Week 1-2 Psychometrics Flashcards
Psychometric soundness
the reliability and validity of a test
Reliability
Consistency of measurement, Precision with which the test measures and the extent to which error is present in the measurement
Validity
The ability of a test to measure what it is intended to measure
Cronbach’s definition of psychological test (3 criteria)
- Test involves behavioural samples
- The behavioural samples are collected in a systematic way
- The purpose of the test is to compare the behaviour of two or more people
Scaling
how numerical values are assigned to psychological attributes
3 key properties of numbers
The property of identity
The property of order
Th property of quantity
Norm-referenced tests
Norm referenced tests are used to compare a person’s test score with scores from a representative reference sample
e.g intelligence tests
Criterion referenced tests
Criterion referenced tests are used to evaluate an
individual’s test score with reference to a set standard
mastery tasks
Criterion referenced tests are typically used to gauge
achievement or mastery, so they are sometimes called
mastery tasks
inter-individual differences
comparing the behaviour of different people
intra-individual differences
comparing the behaviour of the same people at different times in different contexts
4 factors that can negatively affect the
interpretation of test scores as valid
- demand characteristics
- social desirability effects
- malingering
- experimenter bias
The property of identity
Represents “sameness” vs. “differentness” by sorting people into categories based on similarity of psychological features
The property of identity: 3 rules
1 must satisfy the property of “identity”
2 the categories must be mutually exclusive
3 the categories must be exhaustive
The property of order
- represents information about the relative amount of an attribute people possess
- When numerals have this property, they indicate the rank order of individuals on some psychological attribute
The property of quantity
-conveys information about the magnitude of differences between individuals
-Real numbers can be used to represent the quantity of an attribute
-cases can be compared with each other in a
meaningful and informative way
absolute zero
Absence of an attribute
arbitrary zero
does not indicate the absence of an attribute
Additivity
Unit size should remain constant—all units being counted should be equal
Intelligence, aptitude, and personality test scores are ordinal in nature because
they indicate not the amount of intelligence, aptitude,
and personality traits, but rather the rank order of individuals
In psychological measurement is additivty satisfied?
rarely, for example, in tests each question does not have equal difficulty
Counting is a __________ but not _____________ condition for measurement
necessary, sufficient
Is counting construed as measurement?
Simply counting various things is not necessarily construed as measurement.
The total sum needs to be construed to represent some sort of attribute (e.g., knowledge or competence)
Steven’s (1946) four scales of measurement
Nominal
Ordinal
Interval
Ratio
nominal
-categorising based on one or more attribute into mutually exclusive and exhaustive categories
Ordinal
- allows classification
- rank ordering is possible
- does not indicate how greater one ranking is to another
- numbers do not indicate units of measurement
- no absolute zero point
interval
- scales contain equal intervals between numbers
- averaging possible
- no absolute zero
Ratio
- has absolute zero point
- all mathematical operations can be performed
Psychologists prefer to think of their data as interval because
it gives them more flexibility to manipulate and analyse the data and use more powerful operations such as ANOVA which can only be used on interval or ratio data
what levels of measurement can you compute the average for?
interval- or ratio-level
but not if they are
ordinal- or nominal-level
Correlation coefficients are bounded within a range of
-1 and +1
Two distributions can have the same _______ but very different dispersion of _______
Mean, Scores.
Three meaningful ways to describe distributions
1 central tendency
2 variability
3 shape
variability
The degree of variance in scores in a single distribution
Various ways we can represent variability in a distribution of scores
range, interquartile and semi-interquartile ranges, average deviation
variance and standard deviation reflect
variability, as the degree to which scores in a distribution deviate (differ) from the mean of the distribution.
Variance
the mean of the squared deviations between scores on a distribution and their mean.
Calculating variance
- Subtract each score from the mean
- Square every deviation
- Sum the squared deviations and divide by the total number of scores in the distribution.
The SD is equal to
To the squared root of the “average of the squared deviations about the mean”,
hence the SD is the square root of the variance.
SD reflects variability in terms of the ____ ______________ _________, whereas variance reflects variability in terms of _______ _________ ______.
raw deviation scores, squared deviation scores.
Covariability
The degree to which variability in scores in one distribution co vary with scores in another distribution
simple frequency distribution
A table in which scores are listed according to the frequency they occur.
Grouped frequency distribution
A table in which test scores are grouped into intervals/bins known as class intervals e.g 70-78
Central tendency
Tells us about the typical response e.g median, mode or mean
shape
e.g positively skewed or negatively skewed
Approximately what percentage of scores occur between the mean and +1 standard deviation?
34%
the variance of a composite variable is determined by:
1 the variance of each item within the composite
2 the correlations among the items
Approximately what percentage of scores occur between the mean and -1 standard deviation?
34%
A standard score is a
raw score that has been converted from one scale to another scale
Approximately what percentage of scores occur between ±1 standard deviations
68%
The scale used for computing T scores ranges from __ standard deviations below the mean to __ standard deviations above the mean
5, 5
Approximately what percentage of scores occur between ±2 standard deviations
95%
The convariance only provides information about:
a) the magnitude
b) the direction
b) the direction
Calculate percentile rank with access to the entire distribution of scores
An individual’s percentile rank is the number of scores in the distribution that are lower than the individual’s score divided by the total number of scores times 100
The size of the covariance is influenced by the strength of the association between
the metrics of the two variables, such that between two variables that produce large scores (“large-scale” variables) will tend to be larger than a covariance that involves one or more variables that produce small scores (“small-scale” variables
The correlation coefficient provides a clear representation of both
the direction and the magnitude of an association between two variables