Chapter 7: Bivariate Correlational Research Flashcards
What are bivariate associations? Variable(s) are they measured or manipulated?
research involving two variables and they are measured not manipulated…because it would be hard to manipulate them
What is an association claim?
describes the relationship found between two measured variables. They can be positive, curvilinear, negative or zero.
To investigate associations researchers need to measure what?
- the first and second variable in the SAME group
Correlational studies that investigate more than two variables can or cannot be considered bivariate?
can be…they can examine the relationship between two variables at a time….
What are the ways associations between two variables can be described? (2)
- scatterplots and the correlation coefficient r
Categorical variable?
values fall in either one category or another
How are correlational data described when both variables are quantitative?
scatterplot
When one variable in the study is categorical what kind of graph will researchers use?
- bar grap
L> shows the avg of each category
What do bar graphs allow us to examine ?
- allow us to examine the difference between the average scores to sees whether there is a difference.
When a difference is noted on bar graph for when one of the two variables is categorical what does this mean?
- that there is an association! ….
What statistics can be used when at least one of the variables is categorical?(3)
- sometimes r
- t test
- point-biserial correlation
What is a t test used for when at least one variable is categorical?
- whether the difference between the means is statistically significant
What is a point-biserial correlation used for when at least one variable is categorical?
- correlational coefficient that is similar to r but it is especially intended for evaluating the association between one categorical variable and one quantitative variable.
What is the phi coefficient used for?
- designed to evaluate the association between two categorical variables!
Are association claims characterized by a particular statistic or graph?
NO
- its characterized by a study design in which BOTH of the VARIABLES are MEASURED
What are the two most important validities for interrogating an association claim?
- construct validity
- statistical validity
Aside from the two most important validities used for interrogation of an association claim what others can be looked at?
- External validity
Is internal validity relevant to association claims?
NO
Construct validity??
L>q’s?
- ask about this for BOTH variables…….how well each were measured aka the operationalizations of the variables
L> does it have good reliability? Is it measuring what it is suppose to?
L> evidence for face validity( is the measure an appropriate operationalization of the conceptual variable) , concurrent validity (is the measure correlating with a simultaneously occurring outcome it should be related to) ,discriminant validity ( extent a measure does not associate with measures of a measure that is theoretically different and it should not be related to) and convergent validity( extent a measure is associated with otter measures that are theoretically similar in constructs).
Statistical validity???
l> what should you consider?
asking what factors might have affected the scatterplot, correlation coefficient (r), bar graph or difference score that led to your association claim
L> considering effect size and its statistical sig, outliers, is a zero relationship possibly curvilinear?
Effect size?
describes the strength of an association
What are cohen’s values for identifying effect sizes?
- 1 (or -0.1) = weak/small
- 3( or -0.3) = medium or moderate
- 5 (or -0.5) = large or stronge
What does a strong effect size mean?
- that we can make more accurate predictions of one variable from another