Ch11: factorial designs Flashcards
What do factorial designs address?
factorial designs address research questions involving multiple IVs that potentially work together
“factors” in a factorial design refer to
independent variables and/or participant variables
simplest for of factorial design?
2x2
how many factors and how many levels are in a 2x2 design? How many conditions?
2 factors with 2 levels each. 4 conditions.
How many factors and levels in a 2x3 design? How many conditions?
2 factors, one with 2 levels and one with 3. 6 conditions
A researcher was interested in the effects of sexual arousal on the
ability to concentrate, and also wondered whether gender and age are important factors. The researcher had participants read passages that were low, medium, or high in sexual arousal content. The participants included both males and females and were divided into three age categories (18-24, 25-35, and 36-50 years). After reading
the passage, participants were asked to perform a proofreading task;
the researcher measured the number of errors detected on the task.
What are the variables of interest? What kind of design?# of conditions?
Variables of interest: gender, age, level of sexual arousal in content
2x3x3 factorial design
18 conditions
Define main effect
The OVERALL effect of each FACTOR (ignoring the other) on the dependent variable.
how many main effects are in a 2x2 factorial design?
2, because there are 2 factors
main effects are qualified by an ____
interaction
Define interactions in factorial designs
The influence of one factor depends upon the level of the second factor.
When looking at a line graph comparing factors, how can you tell whether or not there is an interaction?
If the lines are NOT parallel, there is an interaction. Especially if they cross.
——B1—B2
A1—30–30
A2–20–40
Is there a main effect?
Mean A1=30 Mean A2=30 NO MAIN EFFECT OF A Mean B1=25 Mean B2=35 THERE IS A MAIN EFFECT OF B
in a 3x3 factorial design in which an interaction existed between age and volume level, how would you word answer?
effect of age depends on volume level (volume on x axis)
How to determine interaction in three factor designs
- determine 2-way interactions by making 3 graphs comparing each pair (eg., age x location, location x gender, gender x age)
- determine 3-way interactions by plotting 2 graphs (eg., age x location for men and age x location for women)
How do you know whether or not there is a 3-way interaction?
- if both graphs look different from eachother
- different interaction of A and B at level C
When all factors/IV’s in the design are between subjects, this is called a
between-subjects factorial design.
eg., gender and age
within-subjects factorial designs
all factors/IV’s are within-subjects.
eg., type of music and type of stimulus (words/images)
Between-subjects factorial designs can be useful for avoiding ____ issues
variability
What could you do to see whether who the experimenter is is a factor in the experiment?
make who the experimenter is a variable to see if it’s a factor
How many total participants do you have in a between-subjects design and in a within-subjects design if there are 10 participants per condition (4 conditions)?
In a within subjects design: 10 participants per condition = 10 total participants
In a between-subjects design: 10 participants per condition = 40 total participants
What is a mixed-methods factorial design?
some factors are between-subjects and some are within-subjects.
-adding between-subject factors to reduce variability
In an experimental design where there are 3 factors:
-order of treatment
-experimenter
-music type
Which factors are between subject and which are within?
Between: -order of treatment -experimenter Within: -music type
What kind of factorial design is a pretest-posttest nonequivalent group design?
2x2 factorial design
-Quasi-experimental
T/F: time series design with nonequivalent control group is a factorial design
True. It would be a ? x 2 factorial design.
Quasi-experimental
Why is a pre-test post-test with a nonequivalent control group design considered 2 x 2 factorial?
factor 1: When are they tested? (pretest and posttest)
factor 2: group (treatment group and control group)
What are the null and alternative hypotheses in factorial designs?
For each individual factor: is thee a main effect? Is there an interaction? eg: 2 x 2 factorial design Hypothesis 1: null: no main effect of "A" alternative: main effect of "A" Hypothesis 2: null: no main effect of "B" alternative: main effect of "B" Hypothesis 3: Null: No interaction of A x B alternative: interaction of A x B 3 HYPOTHESES TOTAL FOR 2x2 FACTORIAL DESIGN
What would be the hypotheses for the following eg:
A researcher is interested in whether students in different fields of study experience different levels of stress regarding exams. She is also interested in whether stress levels change as a function of how close exams are. (2x2)
- Hypothesis 1:
null: proximity to exam has no effect on students’ stress level
alternative: proximuty to exam has an effect on students’ stress level - Hypothesis 2:
null: stress level of psychology and business students are the same
alternative: stress levels of psychology and business students are not the same - Hypothesis 3:
null: no interaction between field of study and proximity of exam
alternative: there is an interaction between field of study and proximity to the exam
One-way ANOVA vs factorial ANOVAS
- One-way ANOVA: includes only one factor
- factorial ANOVAs: have multiple effects to be tested statistically (main effects of each factor and interactions)
p for field of study is 0.682
p for proximity to exam is 0.029
p for field of study x proximity is 0.039
Can we reject the null hypotheses?
- not for field of study but yes for main effect of proximity and for interaction, because they are lower than 0.05.
- The main effect of proximity was QUALIFIED by interaction
In an experiment where the null failed to be rejected for gender, was rejected for education level, and was also rejected for an interaction between gender and education level, what does this mean?
- There is no main effect on scores for gender.
- there is a main effect on scores for educational level
- There is an interaction between gender and ed. level, therefore, the influence of gender depends on what level of schooling the participants are at
Is it possible to find an interaction with no main effects? What about main effects with no interactions?
Yes and yes