# Significance Test for Attribute Data

Six Sigma – iSixSigma › Forums › General Forums › General › Significance Test for Attribute Data

This topic contains 4 replies, has 2 voices, and was last updated by Yasmani Ruiz Bello 9 months, 2 weeks ago.

- AuthorPosts
- February 26, 2018 at 5:22 pm #55949
Hello,

How are you?

I was wonder if you guys can help me with the below:I have sampled a total number of defectives orders per month for one year (12 orders per month) in order to identify where (what part of the day) the majority of the errors were happening.

After plotting the data I observe that most the defectives orders happened between 1:00 pm and 3: pm.Is there anyway I can prove this is applicable to the population?

Thank you in advance

February 26, 2018 at 5:49 pm #202323If your sample size is adequate, you certainly can. Look up the sample size formulas to determine your confidence level in your measurement. We typically want 90% or better. But even if that isn’t the case you may have found something that needs further investigation. Does something happen during or just before this time period that’s different from the rest of the day? Is there a shift change? Is there some other scheduled activity? Could lunch break have some influence?

February 26, 2018 at 6:05 pm #202324Thank you for the quick reply.

shift ends at 3 PM..I have been trying to identify possible root cause but nothing yet.

so, for me to assume this is applicable to the population I will have to determine the value of Z on below formula right?n=(Z squ x p-bar x (1-p-bar))/ delta p squa

If this is the case what would my Delta p(desired proportion interval)be? 1%?

Thank you very much

February 26, 2018 at 6:37 pm #202326Before you get too far down in the weeds why not use one of the sample size calculators you can find free on-line? That’ll tell what your adequate sample size should be given your population. Or, given that you know both, your confidence level. Search for “confidence level calculator”.

February 26, 2018 at 7:15 pm #202327Thank you very much

- AuthorPosts

You must be logged in to reply to this topic.