CHI-SQUARED FOR PERCENTAGES

This is a short summary for a lot of work. So, can one use chi-square for percentages? In very few cases, it MIGHT be acceptable. However, it is best not to use percentages.

Please see the topic in Wikipedia. Further, when one has even raw numbers and you have a 2 by 2 test, you can not use real numbers like 50 in the lower third and 50 in the lower fourth.

This is like using a ”dummy” variables in multiple regression. Unfortunately, we now have meta-analysis. You are applying statistics like chi-square on literature reviews. All the above will come to pass that the behavioral methodologies use the simple statistics because the numbers are heterogenous. We may even gain the respect of the hard sciences.

This author believes that as the soft sciences should not try to stretch soft into hard statistics. I have done this myself. I apologize.