Looking for some insight on how to property analyze the data set attached to this post. The company has been tracking the amount of scrap in pieces each week and the total number pieces produced to get a % scrap for each week. The first 26 weeks are the baseline data, before any improvement actions were taken. Weeks 27 through 31 represent the weeks after some improvement actions were taken.
It would be a mistake to use the overall average of the % scrap values in the Before and After group because the denominators in each row of the calculated percentages are different so I've used Summary to calculate the overall % scrap of the Before group and the After group as shown here:
The % scrap values for both groups are close so it doesn't look promising for us to detect a significant difference between the two groups. My specific question is what is the best way to calculate that (lack of) significance?
I'm also wondering if I should treat Qty Produced as a frequency column since each weeks that produced hundreds of parts should contribute more to the data set than weeks that only had 2 parts produced. Without incorporating frequency when we run 2 parts and have 0 scrap that is treated the same as a week where we run 142 parts and have 0 scrap. If I use the fit Y by X platform and put qty produced in Freq I get a significant result, with the Before group showing more instances of scrap on average than the After group.
