Psych Exam 2: Week 6 powerpoint notes (Factorial Research Designs)

Example 1: - IV: Gender and IQ - DV: Conversational style (# of interruptions) - Design: 2 (gender: men, women) x 2 (IQ: high,low) - MM: Marginal Means Low IQ High IQ MM Men: 8.0 8.0 8 Women: 5.0 5.0 5 MM: 6.5 6.5 NA

MM for gender: > Men: 8.0+8.0= 16.0/2= 8 >Women: 5.0+5.0= 10/2= 5 MM for IQ: > Low: 8+5= 13/2= 6.5 High: 8+5= 13/2= 6.5 Results: * There is a significant main effect of gender such that, across levels of IQ, men (M = 8) interrupted more than women (M = 5) * There is NOT a significant main effect of IQ. People with high IQ (M = 6.5) did not differ from people with low IQ (M = 6.5) in how often they interrupted *There is NOT a significant interaction effect

Example 2: - IV: Low and High vitamins - DV: mg protein - Design: 2 (vitamin level: low, high) x 2 (mg of protein: 100,300) - MM: Marginal Means 100mg protein 300mg protein MM Low V: 0.80 1.5 1.15 High V: 1.10 2.3 1.7 MM: 0.95 1.9 NA

MM for vitamin level: > Low: 0.80+1.5= 2.3/2= 1.15 >High: 1.10+2.3= 3.4/2= 1.70 MM for mg of protein: > 100mg: 0.80+1.10= 1.9/2= 0.95 >300mg: 1.5+2.3= 3.8/2= 1.9 * Results: - There is a main effect of vitamin intake such that, regardless of protein intake, rats who receive more vitamins (M = 1.7) weighed more than rats who received fewer vitamins (M = 1.15) - There is a main effect of protein such that, regardless of vitamin intake, rats who received large doses of protein (M = 1.9) weighed more than rats who received smaller doses (M = .95) - There is a spreading interaction. The increase in weight due to protein is greater among rats who received high vitamins than among rats who received low vitamins.

Factorial Design: Information obtained

* 1) Main effect of each IV - Overall effect of IV across all levels of other IVs *2) Interactions effect between IVs - Interactions are qualifications - These tell us that an effect depends upon a certain condition

Factorial Design Practice

* 2 x 4 factorial design: = 2 variables: 2 levels and 4 levels = 2*4--> 8 cells > 2(IV1 : a, b) x 4(IV2 : a, b, c, d) * 4 x 2 x 3 > How many variables?: 3 (4,2,3) > How many levels of each variable?: 4,2, and 3 levels > How many cells?: 4*2*3= 24 cells *2 (Gender: male, female) x 3 (Mood: happy, sad, angry) > How many variables? What are they?: 2- Gender and Moof > How many levels of each variable?: 2 and 3 levels? > How many cells?: 2*3= 6 cells

Factorial design

* Contains two or more independent variables that are completely crossed - Every level of every IV appears in combination with every level of every other IV - Specifies a) how many IV exist in the design and b) how many levels of each IV exist in the design - allows researchers to examine the boundary conditions of a research finding - Chess example: Originally researchers were interested in assessing expertise and memory. They thought the more expertise, the better memory. * Notation indicates # of VARIABLES and # of LEVELS of each variable

Interaction effect or moderation

* Exists when the effect of one IV depends on the level of a second IV - Relation between IV and DV is not constant across second IV *Tested with simple effects - These are like main effects, but within one level of a second IV - look at everything but the marginal means resutls/ the numbers you are adding up to get the marginal means - Think of it as a qualification - We are comparing the means of the conditions themselves when looking at the interaction effect - The effect exists in some conditions but not in other condition - Effect depends on one level of the 2nd independent variable?

Factorial Designs: Partioning

* In essence, as long as you have factors that are completely crossed, you can partition the data in meaningful ways *Partitioning the variance is key! - Virtually all of the things we have been doing are based on the notion of partitioning variance into systematic variance vs. error variance - This is the crucial component of parametric statistics

One-way design experiment

* Only has one independent variable - Can have two or more levels - Simplest possible experimental design= 2 groups design - Control and experimental group - Allows us to test a lot of social psychological stuff w/ simplicity - Limitation: They allow researchers to look at only one IV at a time

One-way, Multiple-groups design

* Only one single IV IV takes on 3 or more levels

interactions

* Spreading/ordinal - Effect exists at one level of a second IV, but is weaker or nonexistent at a different level - Lines are diverging > Pink line would be one condition and that there is an effect bc it is increasing after time; real position type? > they don't need to cross *Crossover/disordinal - Effect is opposite at different levels of the 2nd IV - No main effects of either variables - Lines cross in the middle - No main effects, but both simple effects are significant > Green line starts low but increases. Pattern is reverse with Pink bc it starts higher on the left but decreases. When Pink starts high, green starts low and when pink ends low the green line ends high > Lines will always be nonparallel but don't have to necessarily touch to be spreading?

Main effects

- The simple/straight-forward effects of IVs in factorial studies , averaging across all levels of the other IV(s) * The overall effect of one IV - Averages across levels of other IVs * One main effect per IV *Add and divide by how many there are *To test main effects look at marginal means