Research Methods: Validity, Measurement, Sampling, and Correlational Research

Comprehensive vocabulary flashcards covering measurement validity, survey and observational designs, sampling techniques, bivariate correlations, and multivariate correlational methods.

Validity

How well a measure actually measures what we want it to measure, or whether it actually measures the intended construct.

Reliability

The consistency and reproducibility of a measure.

Face validity

The extent to which a measure plausibly looks like it measures the target variable.

Content validity

The extent to which a measure captures all important parts and dimensions of the construct.

Criterion validity

How well a measure relates to an important, concrete behavioral or real-world outcome.

Predictive validity

A subtype of criterion validity evaluating whether a measure successfully predicts a future outcome.

Concurrent validity

A subtype of criterion validity evaluating whether a measure relates to a current outcome assessed at the same time.

Known-groups paradigm

A method to test validity by examining whether scores on a measure can distinguish between groups whose behavior or status is already known to differ.

Convergent validity

The degree to which a measure correlates strongly with other measures assessing the same or a closely related construct.

Discriminant validity

The degree to which a measure has a weak correlation with measures of different or distinct constructs that could reasonably be confused with it.

Construct validity

Broad evidence that a measure actually represents and behaves like the theoretical construct it claims to measure.

Open-ended questions

Survey questions where participants generate their own answers in their own words rather than selecting from predetermined options.

Forced-choice questions

Survey questions that require participants to choose their response from a fixed set of provided options.

Likert-type scale

A rating scale in which participants indicate their level of agreement, frequency, satisfaction, or intensity along an ordered set of response options.

Exact-point labeling

A scale design where every numerical response option is given an explicit text label rather than only labeling the two endpoints.

Response set

A tendency for participants to respond to survey items in a patterned way regardless of the specific content of the questions.

Acquiescence

A response set in which a participant has a consistent tendency to agree with almost every statement.

Socially desirable responding

A reporting bias where participants provide answers that make them look favorable or socially acceptable rather than reporting their true attitudes or behaviors.

Implicit measure

A task designed to assess attitudes or automatic associations indirectly without asking the respondent to state them explicitly.

Observer bias

When an observer's expectations, beliefs, or prior knowledge influence how they perceive, interpret, record, or categorize behavior.

Blind observation

An observational design in which observers are kept unaware of participant group assignments or study conditions to prevent expectations from biasing data collection.

Reactivity

A change in participants' behavior caused by the awareness that they are being watched or recorded.

Population

The entire collection of individuals or objects that a researcher is interested in studying.

Sample

A smaller subset selected from the population that actually participates in the research study.

Selection bias

A sampling error that occurs when the selection procedure causes certain members of the population to be overrepresented or underrepresented.

Simple random sampling

A probability sampling technique in which every single member of the population has an equal chance of being selected.

Stratified sampling

A probability sampling method where the population is divided into subgroups (strata) and individuals are randomly sampled from each subgroup.

Proportionate sampling

A sampling method where subgroup sizes in the sample are intentionally chosen to match the exact percentage proportions present in the population.

Random sampling with replacement

A sampling procedure where each selected person or item is placed back into the population pool before the next draw, allowing them to potentially be chosen again.

Random sampling without replacement

A sampling procedure where each selected person or item is kept out of the pool and cannot be chosen again.

Convenience sampling

A nonprobability sampling method in which participants are chosen based on ease of accessibility and availability.

Snowball sampling

A nonprobability sampling method where initial participants recruit or refer additional participants who meet the eligibility criteria.

Pearson's correlation coefficient ($r$)

A statistic that quantifies the direction and strength of the linear relationship between two quantitative variables measured on interval or ratio scales.

Statistical significance ($p < .05$)

The criterion indicating that the probability of observing a result at least as extreme due to random sampling variation alone, assuming the null hypothesis is true, is less than $5\%$.

Outlier

An observation that falls unusually far from the rest of the data points and has the potential to substantially distort the correlation coefficient $r$.

Temporal precedence

A fundamental requirement for establishing causation stating that the presumed cause must occur prior to the observed effect in time.

Third-variable problem

The plausibility that an unmeasured alternative variable explains the statistical association between variable $X$ and variable $Y$.

Moderator

A third variable that changes or modifies the strength or direction of the relationship between two variables, addressing for whom or when a relationship exists.

Multiple regression

A statistical analysis that evaluates the relationships between several predictor variables simultaneously and a single outcome variable.

Coefficient of determination ($R^2$)

The square of Pearson's $r$ ($r^2$), representing the proportion of variance in one variable that is accounted for by another in a linear relationship.

Line of least squares

The best-fitting straight regression line that mathematically minimizes the sum of squared residuals.

Residual

The numerical discrepancy between an actual observed value and the value predicted by the regression line ($\text{Observed } Y - \text{Predicted } Y$).

Longitudinal research

A research design that tracks and measures the same variables or participants across multiple time points to help establish temporal precedence.

Cross-sectional correlation

In longitudinal studies, a correlation assessing the association between two different variables measured at the exact same point in time.

Auto-correlation

In longitudinal designs, the correlation of a single variable with itself across two different points in time.

Cross-lag correlation

In longitudinal designs, a correlation investigating whether an earlier measurement of one variable predicts a later measurement of a different variable.

Beta coefficient ($\beta$)

A standardized statistic in multiple regression describing the unique contribution of a specific predictor to the outcome when all other predictors are statistically controlled.

Control for

To statistically remove or hold constant the influence of a variable in order to observe the remaining relationship between other variables.

Parsimony

The scientific guideline prioritizing the simplest plausible explanation that sufficiently accounts for the accumulated empirical evidence.

Mediator

An intervening variable that explains the mechanism or pathway through which an independent variable affects an outcome variable (addressing how or why).