chapter 8 Flashcards
What is a bivariate correlation? What are its conditions?
An association that involves exactly 2 variables.
The study itself can have more than 2 variables – there will be 2 measured variables in the data set.
How can we visualize the association between 2 quantitative variables?
Through a scatterplot and a data set.
What are the possible associations we can see between 2 quantitative variables?
Positive association ; negative association ; null association.
True or False : I can visualize categorical variables through scatterplots.
False : Scatterplots may be more complicated and not the best option for categorical variables ; they are not as precise in displaying the data. Best to use rectangle graphs.
What makes a study correlational?
Having 2 or more measured variables, but NO manipulated variables.
What is the use of an effect size?
An effect size describes the strength of an association.
True or False : A larger effect size is more important/significant.
True : Small effect sizes can compound over many observations. However, even though it’s small, it might still be meaningful over time.
What question is asked when looking at construct validity in an association claim?
How well was each variable measured? Is it measuring what it’s intended to measure? Does the measure have good reliability?
What question is asked when looking at statistical validity in an association claim?
How well do the data support the conclusion? How strong is the relationship? How precise is the estimate (i.e : how wide is the confidence interval)? Has it been replicated?
What question is asked when looking at internal validity in an association claim?
Can we make a causal inference from an association?
What question is asked when looking at external validity in an association claim?
To whom can the association be generalized?
True or False : If the confidence interval includes 0, it is likely that the association is statistically significant.
False : If it includes 0, it is NOT statistically significant.
What is an outlier?
An outlier is an extreme score distorting the association. It can make a correlation appear much stronger than it is in reality.
What can happen if there is a restriction of range?
If there is a restriction of range, there could be a correlation, but not particularly strong.
In relation to internal validity, how can we apply the 3 causal criteria (i.e : which questions to ask for each of them)?
Covariance : Do the results show that the variables are correlated?
Temporal precedence : Does the method establish which variable came first in time?
Internal validity (third variable problem) : Is there a C variable that is associated with both A and B, independantly?
