Psychology Research Methods: Variables, Designs, and Statistics

Mills' criteria (covariance)

two variables move together in some predictable way (they co-vary)

Mills' criteria (temporal precedence)

the independent variable must come first in time before the outcome variable

Mills' criteria (ruling out alternatives)

researchers must rule out alternative explanations so a third variable is not the reason why the dependent variable is changing

Selection effect

unequal groups to begin with because of differences in participant variables; avoided with random assignment and selection

Maturation effect

changes due to time spent in the experiment (duration or time of day) that differ between groups; happens inside the lab

Testing effect

changes due to measuring the outcome variable more than once such as practice or expectations; avoided by giving both groups the same learning and test experience

Mortality effect (attrition)

differences between groups because some people drop out of the study; asymmetrical attrition is a confounding variable

History effect

changes due to something in the external world (e.g., holidays, disasters) that affects one group more than the other; happens outside the lab

Confounding variable

a variable that affects one group differently than another and provides an alternative explanation for changes in the dependent variable

Extraneous variable

a variable not related to the treatment that influences both groups equally and acts like noise or error

Between-groups design

each participant experiences only one level of the independent variable and different groups are compared

Within-groups design

each participant experiences all levels of the independent variable and acts as their own control

Posttest-only design

participants are exposed to one level of the independent variable and measured only once after the experiment

Pretest-posttest design

participants are measured before and after the independent variable to compare baseline and post-treatment scores

Counterbalancing

randomly varying the order of levels of the independent variable across participants to avoid order effects as a confound

How to decide: between-groups

use when participating in one condition makes the other impossible, when there is a lasting effect of the independent variable, or when participants should not know the research design

How to decide: within-groups

use when you want to reduce individual differences and compare each participant to their own performance to detect small changes

Single-factor design

an experimental design with one independent variable that has multiple levels and one dependent variable

Multifactor (factorial) design

an experimental design with more than one predictor or independent variable that examines their separate and combined effects on the outcome

Main effect

the overall effect of one predictor variable on the outcome variable, ignoring the other predictor(s)

Interaction effect

when the effect of one predictor variable on the outcome depends on the level of another predictor variable; lines are not parallel

Additive effect

when the effects of predictor variables add or subtract in a linear way and the effect of each does not depend on the level of the other; lines look roughly parallel

No effect (null result)

when the independent variable does not affect the dependent variable or differences are obscured by ineffective manipulation or too much within-group variability

Statistical validity (general)

how well the numbers support the claim, including strength of effect, precision of estimates, and replication

Evaluating statistical validity (real relationship)

asking whether researchers have identified a real relationship and how strong the relationship is

Evaluating statistical validity (inferences)

asking whether appropriate inferences or conclusions were made from the data

Descriptive statistics

statistics used to summarize and describe data such as frequencies, measures of central tendency, and dispersion

Inferential statistics

statistics used to make inferences from a sample to a population and to test hypotheses (implied by using the descriptive vs. inferential distinction)

Measures of central tendency

statistics that describe a representative outcome for the sample (mean, median, mode)

Mean

the average score, used when data are normally distributed

Median

the middle score, used when the distribution is skewed or has outliers

Mode

the most common score in a distribution

Standard deviation

how spread out scores are from the mean and how far they are on average from the mean

Measures of dispersion

statistics that describe how spread out scores are, such as standard deviation and the spread of the distribution

Outliers

scores that are far from most other scores and can make a distribution skewed

Skewed distribution

a distribution where data have outliers on one end so the tails are not symmetrical and mean, median, and mode are separated

Standardized scores

general term for scores that have been converted to a common scale (like z-scores) to compare across different scales

z-score

a standardized score where 0 is the mean, positive values are above the mean, and negative values are below the mean, expressed in standard deviations

Converting to z-scores

subtract the mean from the raw score and then divide by the standard deviation to get how many standard deviations the raw score is from the mean