Hi all,
I'm trying to set up some statistical tests to compare against some desired specs.
I'm currently using this handbook to do so:
http://www.itl.nist.gov/div898...rc/section2/prc253.htm
Basically, you need to decide how many samples you will test, the proportion of the samples that would fall within those limits, and the probability that this will happen. From that you can get k values which help to define the Lower and Upper Tolerance limits
I simply followed the example they have listed on an excel chart (and using the other two pages they have listed- critical value of chi square, and critiical value of normal distribution which is just a fancy way to say z score) and got the same K value needed
Now I already have a lot of these determined: I have a 95% confidence interval, I want 99% probability of the samples falling in between the limits, and I'm probably going to get about 30 data points (this might be up for grabs).
Actually, I even have my K value from a table that is listed for 2 sided normal tolernace limits: 3.350
Now...I'm not the type who is fine with just using tables; when I can, I like to double check and confirm the data (partially b/c I'm also reviewing stats on my own, but I haven't reached confidence intervals, etc. yet...and the last time I took stats was 4 years ago...which is why I'm studying on my own
).
so I went ahead and checked the K value and I'm getting 3.284242673 (lol cut and paste from excel) which is a far cry from 3.350
Who is right and who is wrong? Am I doing this wrong? Does anyone else also get 3.350?
I'm trying to set up some statistical tests to compare against some desired specs.
I'm currently using this handbook to do so:
http://www.itl.nist.gov/div898...rc/section2/prc253.htm
Basically, you need to decide how many samples you will test, the proportion of the samples that would fall within those limits, and the probability that this will happen. From that you can get k values which help to define the Lower and Upper Tolerance limits
I simply followed the example they have listed on an excel chart (and using the other two pages they have listed- critical value of chi square, and critiical value of normal distribution which is just a fancy way to say z score) and got the same K value needed
Now I already have a lot of these determined: I have a 95% confidence interval, I want 99% probability of the samples falling in between the limits, and I'm probably going to get about 30 data points (this might be up for grabs).
Actually, I even have my K value from a table that is listed for 2 sided normal tolernace limits: 3.350
Now...I'm not the type who is fine with just using tables; when I can, I like to double check and confirm the data (partially b/c I'm also reviewing stats on my own, but I haven't reached confidence intervals, etc. yet...and the last time I took stats was 4 years ago...which is why I'm studying on my own
). so I went ahead and checked the K value and I'm getting 3.284242673 (lol cut and paste from excel) which is a far cry from 3.350
Who is right and who is wrong? Am I doing this wrong? Does anyone else also get 3.350?
Facebook
Twitter