Measuring agreement between categorical and continuous variables

Hi all,

First post here. I'm currently measuring agreement between continuous and categorical variables that are also in different units of mesurement (and non normally distributed). I have physical activity measures from a questionnaire (low, moderate and high activity levels) and 9 measures from a body worn sensor. Bland altmans obviously isn't an option due to the categorical variables and cohens kappa isn't either as measures are in different units.

I've used spearmans rho to measure correlation and kruskall wallis to show continuous variables from the body sensor and sognificantly different across low, moderate and high physical activity cateogies.

Any other thoughts on demonstrating agreement between categorical and continuous variables in different units of measurement?