I need to model the variable Total motile Count which describe how many million sperm cells in an entire ejaculate are motile. It is not a proper count since it is calculated as a product of other variables: sperm count(x10^6/mL) x motility(%) x volume(mL).

Therefore, it is not continuous. There are no zeroes (min=2 max=650, mean=129.56, var=10916.37). What family of distribution should I consider to model Total motile Count as response variable? I considered Gamma because the number of counts is very high and quasi-poisson distribution. What do you suggest? I tried Poisson and negative binomial (overdispersion parameter = 0.6063, CI: 0.4375-0.8403), suggesting that the latter is preferred (smallest AIC, also over Gamma). Thanks!

enter image description here


If motility is constant, Total motile Count is also a Poisson distribution.

Y = sperm count(x10^6/mL) x motility(%) x volume(mL).

E[Y] = E[sperm count]*(x10^6/mL) x motility(%) x volume(mL).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.