Re-estimate the logistic regression without observation 14


Refer to the Challenger data from Exercise 4 in Chapter 2. Fit a model including only temperature as an explanatory variable.

(a) Examine the deviance/df statistic. Also conduct a graphical assessment of the residuals. Discuss the fit of the model.

(b) Examine the goodness of fit for the model and draw conclusions.

(c) Conduct an influence analysis and draw conclusions.

i. Explain why h14 is so large.

ii. Re-estimate the logistic regression without observation 14 in the data set. For both the model with and without this observation, compare confidence intervals for the parameters, and plot the estimated probability of O-ring failure against temperature on one plot. Do the results change in a meaningful way?

Exercise 4

The failure of an O-ring on the space shuttle Challenger's booster rockets led to its destruction in 1986. Using data on previous space shuttle launches, Dalal et al. (1989) examine the probability of an O-ring failure as a function of temperature at launch and combustion pressure. Data from their paper is included in the challenger.csv file. Below are the variables:

• Flight: Flight number

• Temp: Temperature (F) at launch

• Pressure: Combustion pressure (psi)

• O.ring: Number of primary field O-ring failures

• Number: Total number of primary field O-rings (six total, three each for the two booster rockets)
The response variable is O.ring, and the explanatory variables are Temp and Pressure. Complete the following:

(a) The authors use logistic regression to estimate the probability an O-ring will fail. In order to use this model, the authors needed to assume that each O-ring is independent for each launch. Discuss why this assumption is necessary and the potential problems with it. Note that a subsequent analysis helped to alleviate the authors' concerns about independence.

(b) Estimate the logistic regression model using the explanatory variables in a linear form.

(c) Perform LRTs to judge the importance of the explanatory variables in the model.

(d) The authors chose to remove Pressure from the model based on the LRTs. Based on your results, discuss why you think this was done. Are there any potential problems with removing this variable?

Request for Solution File

Ask an Expert for Answer!!
Basic Computer Science: Re-estimate the logistic regression without observation 14
Reference No:- TGS01645112

Expected delivery within 24 Hours