Deck 11: Moderation,mediation and More Regression

Question 1

A doctor conducted research into the influence of consumption of coffee (measured in cups consumed) and social media (measured in minutes daily spent on social media) on students' academic performance (measured by an end of term test score). She reported the following: Coffee (centred), b = 0.190, t = 3.23, p = 0.015. How would you interpret her findings?

A) At the mean value of social media usage there is a significant negative relationship between coffee consumed and test score.
B) At the mean value of social media usage there is a significant positive relationship between coffee consumed and test score.
C) At the mean value of social media usage there is no relationship between coffee consumed and test score.
D) At the mean value of social media usage there is a significant relationship between coffee consumed and test score.

Answer 1

At the mean value of social media usage there is a significant positive relationship between coffee consumed and test score.

Question 2

What does a significant result in a Sobel test mean?

A) That the predictor does not affect the outcome variable due to the introduction of the mediator variable
B) That the mediator variable significantly effects the outcome variable via the predictor variable
C) That the predictor significantly effects the outcome variable via the mediator variable
D) That the mediator variable does not affect the outcome variable via the predictor variable

Answer 2

That the predictor significantly effects the outcome variable via the mediator variable

Question 3

What would you have to create in order to include categorical predictors in a moderation or mediation analysis?

A) Dummy variables
B) Zombie variables
C) Z variables
D) Extraneous variables

Answer 3

Dummy variables

Question 4

A doctor conducted further clinical trials on the new drug's relationship to patient wound healing times; building on her previous study, she was interested in the potential mediating effects of patient age. She conducted a mediation analysis and had the following output: b = 0.31, 95% CI [0.01, 0.30], p = 0.054. How would you interpret her results?

A) There is no indirect effect between the new drug and patient wound healing times mediated by patient age
B) There is a direct effect between the new drug and patient wound healing times not mediated by patient age
C) There is an indirect effect between the new drug and patient wound healing times mediated by patient age
D) There is an indirect effect between the new drug and patient wound healing times not mediated by patient age

Question 5

A sports scientist conducted an analysis of the relationship between consumption of energy drinks (measured in units consumed daily) and athlete performance in a physical test (measured by test score). He had found that there was a positive relationship between the predictor and the outcome variable. However when he introduced the variable consumption of protein (measured in daily calories obtained from protein) he found that the strength of the original relationship weakened. How might he now summarise the relationship between energy drinks and physical performance?

A) It has an indirect effect mediated by protein intake
B) It has a direct effect mediated by protein intake
C) It has no effect at all
D) It has no significant relationship.

Question 6

A doctor conducted clinical trials on a new drug's relationship to patient wound healing times and had the following result: b = 0.46, p = 0.02. How would you interpret her results?

A) A direct effect between the new drug and patient wound healing
B) No direct effect between the new drug and patient wound healing
C) No indirect effect between the new drug and patient wound healing
D) An indirect effect between the new drug and patient wound healing

Question 7

What are two of the advantages of using the PROCESS tool instead of the usual regression menu in SPSS?

A) It allows manual computation of predictors and predictor centring
B) It centres predictors and produces simple slopes analysis
C) It produces simple slopes analysis and errors scores
D) It centres predictors and produces complex slopes analysis

Question 8

Why is it better to use confidence intervals to identify indirect effects rather than the Sobel test?

A) It allows for more sophisticated analysis of degrees of mediation
B) It allows for error to be identified within our analysis
C) The Sobel test is outdated
D) The Sobel test only identifies direct effects

Question 9

In the researcher's analysis of what influences self-esteem in adolescent girls he found that the interaction of time spent in daily physical activity and body image level was a significant predictor of self-esteem. What sort of analysis does he need to do to work out the nature of the moderation effect?

A) Complex Slope Analysis
B) Multi-slopes Analysis
C) Simple Slopes Analysis
D) Linear Slopes Analysis

Question 10

How would you interpret the following Sobel Test results: b = 0.134, z = 1.29, p = 0.510?

A) There is a direct effect between the mediator variable and the outcome and predictor variables.
B) There is an indirect effect between the mediator variable and the outcome and predictor variables
C) There is no indirect effect between the mediator variable and the outcome and predictor variables
D) There is no direct effect between the mediator variable and the outcome and predictor variables

Question 11

A health researcher had conducted trials testing the influence of age and cancer survival rates. She found a significant relationship between the two variables, with younger patients more likely to survive for two years following the completion of treatment than older patients. However, during the clinical trials she observed that a significant number of her older participants had good survival rates, irrespective of their ages and she suspected that their positive mental attitudes might also have an influence on their likelihood of survival. What sort of analysis should she conduct to explore this further?

A) Mediation
B) Moderation
C) Multi-Level Modelling
D) Multiple Linear regression