Second Test Flashcards
Categorical Data
Observations fall in one, and only one, of a set of non-overlapping categories
Continuous Data
observations can take on any value within a range of possible values
some research questions are better address with continuous predictor designs
for instance, when the predictor variable cannot or should not be directly manipulated
Correlation Data
Captures the degree to which two variables “vary together” - covariance
Effect (covariance)
Degree to which they co-vary
Error (covariance)
Degree to which they vary
Regression
A model (i.e., equation) where one variable predicts another (outcome) variable
“Line of best fit” is called a ___________
Regression line
The error between the prediction and the observed data is called ___________
least squares regression
Standardized regression coefficient (β):
When the data have been “standardized” (Z-scored),
than the slope of the equation is equal to the correlation coefficient (r) – one predictor
Standard error of the estimate:
like “standard error” in t-tests, this can be interpreted as a
standard deviation of data points around the regression line (i.e., residual or error variability)
r-squared: (in regression)
captures the degree of variability
accounted for by the regression equation (“percentage of variability explained”)
if p < α: Results are ___________
“significant” – relationship will hold in the population
“Effect” term (ANOVA)
“significant” – relationship will hold in the population
“Error” term (ANOVA)
differences within groups – individual differences/error
Within-subject designs:
Same group experience different conditions
Between-subject designs:
Different groups experience different conditions
Cross-sectional design:
measure attention from a group of kids at different ages
Longitudinal design:
measure attention from the same group of kids as they get older
Advantages of within-subject design (1-4)
1: More “direct”…applies to highly idiosyncratic behaviors
2: Avoids potential confounds inherent in between-subject designs
- differences on DV between groups prior to the experiment confound the results
- avoided in within-subjects designs / same people in every condition
3: More efficient use of research resources
- maximal data collection from each participant
- can be a big advantage when participants are costly/difficult to come by
4: Allows for more powerful statistical tests
- each participant serves as their own “baseline”
- allows statistical test to “account for more variability” (less error variance)
Repeated-measures ANOVA:
Testing difference for multiple means from within-subject design
Between-subjects test (i.e., one-way ANOVA)
includes individual differences in the “error term”
