1.4b, 12.2a Associations on Two Variables

Question 1

Define explanatory and response variables. Define lurking varaibles.

Answer 1

*the explanatory variable explains/causes a change in/predicts the response variable
*additional variables that might cloud/confuse a study

Question 2

What should be used to display 2 categorical variables? for quantitative variables? for 1 categorical and 1 quantitative variable?

Answer 2

*contingency tables
*scatterplots
*side-by-side bar graph, multiple box plots, back to back stem and leaf chart

Question 3

Define contingency tables.

Answer 3

*used for 2 categorical variables
*column = explanatory variable
*row = response variable

Question 4

Define marginal proportions.

Answer 4

the margins of the table (right and bottom) with totals and frequency distributions for each of the variables

Question 5

Define joint proportions. Define conditional proportions. What can be used to display conditional probabilities?

Answer 5

*“and”= multiply
“or” = add
*conditional proportions - refer to a particular row or a particular column and do not use the bottom right value; take into account only a specific subset of the population
-can be displayed by a table or a stacked bar graph; all totals on right margin will equal 1.0

Question 6

Define independent variables.

Answer 6

variables with no association; occurs when the conditional proportions are the same for reach explanatory variable

Question 7

Define scatterplot. List identifiable features. Define the correlation coefficient, r.

Answer 7

*used for two quantitative variables
*x axis = explanatory variable
*y axis = response variable
*identifiable features:
-form or shape - linear, non-linear, very scattered
-outliers, unusual points, gaps
-strength - none, weak, moderate, strong
*if linear, determine if pos or neg association
*strong associations have an r value close to -1 or 1
*weak associations have an r value close to 0
*correlation does not equal causation