Midterm 2 - Chpt. 11 Flashcards
Simple vs. Complex Experiment
Simple designs great for isolating simple causal relationships
Complex designs have multiple IVs and/or multiple (more than two) levels of an IV
○ Allow us to examine if a relationship is non-linear
○ Better approximation of the real world
○ Multiple IVs allow us to examine interactions
In complex designs, IV referred to as…
a factor
Marginal Means
Averages per condition
Cell Mean
Key Concepts
For each condition (cell) the average is calculated across participants (e.g. total $ spent divided by # of spenders)
Main Effect
Key Concepts
The effect that each IV, by itself, has on the DV (pretend the other IV doesn’t exist!)
In a 2 x 2 design, there are two possible main effects (i.e. one main effect for each IV)
Do the marginal means differ?
Interaction
Key Concepts
An interaction between IVs indicates that the effect of one IV on the DV is different at different levels of the other IV
e.g. the effect of food type on amount spent is different depending on whether people are using cash or credit
ONLY LOOK AT ONE DIRECTION, NOT BOTH
EX: Does the type of food (junk vs. healthy) affect how much money spent
Moderator
Key Concepts
WHEN WE HAVE AN INTERACTION, we call the “other” vairable a moderator (When the effect of one IV on the DV depends on the level of the “other” IV)
- “Other” IV = “Moderator”
To check its effects: are the differences in cell means (per main IV) significant when you compare them to each other? (eg - difference in $ spent on junk/healthy food using cash, versus difference in $ spent on junk/healthy food using credit = COMPARE THE MARGINAL DIFFERENCES)
Benefits of Factorial Designs
- Less participants
- Can tell us whether the effects of one variable are constant across levels of the other variable.
- Is the effect of payment different, for different types of food?
- Closer to reflecting real-world conditions.
Why should we interpret interaction graphs?
- To understand psychological phenomena, we need to examine interactions.
- Learning to visualize data can be tough at first, but once you get it, it makes interpreting your data much easier.
- Easier to read journal articles.
What condition must be met in order to calculate average cell means?
When cells have EQUAL #’s of P’s in each condition
Key Questions (for bar graphs)
For factor 1: Are the average areas of the different colored bars different?
□ Main effect of A
- X axis condition
For factor 2: Are the average areas of the similarly colored bars different?
□ Main effect of B
- Y axis condition
Is there an interaction?
□ If the lines are parallel, then NO (slopes of lines are parallel, even in non-identical heights)
□ If lines are NOT parallel, then YES
Key Questions (for line graphs)
Are the midpoints (average) of the two lines different on the DV?
□ Main effect of B
- Y axis
Are the average values of the DV at each level of the IV on the X-axis different?
□ Main effect of A
- X axis
- Comparing AVG of each endpoint between the two lines
Is there an interaction?
□ If the lines are parallel, then ”No”
REMEMBER - measuring/comparing MARGINAL means
READINGS
Why would researchers want more than 2 levels of an IV?
- Interested in comparing more than two groups (additional controls)
- May not provide enough information about the relationship between the IV and DV
- Can clarify if a relationship is truly linear, or something like a non-linear positive/negative relationship
Variables are sometimes related in a…
- Curvilinear fashion: directions of relationship changes across values of the variable
- AN EXPERIMENT WITH ONLY TWO CONDITIONS CANNOT DETECT THESE RELATONSHIPS; AT LEAST 3 LEVELS MUST BE USED
An experiment with more than one IV:
Factorial Designs: has more than one IV, with each IV also having more than one level
Simplest design: two IV’s, each with two levels
Factorial designs yield two distinct kinds of information..
- Main effect: information about the effect of each IV, taken by itself - the effect that each IV has on the DV by itself
* In a design with two IV’s, there are two main effects - one for each IV - Interaction: the way that one independent variable affects the dependent variable depends on the level of the other variable
Interactions:
Results depend on, or is conditional or contingent on, something else.
* EX: friend asks you to go to the movies
* Likelihood of you going depends on
◊ Any exams coming up
◊ Who stars in the movie
Why are interpretations of interaction important?
- indicates that the main effects must be qualified (i.e. main effects are conditional or contingent on something else)
- Interactions between IVs indicate that the effect of one IV varies at different levels of the other IV
Interactions illuminate…
moderator variables: which influence the relationship between two other variables
* E.g. body type variable is a moderator because it moderates the relationship between other variables
* Can be aspects of the situation, or aspects of the participants
Simple Main Effects
For significant interactions, we must break it down to understand it
…the mean difference at each level of one IV
Results are analysed within each level of the other IV
NOT to be confused with main effects (which takes the avg. across the levels of the other IV)
Variations of 2x2 factorial designs:
IV x PV design
IV x PV design: common type of factorial design that includes both experimental and non-experimental; NOT A TRUE EXPERIMENT - contains a measured, not manipulated, variable; thus, can’t make causal claims
Allows researchers to investigate how different types of people (ie participant variables - age, ethnic group, characteristics - which can’t be controlled/randomly assigned) respond to the same manipulated variable
Design:
- 1 manipulated variable that has two levels
- 1 participant variable with two levels: Different age groups that are low/high on personality traits, short/tall individuals
Variations of 2x2 factorial designs:
Mixed-factorial design
a combination of both
Between subjects:
* In a 2x2, there’s 4 conditions
* If we want a completely between-subjects design, different participants will be assigned to each of the 4 conditions
* Thus, if we were to put 10 participants in each condition, we would need 4 participants total
Within subjects:
* The same people will participate in all conditions
* If you wanted 10 participants per condition, a total of 10 participants would be needed
Mixed factorial design using combined assignment
- Participant variable - psychopathy - is a between-subjects variable, as participants could be both psycopathic and non-psychopathic
- Second IV - story truth - is a within-subjects variable: all participants told a truthful story and a lie
Increasing the complexity of factorial designs - 2 METHODS
- One way to increase experimental complexity - more than 2x2 - is to increase the number of levels of one or more of the Ivs
- Another being to increase the number of Ivs overall
- Expansion of the general format for describing factorial designs:
- Number of levels of first IV x number of levels of second IV x number of levels of third IV….+
