Correlation Research pt 1
The best graph format to examine an association claim between two quantitative variables.
What is a scatterplot?
Internal validity, covariance, temporal precedence
What are the requirements for causality?
Pre-existing group differences
What are selection effects?
What are the 3 potential threats to internal validity?
Can be just as interesting as experiments that show group differences
What are null effects?
r = .50
The conclusion a researcher reaches regarding the likelihood (probability) of getting the calculated correlation just by chance, assuming there’s no correlation in the real world.
What is statistical significance?
Another way to say within-subjects design or dependent measures design or repeated measures design.
What is a within-groups design?
The only way or at least the best way to investigate a causal claim.
What is an experimental design?
A third factor that could additionally inflate variability within groups.
What is situation noise?
Implications for when a small effect might still be important.
What are life-or-death implications?
Tests if the difference between the means of two groups is meaningful.
What is a t-statistic or visual inspection of a bar graph?
A confound that qualifies as an alternative explanation for the results of a study.
What is design confound?
All potential 3rd variables that researchers keep constant in experiments
What is a control variable?
The likelihood that a study will yield a statistically significant result in our sample when the IV has an effect in real life.
What is power?
These are hard to detect with a straight line model.
What are curvilinear relationships?
A correlation that involves exactly two variables that can be either categorical or quantitative.
What is bivariate?
Participants are exposed to all the levels of an independent variable at roughly the same time
What is a Concurrent-Measures Design?
Eliminate alternative explanations (confounds) by keeping all variables (except the independent and the dependent variable) constant.
What is a well-designed experiment?
Measurement error, individual differences, and situation noise
What are the three sources of high within-group variability?
1.What is the effect size? How strong is the association?
2.Is the correlation or the mean differences statistically significant?
3.Are there any outliers affecting the findings?
4.Is there a range restriction?
5.Is the association linear or curvilinear?
What are the 5 questions to ask when investigating aspects of the statistical validity of an association?
p reflects the likelihood or the degree to which the null hypothesis (Ho) is true
What is a common misconception about the p-value?
Occur when exposure to one level of the independent variable influences responses to the next level.
What is an order effect?
Gives researchers more power to notice differences between conditions
What is an advantage of a within-groups design?
Weak manipulations, insensitive measures, ceiling and floor effects, design confounds
Why we might not have enough between-groups difference?