Results and Data
In normally distributed data, what percent of the data will lie within +/- 3 standard deviations from the mean?
99%
I want to compare the heart rate of nursing home residents by insurance type. Which test should I run and why?
ANOVA
*Rejecting the null when the null hypothesis is true is an example of what type of error?
Type 1 error
You are interested in knowing if the cost of a hip replacement is more at OrthoArkansas when compared to UAMS. In this scenario, would you use the one-tail or two-tailed p-value? Why?
One tailed, direction is established.
*Two researchers, Harvey and Chuck conducted correlation tests between two continuous variables. Harvey reported a correlation coefficient of -0.75 and Chuck reported 0.75. Which of the two researchers reports the strongest relationship and why?
They are equally strong but in opposite directions
You want to know if there is a relationship between having high cholesterol and hospital readmission. Which test should you run and why?
Chi-square, both are dichotomous
You run a simple regression on two variables and achieve an R2 of 0.89. What does this mean and how confident are you in the results that are produced?
The model is good at predicting the outcome and I would feel pretty confident in my results.
What is the name of the statistical test that should be used for comparing two dichotomous variables when one of the cells is <5?
Fisher's exact
Calculate the expected values for the first two rows
1st cell: 30.9
2nd cell: 24.1
*You want to analyze the factors that influence whether or not patients die in the hospital while accounting for age, gender, race, insurance, and ICU length of stay. Which type of regression analysis would be more appropriate? Why?
logistic because binary outcome.