PSY 3100 Ch. 10: Experiments

correlational studies

all variables are measured (poor int. validity)

experiments

at least one variable is manipulated (high int. validity)

independent variable

what we think causes differences in another v

dependent variable

what we think is caused by something else

treatment group

condition designed to change lvl of IV

control group

condition designed to NOT change lvl of IV

constant

part of a study that is the same for all participants (to conclude IV causes DV: every participant should experience the exact same thing BUT manipulation)

between-subjects

 every participant is exposed to ONE level of IV

within-subjects

every participant is exposed to EVERY level of IV

carry-over effect

effect of one level of IV “carries over” into other conditions.

practice effect

performance improves

fatigue effect

performance worsens (boredness)

counterbalancing

 presenting diff sequences of conditions to diff participants. Controls for order effects by equating them across levels of the IV.

demand characteristics

any aspect of experiment that indicates nature of experiment (can cause reactivity)

evaluating construct validity

consider whether manipulation is manipulating the construct

evaluating statistical validity

is there an effect? (statistical significance) How large? (effect size) how precise? (sample size/error bars)

P-value

tells us how surprising the data is, assuming there is no effect. Probability of getting the observed or more extreme data, assuming the null hypothesis is true. *

Cohen’s D

 gives difference between groups/conditions in standard deviation units (Z-scores) (.2 .5, .8)

meta-analysis

statistical analysis of several studies to examine if there is overall evidence of an effect

evaluating external validity

who is the population we want to generalize to? Who was sampled? Was it representative?

evaluating internal validity

can we conclude one v. causes another?

placebo group

neutral group that is made to appear similar to treatment group w/o changing level of IV

limits to adding levels to an independent variable

can’t distinguish among diff possibilities; might not reveal full effects of IV. Fleshes out direction of influence