# Confidence interval/statistic question

1. Nov 22, 2013

### abe_cooldude

Hi all,

I am trying to do something that I am not sure how exactly it can be done. Not very good at statistics.

I have a bracket, sample size of 26, that was designed for 120lbs. It holds weight from 120lbs to 150lbs. Only 1 amount of weight gets applied at a time.

Each of the bracket gets tested for predefined number of miles, and each bracket in the sample size only gets tested once. If the bracket doesn't bracket and reaches the predefined miles, then it's good. Higher mileage means the bracket did good.

The goal is to see how long a bracket survives without breaking under a certain weight. Higher weight loading (150lbs vs. 120lbs) have higher weight in determining how good a bracket is and vice versa.

Here's what data looks like:
6 brackets at 120lbs on 12000/12000 miles course (good)
2 brackets at 120lbs on 10600/7600 miles course (better)
4 brackets at 120lbs on 12000/7600 miles course (even better)
4 brackets at 120lbs on 9100/7600 miles course (good)
4 brackets at 120lbs on 17700/7600 miles course (awesome)
2 brackets at 140lbs on 7600/7600 miles course (even better)
2 brackets at 150lbs on 12000/12000 miles course (greatest)
1 brackets at 150lbs on 3150/7600 miles course, broke (bad)
1 brackets at 120lbs on 4150/7600 miles course, broke (very bad)

I am trying to say that the bracket that was designed for 120lbs has a some percentage likely hood of doing well based on the data, and provide some kind of plot, confidence interval.

Any idea where to start or what I need to do? Not at all familiar with stats.

Thanks,
Abe

2. Nov 22, 2013

### abe_cooldude

The more I read on the internet, the more I think this problem requires the use of Weibull plot.

3. Nov 24, 2013

### Stephen Tashi

As I understand the situation, both the weight supported and the mileage travelled are factors that tend to break a bracket. What probability do you want to estimate that involves both factors?

You could look for a table that says P percent of brackets will not break if they support a weight of W for M miles. That type of table would have 3 entries per line: P, W, M.

It's simpler to analyze the durability of a part from tests that are test-till-failure. However, I'm sure analyzing durability from tests that are fail vs no-fail has been studied and has its own jargon . Offhand, I don't know what that jargon is.