The probit model links the stress level to the failure probability through the chosen cumulative distribution function. The probit model can be expressed in terms of the shape and scale (Weibull distribution) or location and scale (all other distributions) of the chosen distribution. For the normal distribution, the location parameter equals the mean and the scale parameter equals the standard deviation.
The probability of unit failing at a given stress level can be restated as the probability that a unit's tolerance is less than a given stress level. The distribution of these probabilities is known as the tolerance distribution.
Standard Error | 95.0% Normal CI | |||
---|---|---|---|---|
Parameter | Estimate | Lower | Upper | |
Mean | 692.416 | 18.3649 | 656.421 | 728.410 |
StDev | 111.612 | 19.4518 | 79.3167 | 157.058 |
For the windshield data using a normal distribution, Mean = 692.416 and StDev = 111.612. The tolerance distribution models the probability that a particular windshield's breakage threshold is less than a given velocity.