Assignment 9
Logit/Probit
The Challenger Disaster
The last flight of the space shuttle Challenger was conducted under conditions quite different from the previous 25 shuttle flights. The morning temperature at launch was 28 degrees, more than 20 degrees colder than any previous launch. Despite pre-flight debate about the effectiveness of the booster rocket joints and seals (O-rings) under frigid weather conditions, the launch was approved and the consequences of the decision explosive.
In "shuttle1.dta" you will find relevant evidence reported by the President's Commission on the Space Shuttle Challenger Disaster (pp. 129-131). These data identify the flight (STS-#) and date of all shuttle missions up to the Challenger accident, whether or not the O-rings showed evidence of distress in terms of erosion or total failure ("blow-by"), the pressure in each solid rocket booster, and the temperature of the joints at launch. (O-ring problems were identified in recovered rocket boosters following each flight.) The potential for major disaster as a result of O-ring failure was known to both the manufacturer, Morton-Thiokol, and to NASA.
TASKS:
- On the basis of past experience, consider the question of whether
temperature is related to the failure of rocket seals. Is "distress" measured
on the recovered booster rockets of past flights related to flight launch
temperature controling for joint pressure. One of several dependent variables
could be examined. The dichotomous variables - did O-ring erosion or blow-by
occur (or did either occur)? Or ordered categorical dependent variables (the
number of eroded O-rings, the number of blownby O-rings or the number of both)
could be examined. You may examine this question from the perspective of
logistic regression or probit or both.
- Does age of the program matter? (How might you evaluate the alternative
hypotheses that the program learned over time or that the equipement
degraded?
- Once you have a predictive model, consider the effect of a 31 degree launch on the predicted failure of O-rings.
Sources:
S. Dalal, E. Fowlkes and B. Hoadley (1989) "Risk Analysis of the Space Shuttle: Pre-Challenger Prediction of Failure." Journal of the American Statistical Association. 84: 945-957.
William Rogers et al. (1986) Report of the Presidential Commission on the Space Shuttle Challenger Accident. Washington DC.
Resources: