Statistics problem: Comparing written work with & w/out use of AI

[TULC]

I want to compare performance on written work under different conditions, for example with and without the use of AI, according to some specified criteria. Assume the written work is a critical analysis of specific content.

The written work will be scored on a number of dimensions, such as creativity etc. The goal is to gain some understanding - based on a large sample of written samples - of the extent to which AI can improve the written work. This will be a way to develop a benchmark against which we can compare individual written samples. If the correlation b/w individual written work with and w/out use of AI is sig. weaker than expected based on the analysis of a larger sample of written work, then one could argue that this warrants a question: is AI being overused by the individual?

Given the above, would calculating correlation coefficients be a good choice here? I want something simple that can be used with ease by almost anyone. At the same time, I acknowledge the fact that I am manipulating some variables, so a correlational approach may not be ideal. If so, what alternatives, if any, would you suggest?

 

[Hornbein]

You might have more luck posting this in the statistics group. The moderators might move it if you ask them. (If you post it there yourself they will complain about a duplicate.) You can make this request by hitting the "Report" button and typing it in.

 

[berkeman]

Thread closed for Moderation...

 

[PeterDonis]

I want to compare performance on written work under different conditions, for example with and without the use of AI, according to some specified criteria.

This is not allowed at PF for two reasons: first, we don't allow discussion of personal research; and second, we don't allow discussions based on AI-generated content.

Thread will remain closed.

 

[PeterDonis]

You might have more luck posting this in the statistics group.

Not this topic, no. See previous post.