Psychometrics and Psychological Measurement

Operational definitions

Procedures used to measure hypothetical constructs in psychological research.

Hypothetical constructs / latent variables

Theoretical concepts used to explain differences in people's behavior, such as intelligence or self-esteem.

Psychological test (Cronbach)

A systematic procedure for comparing the behavior of two or more people, involving behavioral samples.

Criterion-referenced tests

Tests that sort people into groups based on whether their performance exceeds a specific expected decision criterion.

Norm-referenced tests

Tests used to understand how an individual's score compares to a broader representative reference sample.

Speeded tests

Time-limited tests scored by counting the number of questions answered in the allotted period.

Power tests

Non-time-limited tests scored by counting correct answers, with items ranging in difficulty.

Reflective / effect indicators

Observable variables seen as being caused by an underlying latent variable (e.g., intelligence test scores).

Formative / causal indicators

Observable variables that collectively contribute to an underlying latent variable rather than being caused by it.

Francis Galton

Considered the founding father of modern psychometrics; studied individual differences in 1880.

Participant reactivity

When knowing they are being measured influences a participant's psychological process.

Demand characteristics

Cues that let participants know what the researcher expects to find, influencing their behavior.

Social desirability

Behavior change in participants attempting to impress an observer or scorer.

Score sensitivity

A measure's ability to discriminate between meaningful amounts of the dimension being measured.

Property of identity

The most fundamental level of measurement where categories are mutually exclusive and exhaustive.

Property of order

Values of a sample arranged in ascending rank order relative to each other.

Property of quantity

Provides information about the magnitude of differences between people using continuous real numbers.

Absolute zero

A state in which an attribute of an object or event has no existence (e.g., zero length).

Arbitrary zero

A starting point chosen by convention or convenience, such as temperature scales.

Nominal scale

Measures qualitative data by sorting variables into unordered, labeled categories based on identity.

Ordinal scale

Produces ranks in which people are ordered according to the amount of an attribute they possess.

Interval scale

Data where values are ordered, distances between values are equal, and the zero point is arbitrary.

Ratio scale

A scale with ordered values, equal intervals, and an absolute, true zero point allowing multiplicative interpretations.

Interindividual variability

Differences that exist between different people.

Intraindividual variability

Differences that emerge within a single person over time or different circumstances.

Variance

A statistical measure of variability calculated as the mean of squared deviation scores from the mean.

Standard deviation

The square root of the variance, reflecting variability in the size of raw deviation scores.

Positively skewed distribution

A distribution with relatively few values above the mean, having a skew value greater than zero.

Negatively skewed distribution

A distribution with relatively few values below the mean, having a skew value less than zero.

Covariance

A quantitative measure of the connection between variability in one set of scores and variability in another.

Pearson correlation coefficient

An easily interpretable index of linear association bounded between -1 and +1.

Validity

The degree to which evidence and theory support the interpretations of test scores for proposed uses.

Construct validity

The essential validity concept referring to the degree test scores reflect a particular psychological construct.

Karl Pearson

Created Pearson's correlation coefficient, the Chi-Square test, and coined the term 'standard deviation'.

Charles Spearman

Developed factor analysis and made important contributions to correlation and intelligence testing.

Lee Cronbach

Developed generalizability theory and significantly expanded classical test theory in the 1970s.

Observer/scorer bias

When a researcher brings their own biases and expectations into the measurement process.

Conjoint measurement

The quantitative assessment of qualitative attributes based on numerical values or utilities assigned to different attributes.

Variance-covariance matrix

A square matrix presenting variables' variances on the diagonal and covariances between pairs of variables in the off-diagonal cells.

Differential psychology

The study of individual differences, often seen to include intelligence, aptitude, and personality.

Five facets of construct validity

Internal structure, associations with other variables, response process, consequences of use, and test content.

load() function (R)

Opens an R-formatted data set.

names() function (R)

Prints the names of all variables in the data set.

head() function (R)

Prints the first few rows of data.

str() function (R)

Reveals key info about each variable, including the type and meaning of numerical values.

attr() function (R)

Tells us the meaning of each numerical value of a variable to help interpret data.

table() function (R)

Provides frequency counts for each level of a variable.

factor() function (R)

Reclassifies a variable to ensure categories are treated as a special type without quantitative properties.

Central tendency

The score that is most representative of the entire distribution of scores.

Scatterplot

A visual representation of the association between two variables using points plotted at coordinate intersections.

Sum of squares

The sum of squared deviation scores about the mean, used as the numerator when computing variance.

Cross-products

The product of an individual's two deviation scores, used to compute covariance.

Mean (X-bar)

Represents the typical score in a distribution, calculated by summing all values and dividing by N.