Paired Samples T-Test
Used for within-group comparisons. (Comparing 1 sample across 2 separate time periods)
Dependent Group Designs
More control over extraneous variables.
Creates a more powerful test (more likely to have a significant result)
Fewer overall participants needed
Counterbalancing
The order of events is rotated across participants to prevent carryover effects
Fatigue Effects
Performance decreases over time
Practice Effects
Performance increases over time
Demand Characteristics
Clues that alert the participant to guess the nature of the experiment
Paired sample and independent sample t-tests are similar except…
The SE is calculated differently. So is df
df = n-1
Conditional differences formula…
dfB = (number of conditions) -1
Differences due to individuals and same people formula…
dfE = (n-1)(number of conditions -1)
2-Way Factorial Design
2 independent variables. APA format is 2(stimuli) x 2(strategy) factorial design
3-Way Factorial Design
APA format is 2(stimuli) x 2(strategy) x 2(gender) factorial design
Factorial Design
A design that consists of 2 or more IVs with all possible combos of them. Produce main effects of IVs and interaction between IVs
Interactions
When the effect of an IV depends on the presence of a second IV (intersecting lines)
Correlational Design
Does NOT include any IVs. Has at least 2 DVs. Used to identify potential relationships between 2 variables
With correlational data remember…
CORRELATION DOES NOT IMPLY CAUSATION!!!
Positive Correlation
Increases in X associated with increases in Y
Decreases in X associated with decreases in Y
Negative Correlation
Increases in X associated with decreases in Y
Decreases in X associated with increases in Y
Pearson’s R
A measure of effect size. Ranges from -1 to 1. Formula is df = n-2
Random Assignment
The experimenter randomly assigns participants to the experimental or control group (helps to determine causation)
Quasi Experiments
Comparing group differences without random assignment (hard to determine causation)
Ignoring the baseline means what?
Scaling in data can be misleading because…
