What statistical tools to use to analyze 4 datasets?

[#1]

I have 4x production runs with 100 in-process dimensional inspections. All 4 datasets are of the same dimension, same process but different controls/tools, and different machines/operators. What statistical analysis tools can I use to make observations about each production run and compare processes to determine the better performing process?

[#2]

What is the definition of "better performing process", yield, lower defects, etc.?

[#3]

Great question! These are continuous datasets with defined LSL and USL values, therefore I imagine process with least variability. I calculated yield and the results are high for all 4 datasets, so there is not enough information there to differentiate the processes based on it.

[#4]

In addition to @txnelson 's question here is a general workflow/decision tree I used often to decide which analytical tool works 'best'.

In this order:

1. Articulate the practical questions you need to answer or decide.

2. Does the data collection method support answering these questions? Often times happenstance or observational data is very limited wrt to addressing issue 1. above. Am I missing important signals or noises? How were nuisance or noise variables handled? Once satisfied with 2 move to 3. below, otherwise devise a new data collection strategy and then execute that strategy.

3. Now onto analysis...three words of advice: 1. plot the data 2. Plot The Data. 3. PLOT THE DATA in a way that allows answering/addressing 1. above. Move to other analysis techniques as needed.

[#5]

In addition to the invaluable comments from @txnelson and @P_Bartell, I would like to suggest to try for identifying the 4 processes behaviors by using the tools and/or techniques in the files, which I prepared based on specific scenarios, attached. Example shots are below.

  

[#6]

Welcome to the community.  First, make sure all of the data from the 4 production runs is in the same table with columns describing how the data was collected (e.g., machine, operator, controls, tools and response variables (this is what is unclear and what Jim is asking)).  Then you have lots of options.  Graph builder to look at relationships, control charts for stability, fit model for quantitative analysis, etc.

Sorry I had not read all of the responses when I posted this...