Exam 2 Flashcards
Correlated Groups Design
Repeated measures: Uses the same people
Matched Designs: use matched participants
Single subject Design types
ABA Reversal
Multiple Baselines
Single-subject randomized, time-series
Advantages to Correlated Group Design
Each person/pair is their own control
Saves time; Fewer participants needed
Disadvantages to correlated group design
matching can be hard
Sequence effects
Sequence effects reduced by…
Randomization of conditions full counterbalancing partial counterbalancing (latin squares) hold variable constant (pre-training) rest period
How are single subject designs different from case studies?
A variable of interest is manipulated by the experimenter
measures changes in the individual/pair
Ex: Treatment of phobias
Strengths of single subject design
internal validity is strong (confounds are controlled)
Weaknesses of single subject design
external validity weak
only 1 individual so hard to generalize
ABA reversal
Single subject design
A: baseline–not reading text
B: manipulation–reading text
Multiple baseline
Single subject design
IV: teacher attention to problem student
DV: disruptive behavior, reading scores, math scores
Single Subject-randomized-time-series
Behavior is monitored at each point, but what time the treatment is introduced is selective randomly
Assumptions of an ANOVA
Populations are normally distributed
Populations have equal variances
ANOVA’s compared to T-test
you use an ANOVA when there are more that 2 group means to compare
ANOVA compared to Regression
Regression is a more general form of ANOVA
ANOVA Null hypothesis
H0: U1 =U2=U3
ANOVA Research hypothesis
At least two population means differ (largest from the smallest)
Are there directional ANOVA hypotheses?
NO, can only tell us whether there is a significant difference between the largest and smallest mean
Planned Multiple comparison tests (a priori)
Bonferroni, Fisher’s LSD
Unplanned Multiple comparison tests (post hoc)
Tukey’s, Scheffe
Bonferroni, Fisher’s LSD (least significant difference)
(a prior)i idea about which means differ, easier to get sig. results
Tukey’s, Scheffe
harder to get sig because not planned; alpha cut by however many comparisons you’re doing
R^2
Proportion of variance explained
R^2 sizes
Small=.01
Medium=.06
Large=.14
