Show:
Experimental/Research Design
Best Practices
What Test to Use?
Predictions
Assumptions
100
This sampling technique increases the chances that samples are chosen by chance to be in either the treatment or control condition
Random sampling
100
How large should your sample be?
As large as possible
100
A researcher is interested in examining the relationship between two continuous variables.
Correlation
100
You are interested in the relationship between two variables. You get your results, and prior to running any analyses, find that all the scores are near the top end of the possible range of scores on both variables. Will your correlation be strong or weak?
Weak, due to restriction of range
100
What would you do if Levene's statistic was significant?
Interpret the line in SPSS that notes "equal variances not assumed"
200
This type of research design includes a group who receives some form of treatment or manipulation
Experimental Design
200
This should be reported along with any test statistic and p-value
Measure of effect size
200
This is used to examine group differences on a continuous DV
Independent Samples t-Test
200
You are running an independent t-test and find that the standard deviations between both groups are quite small. Will this increase or decrease error? Will this increase or decrease your t-value?
Decrease error, thus increase t-value
200
What would you do if sphericity was violated?
Use either the Greenhouse-Geisser or Huynd-Feldt Correction
300
This design is used to determine the relationship between two variables
Correlational design
300
True or False: A p value of .054 can be considered "marginally significant" or that it "approached significance"
False
300
A design that has one categorical and one continuous IV, with a continuous DV
ANCOVA
300
After running an ANOVA, a student found no significant differences (F-statistic was not significant). She decides to run another study with a very similar sample, but this time around, will include a covariate. All other things being equal, is she more or less likely to find a significant result when using ANCOVA? Why?
More likely, as ANCOVA has more power (reduces individual differences)
300
What are the three usual ANOVA assumptions?
normality of DV scores, independence of observations, and equality of variances
400
In this design, people in the treatment condition are assigned based on some quality or characteristic
Quasi-experimental design
400
If given the option, should one conduct an independent samples ANOVA or a dependent samples ANOVA? Why?
Dependent samples ANOVA. These have more power because we can model individual differences.
400
This design has two categorical (between subjects) variables and one continuous DV
Factorial ANOVA
400
An uninformed researcher runs multiple t-tests whilst keeping alpha at .05. Is he likely to commit a Type 1 or a Type 2 error?
Type 1
400
When the relationship between the DV and the covariate in an ANCOVA are not equal across groups, what assumption have we violated
Homogeneity of regression slopes
500
What three assumptions need to be met in order to make causal statements?
X precedes Y X and Y covary No third variable (or other explanation) influencing the relationship
500
Which would be the better covariate: one that was strongly correlated with the IV or one that was not correlated with the IV? One that was correlated with the DV or one that was not correlated with the DV?
Not correlated with the IV and correlated with the DV.
500
This design incorporates one within subject IV, one between subjects IV, and one continuous DV
Mixed ANOVA
500
You've run a two-way within subjects ANOVA and find epsilon to be .98. Would you predict that you've violated the assumption of sphericity?
Nah, probably not
500
Why do we examine and/or test assumptions
Violations may result in either increased Type I or Type II error rates