--allows testing of causal hypotheses with confidence without randomization

--use of the same subjects in all conditions guarantees equivalence at the start of the study

--also called "repeated measures" or "within subjects" research design

-all subjects exposed to experimental conditions—each subject serves as his/her control

1. each subject is exposed to all levels of the independent variable
2. scores on each condition are correlated with scores in the other conditions
3. each subject is measured under each experimental condition on dependent variable
4. critical comparison is the difference between correlated groups on dependent variable

Condition 1 Condition 2 Condition 3

Group A Measure Measure Measure

**may be confounding because experience under one condition may affect results under other conditions—called sequencing effects

counterbalancing—order of presentation to subjects is systematically varied—ideally, would like to have all possible orders of conditions occur an equal number of times

Analysis uses repeated measures ANOVA [example using SPSS in book]

Advantages of Within-Subjects Design

--no differences between groups due to subject variables

--more sensitive than between-subjects designs to the effects of the independent variable because eliminates variance due to subject differences, thus reducing error variance

--fewer subjects are needed

--instructions need only to be given once because only one group

Disadvantages

--each subject is exposed to each condition (potential for sequencing effects)

--practice effects—subject gets familiar with task or may become fatigued

--carry-over effects—one condition affects score in subsequent condition

Controls for Sequencing

1. train all participants up to same level
2. vary the order of presentation

1. each subject exposed to only one level of the independent variable
2. each subject has a matched subject in each of the conditions so that the groups are correlated
3. only one measurement per subject on the dependent variable is used in analyzing the results, but the analysis also takes into account which subjects were matched with which other subjects
4. critical comparison is the difference between the correlated groups

subjects are matched on relevant variables—those that might affect dependent variable

how do you know which variable are relevant? Look at previous studies with high correlations

Advantages

--same as within subjects except no problem of practice or carryover effects

Disadvantages

--requires extra work—must decide which variables to match on and obtain measures of these variables

--matching process is tedious with more than one matching variable

--many subjects may be eliminated because suitable matches cannot be found

--same subject appears in all conditions and there is only one person in study

--usually a time-series study where repeated measurements are taken over time

factorial designs—allow more than one independent variable

interaction—when two or more variables combine in such a way as to affect dependent variable

factors—the independent variables

design notation—(e.g., 2 x 2) denotes how many variables and how many levels of each variable 2 x 3 x 2 = 3 variables, 2 levels of A, 3 levels of B, 2 levels of C

main effects—impact of each independent variable on dependent variable

interaction effects—effect of any combination of two or more independent variables

leads to multiple hypotheses (tested with ANOVA)

1. there is no difference between the levels of factor A (no main effect A)
2. there is no difference between the levels of factor B (no main effect B)
3. there is no significant interaction of factors A and B

because factorial designs more complex, more potential confounding

--need to randomly assign subjects to each of the matrix conditions

--always begin interpretation with interaction (if present)

1. Within-subjects (repeated measures) factorial

--each subject is measured under each of the conditions

--sequencing effects must still be considered

2. Mixed Designs

--a factorial that includes a between-subjects and a within subjects factor

--a factorial that includes a manipulated factor and nonmanipulated factor

**research designs using nonmanipulated factors are not experiments—they represent differential research

**cannot infer causality from nonmanipulated independent variables

--used the same as ANOVA

--effect of a theoretically unimportant but nonetheless powerful variable are removed from the dependent measure scores

--have more than one dependent variable