Bivariate, Multivariate, and Experimental Research Methods

Comprehensive practice flashcards reviewing bivariate correlations, multivariate regression and longitudinal designs, and simple experimental methodologies.

What is a multivariate design in psychological research?

A study designed to test an association involving more than two measured variables, conducted to help researchers get closer to evaluating causal criteria.

What is a longitudinal design, and which causal criterion does it specifically help address?

A study in which the same variables are measured in the same individuals across multiple points in time; it specifically helps establish temporal precedence.

In a longitudinal study, what is a cross-sectional correlation?

A correlation between two variables that are measured at the exact same point in time.

In a longitudinal study, what is an autocorrelation?

The correlation of a single variable with itself measured at two different points in time, showing how consistent that variable is over time.

In a longitudinal design, what is a cross-lag correlation?

A correlation between an earlier measure of one variable and a later measure of another variable, which directly helps establish temporal precedence.

In the longitudinal study examining violent TV watching and aggression from 3rd grade to 12th grade, what did the cross-lag correlations show?

Watching violent TV in 3rd grade was significantly correlated with aggression in 12th grade ($r = .31^*$), whereas aggression in 3rd grade was not significantly correlated with violent TV watching in 12th grade ($r = .01$).

In the longitudinal study investigating sports activities and friend count, which causal explanation was ruled out by the cross-lag correlations?

The hypothesis that spending more time early on at sports causes individuals to have more friends later was ruled out ($r = .11, \text{n.s.}$), whereas having more friends in 3rd grade was significantly associated with higher sports participation in 12th grade ($r = .41^*$).

What is multiple regression analysis, and which causal criterion does it help address?

A statistical technique computing the relationship between two variables while statistically controlling for one or more additional variables; it helps address internal validity by ruling out measured third variables.

In multiple regression, what is the distinction between a criterion variable and a predictor variable?

The criterion variable (or dependent variable, $y$) is the primary outcome researchers seek to understand or predict, while predictor variables (or independent variables, $x$) are used to explain variance in the criterion variable.

What does the phrase 'controlling for' or 'adjusting for' mean in a multiple-regression analysis?

Holding a potential third variable at a constant level statistically while assessing the unique association between two other variables.

How is beta ($\beta$) interpreted in a multiple-regression model?

It indicates the slope and direction of the relationship between a predictor and the criterion variable while holding all other predictors in the model constant.

Why does multiple regression fail to definitively establish causation?

Researchers can only control for third variables that they measured; they cannot control for unmeasured confounds, nor does multiple regression necessarily establish temporal precedence.

What is a mediator (mediating variable) in psychological research?

An intermediate variable that provides the causal mechanism explaining why two variables are related, specifying the sequence where $A$ leads to $C$, which in turn leads to $B$.

What is a moderator (moderating variable)?

A variable that, depending on its level, changes the strength or direction of the relationship between two other variables.

What is the key conceptual difference between a mediator, a moderator, and a third-variable problem?

A mediator explains the underlying mechanism of why two variables relate; a moderator identifies for whom or when the relationship changes; and a third variable is an extraneous confound that creates a spurious relationship between two unrelated variables.

What defines a bivariate correlational study?

An investigation that tests the statistical association between exactly two measured variables.

What statistical tests are used to evaluate an association when one variable is categorical and the other is quantitative?

A $t$ test is used if the categorical variable has two levels; an ANOVA is used if the categorical variable has more than two levels.

What statistical test is used to assess the association between two categorical variables?

A chi-square ($\chi^2$) test.

What are the three essential rules for establishing causality?

1. Covariance (the variables must be significantly associated). 2. Temporal precedence (the causal variable must precede the outcome). 3. Internal validity (there must be no alternative plausible explanations or confounds).

What is the directionality problem in correlational research?

An ambiguity in temporal precedence that occurs because both variables are measured at approximately the same time, making it unclear whether variable $A$ caused variable $B$ or variable $B$ caused variable $A$.

What is a spurious correlation?

An apparent statistical relationship between two variables that is not causally linked but driven entirely by an unmeasured third variable or coincidence.

How do outliers impact the correlation coefficient ($r$), and how does sample size influence this effect?

Outliers can disproportionately distort the strength and direction of $r$; their distorting influence is substantially stronger in small samples than in large samples.

What does a 95% confidence interval for $r$ that contains zero indicate about statistical significance?

It indicates that the correlation is not statistically significant ($p \ge 0.05$) because zero is a plausible true population value.

If bivariate data produce the scatterplot shown in the figure, how will the Pearson correlation coefficient ($r$) evaluate the true strength of the relationship?

The correlation coefficient ($r$) will underestimate the relationship because $r$ describes the slope of a straight line and cannot properly capture curvilinear relationships.

Why is data visualization essential when evaluating bivariate relationships with identical correlation coefficients and summary statistics?

Different datasets can share identical means, standard deviations, and correlation coefficients ($r$) while exhibiting completely different curvilinear or patterned distributions.

How can moderators inform the external validity of an association claim?

When an association is moderated by an outside variable (such as residential mobility), it demonstrates that the finding does not generalize equally across all subgroups or contexts.

What is the primary advantage of experimental designs over correlational designs?

Experiments manipulate an independent variable to establish temporal precedence and control extraneous variables to establish internal validity, thereby supporting causal claims.

What is the distinction between an independent variable and a dependent variable in an experiment?

The independent variable is the variable manipulated by the experimenter (consisting of two or more levels), whereas the dependent variable is the measured outcome variable.

What is a control variable in an experiment?

A potential third variable that a researcher purposefully holds constant across all conditions to eliminate alternative explanations and maintain internal validity.

What is a design confound?

An experimenter's error where an extraneous variable happens to vary systematically alongside the independent variable, serving as an alternative explanation for the outcome.

What is the difference between systematic variability and unsystematic variability regarding potential confounds?

Systematic variability occurs when an extraneous factor coincides predictably with specific experimental conditions (creating a confound), whereas unsystematic variability fluctuates randomly across all conditions and does not threaten internal validity.

What is a selection effect in an experiment, and how is it prevented?

A threat to internal validity occurring when participants in one condition systematically differ from participants in another condition; it is prevented through random assignment or matched groups.

What is a matched-groups design, and when is it most advantageous?

A procedure where participants are measured on a relevant variable, sorted into matched subsets based on that variable, and then randomly assigned across conditions; it is typically used with small sample sizes to ensure equality on critical individual differences.

What defines an independent-groups (between-subjects) design versus a within-groups (within-subjects) design?

In an independent-groups design, different participants are assigned to each condition so each person experiences only one level; in a within-groups design, every participant experiences all levels of the independent variable.

What is the difference between a posttest-only design and a pretest/posttest design in independent-groups experiments?

A posttest-only design measures participants on the dependent variable once after exposure to the independent variable, whereas a pretest/posttest design measures the dependent variable twice: once before and once after exposure.

What is a repeated-measures design?

A within-groups experimental design in which participants are measured on the dependent variable multiple times, following sequential exposure to each level of the independent variable.

In the puppy communication study shown in the photograph (Bray et al., 2020), what type of experimental design was utilized, and what were the key variables?

It was a concurrent-measures within-groups design where the independent variable was whether a cup was pointed at or not, and the dependent variable was whether the puppy knocked over the correct cup.

What are order effects in within-groups designs, and what are their three primary forms?

Threats to internal validity occurring when exposure to earlier conditions alters performance in later conditions; the three forms are practice effects, fatigue effects, and carryover effects.

How does counterbalancing prevent order effects in within-groups designs?

By presenting experimental conditions in varied sequences across participants, either through full counterbalancing (all possible sequences) or partial counterbalancing (such as a Latin square or random ordering).

What are demand characteristics?

Cues in an experiment that lead participants to infer the researcher's hypotheses, causing them to alter their behavior accordingly.

Based on the comparison in the diagram, how do sample size requirements differ between an independent-groups design and a within-groups design?

Independent-groups designs require larger overall samples ($N = 36$ in the diagram) because unique participants are needed per condition, whereas within-groups designs require fewer participants ($N = 12$ in the diagram) because each person participates in every condition.

What techniques do researchers use to evaluate the construct validity of an independent variable manipulation?

Manipulation checks (inserting an extra dependent variable to verify that the manipulation worked) and pilot studies (pretesting manipulations on a separate group before the main experiment).

What standardized effect size metric is conventionally computed to quantify the difference between two group means in an experiment?

Cohen's $d$, which represents the difference between condition means scaled in standard deviation units to show group overlap.