Please help me figure out what are the best test to run for the questions below using the informations provided.

Students at Tufts recently studied contemporary attitudes on cost-effectiveness and equity by presenting the following scenario (created by Ubel, NEJM 1996) to a sample of medical students, medical residents, and physicians:

"A group of doctors was formed to help the government decide which of two tests to offer low-risk people. Test 1 is inexpensive but does not always detect cancers in their early stages. Test 2 is more expensive but is better at detecting early cancers. The decision is complicated by budget limitations: the government has only a certain amount of money available to pay for the screening tests. After evaluating the costs and benefits of each test, the doctors have reached the following conclusions. The budget is just large enough to offer Test 1 to all the low-risk people. With this approach, everyone can receive the test, and 1000 deaths from colon cancer will be prevented. The budget is just large enough to offer Test 2 to half the low-risk people. With this approach, half the people can receive the test and half cannot, and 1100 deaths from colon cancer will be prevented"

Respondents were asked to select which test they would rather offer, Test 1 (the "equitable" option) or Test 2 (the "cost-effective" option). [Which one would you choose?]

The data are found in the file ScreeningEquityHW.dta. Variables include Response (Test1 or Test2), M1F2 (gender, male=1, female=2), Group (1=students, 2=residents, 3=physicians), and UbelTest1 (1=respondent chose Test1 ["the equitable option"], 0=respondent chose Test2 ["the cost-effective option])

1. Is the proportion of people who prefer the equitable option the same across medical students, medical residents, and physicians?

2. Are the odds of selecting the equitable option equal for physicians and medical students?

3.Are men and women equally likely to select the equitable option?

4.Looking at the distribution of gender across the groups, and considering your answer to part C, do you want to modify your answer for B?

Does the story change when you stratify by gender?Researchers in the Tufts Orthopedic Department noticed that, when recommended certain surgeries, Asian patients were more likely to decline the surgery compared to White patients. The dataset DistalRadiusFractureHW.dta contains information on patients who were seen at Tufts after breaking their wrist. All patients were given the recommendation that they undergo surgery.Variables include age in years, male (0=female, 1=male), ethnicity (0=white, 1=asian), englishPrimaryLanguage (0=no, 1=yes), insured (0=no, 1=yes, 2=unsure), smoker (0=no, 1=yes), provider (a categorical variable for the 4 different surgeons in the study), and underwentSurgery (0=no, 1=yes).

1. Does the proportion of White and Asian patients who elect to take the surgery differ?
2. Does that observed relationship hold after controlling for potential confounders?
3. Does that observed relationship appear to be the same in men and women? (Hint: stratify by sex or make an interaction term)

The Tufts Orthopedics department conducted exploratory research to investigate which factors predicted whether a patient would not be able to attend a schedule appointment. A subset of these data can be found in MissedAppointmentsHW.dta

1. Which factors are significant predictors of whether a patient will miss an appointment?

Variables of interest include age, meantempf (the temperature in Boston on the day of the appointment), month (the month of the appointment), ins1 (the type of insurance held by the patient), problem (whether the appointment was for hip or knee complaints), race1 (a categorical variable for the race/ethnicity of the patient), male (0=female, 1=male), dayofweek, hour (the scheduled time of the appointment), zipsalcat (a categorical variable for the median income of the patient's home zipcode), make (0=missed appointment, 1=made appointment), miss (0=made appointment, 1=missed appointment), english (a binary variable for whether the patent speaks english as primary language), rainsnow (a binary variable for whether there was precipitation on the appointment day), and logzipdist (the log-transformed distance from the patient's home to Tufts).