Multiple Regression Assumption Checks
IMAGE 1
Interpret this graph
There is likely no auto-correlation, and the variances appear to be homogeneous.
What test should we run to test this?
*A student nicknamed Pikachu wants to know if there is a difference between legendary pokemons and non-legendary pokemons in their speed. Assume speed is a continuous variable.
*used from class
Independent-samples t-test
Regarding the following scenario: A student nicknamed Pikachu wants to know if there is a difference between legendary pokemons and non-legendary pokemons in their speed. Assume speed is a continuous variable.
After running the t-test, researchers found that legendary pokemons had a mean of 80 and non-legendary pokemons had a mean of 82.5. The 95% confidence interval is [-0.6, 2.3]. Do we reject or fail to reject the null hypothesis?
Fail to reject the null hypothesis because the confidence interval contains 0
Every dummy variable must be dichotomous. True or False?
True
IMAGE 2:
The correlation coefficient of the observations in blue is -0.96. What would happen to the correlation coefficient if we added the red dot to the scatterplot of blue dots?
gets closer to -1
What test should we run to test this?
*Nancy wants to know if adolescents’ executive function (EF) performance differs under stressful versus regular situations. She has her participants play a game measuring EF regularly versus under a timed, stress-inducing environment. Every participant will experience both conditions and the order of the two is counterbalanced. She is taking a conservative two-tailed approach, and sets the alpha at .05.
*Taken from class modules
Paired t-test
When you make dummy variables from 1 categorical variable with 4 categories, how many dichotomous dummy variables do you end up with?
4
Image 3
Interpret this graph
There is likely auto-correlation, and there is likely no homogeneity of variances
What test should we run?
*
You are trying to see if the Psychology department makes its students spend more/less time in statistics compared with all departments in the university in general. You collected data from 500 Psychology students and found that they are required to spend around 2,000 minutes in stats class on average (SD = 150). There is somehow a database of statistics education for the entire university that shows that students typically spend 2,500 minutes in stats class during their college years (SD = 200).
*Taken from class modules
Z-test
What is wrong with the following report of an F-test, with df=53?
F(3, 53)
It is missing an associated p-value
What test should we run?
*
Elmo is still interested in knowing if tickling and laughter and eating healthy all contribute to children’s learning of the alphabets. With the help of his friends, Elmo collected data on how many times children were tickled per week, how many times they laughed, how healthy their eating habits are.
However, unlike last week, how healthy their eating habits are is now operationalized as a categorical variable:
The outcome variable is still how well they learned the alphabets (continuous variable).
*Taken from class material
What is the proper interpretation of the standardized coefficients in the following table?*
IMAGE 4
*Table taken from class notes
After controlling for the number of flavors sold at a store, a one standardized unit increase in temperature is associated with .958 standardized unit increase in revenue (𝛽= .958, SE = .011, p < .001). Accounting for temperature, one standardized unit increase in the number of flavors sold at a store is associated with an average revenue increase of .268 standardized unit (𝛽= .268, SE = .011, p < .001).