Hello everyone,

i am curious, if it is recommandable to estimate the heritability of a quantitative trait (where actually exact data are known) in a case-control study. Like i have 300 german shepard and i know exactly the length of the tongue of each dog in cm. Would it be more accurate to determine a threshold and say everything below this treshold is short and everything above is long, encoding it with 0 and 1 and run a GREML. I did this and i have a nice result, whereas in my first estimation (where i inputed the length in cm), i needed to specify the option --reml-no-constrain and got a heritability above 1

Thanks in advance