Hi Marie, thank you for bringing this to our attention. You are correct. We apologize for the confusion. Technically, Mann Whitney U or Chi-squared test could be correct for this question as they are both non-parametric tests (and t-test and pearson correlation are parametric tests). Mann Whitney U is recommended for ordinal-level data that can be ranked (such as attributes that can be ranked: IVH grades, MMT grades), although we don’t know the exact distance between the ranks/values. Chi-square is recommended for nominal or categorical outcomes where you can label an attribute, but they aren’t ranked (lowest level of measurement, such as race, gender).

Regarding use of parametric tests, ratio level data has a natural zero and is the highest level of measurement (time, distance, etc) while interval level data has a designated order and distance between values but not a natural zero (temperature).

Most of our therapy measures are ordinal or interval level. The best example is that the GMFM-88 is an ordinal scale because the exact difference between item scores can’t be determined. However, the GMFM-66 underwent Rasch analysis to determine item difficulty and convert the test to an interval scale: the GMAE software can calculate score changes based on item difficulty because the difference between score values is known. I have often read that many of our pediatric measures (PDMS, Bayley, etc) are “interval-like” and sort of hover between interval and ordinal data. That is why you often see parametric tests used to analyze that data in studies.