Chapter 8 Experimental Design
Confounding variables and internal validity The means by which independent variables are kept constant Confounding variable A variable that fluctuates along with the independent variable Confounding occurs when the effects of 2 IVs are intertwined making it unsure of which IV is responsible for the observed effect Eliminating possible confounds makes for good experimental design Internal validity When the results of an experiment can confidently be attributed to the effect of the IV
Find the confounding variables A researcher is interested in the ability of schizophrenic patients to judge different time durations. It is hypothesized that loud noise will negatively affect judgment. Subjects are tested Quietly in a sound proof room or Noisily in a nurse’s office w/ a stereo playing music at a loud and constant volume Due to scheduling problems, Risk III patients are available only on Mondays. Risk I patients are available only on Thursdays. Also, hearing tests are scheduled for Thursdays, so the sound proof room is only available on Monday.
Basic experiments Posttest-only design Researcher must Obtain 2 equivalent groups Introduce the independent variable Measure the effect of the independent variable on the dependent variable R = random assignment Experimental group R Measure Participants Control group R Measure
Basic experiments Posttest-only design Avoid selection differences where people selected differ in any way Accomplish equivalence by randomly assigning participants If groups were equally selected and there were no confounds, any difference in the DV groups must be attributed to the IV
Basic experiments Pretest-posttest design Researcher must Give pretest before experimental manipulation The larger the sample, the less likely groups will differ before manipulating the independent variable (IV) R = random assignment R Experimental group Pretest Treatment Posttest Participants Posttest Control group Pretest R
Basic experiments Advantages . Disadvantages
Basic experiments Solomon four-group design ½ participants = only posttest ½ participants = pretest and posttest The effect of the pretest can be directly assessed Possible outcomes Results are the same with or without a pretest A difference b/w the experimental and control groups w/ a pretest BUT no difference with no pretest
Assigning participants Two methods Random assignment (independent groups design) Participants are in all conditions (repeated measures design Advantages Needs less participants; Less training of participants Sensitive to finding statistically significant differences Individual differences can be seen and explained Disadvantages Order effects: the order of presenting the treatments affects the dependent variable. Performance on the 2nd task may be better than the 1st
Assigning participants Types of order effects Practice effect Fatigue effect Contrast effect
Should the IV be between- or a within-subjects design? A neuroscientist hypothesizes that damage to the primary visual cortex is more permanent in older animals A sensory psychologist predicts that it is easier to distinguish slightly different shadings of grays under daylight than under fluorescent light A developmental psychologist predicts cultural differences in moral development
Counterbalancing Used in repeated measures designs Complete counterbalancing All possible orders of presentation are included in the experiment Latin square A limited set of orders where Each condition appears at a ordinal position Each condition precedes and follows each condition at one time
Counterbalancing Randomized blocks Each repetition in the basic experiment is called a block of trials To control for order effects the order of presentation can be randomly determined
Matched pairs design Match people on a certain characteristic Rank participants based on scores on the matched variable Members of each pair are randomly assigned to the conditions in the experiment Each treatment is given to one subject in each pair In the two different groups one of which will act as the control, the other as the experimental. Ensures that groups are equivalent Drawbacks Time consuming Needs a large sample pool
Developmental research designs Cross sectional method Persons of different ages are studied at only one point in time Developmental change is not observed but based on comparisons of different cohorts More common and yields immediately useful results Longitudinal method The same group of people is observed at different time points in time as they grow older Over the course of the study people may move, die, drop out… Sequential method Combines both methods First phase is cross sectional then they are studied longitudinally Two or more age cohorts are tested at two or more times.