Pre and post comparison for nonparametric unpaired sample

#1
Hi. I am not sure how to compare the results from the following study.
Number of wrong orders out of total orders placed every day for 60 days as the pre sample (significant variation eg 2 wrong out of 6 on day 1, 5 wrong out of 15 on day 2, and so on till day 60)
Intervention to decrease the number of wrong orders placed
Similar data collected for next 60 days

What test do I use? Mann Whitney? Unpaired t test? Some other test? How do I analyze this data?
Thanks for your help
 
#2
One suggestion is to accumulate the data before and after. e.g. Before 125 right out of 180, After 155 right out of 175. Then use the Chi Square test, or test for the difference between two proportions.
 

Miner

TS Contributor
#3
The following suggestion may not be accepted in the biostatistics field, but is common in my field of industrial statistics.

Since your samples are taken in time series order, a common tool in my field would be a p-chart. See example below. This is a general explanation of what a p-chart is and what it is used to monitor. The second image are tests to detect different types of instability. The one of most interest to you would be the 9 points in a row on the same side of the center line. The center or mean line is calculated from the first 60 days data.

Another approach would be to use ANOM. While ANOM and the suggested chi-square are legitimate approaches, they both ignore the information inherent in the time series structure.