STA429/1007 Assignment 9

Quiz on Friday March 23 at 12:10 p.m.

The file bweight.dat has these variables for a sample of new babies:

 1. Identification Code  
 2. Low birth weight (1=Yes: Under 2500 grams, 0=No)
 3. Mother's age (In years)
 4. Mother's weight at last period (In pounds)
 5. Mother's race (1 = White, 2 = Black, 3 = Other)
 6. Did mother smoke during pregnancy? (1=Yes, 0=No)
 7. History of Premature Labour (Number of times)
 8. History of Hypertension (1=Yes, 0=No)
 9. Presence of Uterine Irritability (1=Yes, 0=No)
10. Visits to Doctor During 1st trimester (Number of visits)
11. Birth Weight (In Grams)

Do a logistic regression, with low birth weight as the dependent variable. Set up your own dummy variables for race, making White folks the reference category. There should be no interaction (product) terms in your model. When a significance test is necessary to answer the questions below, use a Wald test.

  1. Why would it be silly to use birth weight as one of the independent variables?
  2. To check against errors in reading the data and gross errors in setting up the model, does your overall test of all the regression coefficients except the intercept have a Wald Chisquare of 25.7025 with 9 degrees of freedom?
  3. Controlling for 7 other independent variables, do the three "race" groups differ in their chances of having a low birth weight baby?
    1. Answer the question YES or NO.
    2. What is the numerical value of the test statistic? The answer is a number.
    3. What is the p-value? The answer is a number.
    4. Do you reject the null hypothesis? Answer Yes or No. Are the results statistically significant? Answer Yes or No.
  4. If the answer to the preceding question was "Yes," carry out all pairwise tests between race groups to see where the difference comes from. Protect the three follow-up tests with a Bonferroni correction, so that the joint significance level is 0.05. You will only declare the results significant if the p-value on your printout is less than … What value? The answer is a number. For each of the three tests,
    1. What is the numerical value of the test statistic? The answer is a number.
    2. What is the p-value? The answer is a number.
    3. Do you reject the null hypothesis? Answer Yes or No. Are the results statistically significant with the correction? Answer Yes or No.
    Finally, what do you conclude from the follow-up tests? State your findings in plain, non-statistical language.
  5. Be able to interpret all the significant Wald tests that are produced by default.
  6. Be able to interpret all the estimated odd ratios that are produced by default. For example, the controllng for all other independent variables, the estimated odds of a low birth weight baby are ___ times as great for Black mothers, compared to White mothers. Weestimate that mothers who smoked during pregnancy are ____ times as likely to have a low birth weight baby.
  7. In your opinion, what is the most obvious independent variable that should be in this data set, but is not?

Please bring your log file and list file to the quiz.