability
Test-retest reliability
What is intraclass correlation coefficient (ICC)?
Used when multiple independent raters/observers
Less than 0.5 poor reliability
Between 0.5-0.7 mod reliability
0.7 + good reliability
What is Internal consistency?
Correlations between different items within a measure
What 2 main factors influence internal consistency?
1) consistency between items
2) length of test
What are the 3 methods of internal consistency
1) split half
2) raw alpha
3) standardised alpha
What is the split half approach?
Split items in half add all scores together compute correlation
What is the spearman brown split half formula
Reliability = 2(correlation value)/1 + correlation valve
What is the raw alpha approach
Evaluates the consistency across all scores
Checks if all questions answered in similar manner
Most important to report cronbachs alpha
What is an acceptable level of cronbachs alpha
0.7
What is the standardised alpha approach
Calculated after converting scores to z scores
What is item discrimination?
Degree to which item differentiates people who score high/low on a teat
What is the discrimination index? (D)
Proportion of high scores that get item correct
High d= high/low scores differ
What are confidence intervals?
Indicates accuracy if point of estimate
Large range = less confident
What is regression?
A way to explore relationship between 2 numerical scores
What is validity?
Degree to which test measures what its supposed to
What are the 5 facets of construct validity?
Test content
Internal structure
Response process
Associations with other variables
Consequences of use
What does test content validity involve?
Face validity and content validity
What is face validity?
Degree to which measure appears related to specific content
Does the test make sense to participants
What is content validity?
Degree to which content reflects the construct.
What is internal structure?
For test to be valid actual structure should align with theoretical structure
What is dimensionality?
The number of dimensions included in a test
What does unidimensional mean?
All items relate to one dimension
What does multidimensional with correlated dimensions mean?
Different dimensions are correlated
What does multidimensional with uncorrelated dimensions mean?
Dimensions are separate so average score needs to be collected for each dimension
What is factor analysis?
set of statistical procedures designed to determine the number of distinct unobservable constructs needed to account for the pattern of correlations among a set of measures
What are the 2 main types of factor analysis?
Exploratory factor analysis and confirmatory factor analysis
What is exploratory factor analysis?
There is no assumptions about how many factors in data
Data indicates how many factors present
What is confirmatory factor analaysis
We assume certain items relate to different factors then try to confirm it
What are factor loadings?
Indicates which item associated with which factor
Ranges from -1 to 1
Higher than 0.4 is acceptable
What is associative validity?
Does a measures actual association match those it should have with others?
What is convergent validity?
Degree to which test scores correlate with measures of related constructs
What is discriminant validity?
Degree to which scores are uncorrelated with measures of unrelated constructs
What is criterion validity
Test scores being related to particularly important criterion variables
What are 2 types of criterion validity
Concurrent (cross sectional)
Predictive (longitudinal)
What is discriminative and convergent evidence?
Assesses if measures shows correct pattern of association with other measures
What are focused associations?
A method to eval convergent/discrim validity
What is a multitrait-multimethod matrix? (Mtmmm)
A stats method to eval convergent and discrim validity
Involved obtaining scores for several (connected) traits using multi methods
Eg/ parent scales teacher scales etc
What is trait variance?
Whether traits are actually associated with one another
What is (Shared) method variance?
Do correlations between traits only exist when scored in same way?
