topic Re: Probit Slope Determination in Discussions

Probit Slope Determination

tox — Thu, 01 Mar 2018 18:40:33 GMT

I have a toxicology dataset and I want to calculate the median lethal dose and probit slope. I have been able to follow some other threads and calculate the median lethal dose using probit analysis. The results of an example analysis I performed are shown below. I am looking to estimate the probit slope as well as the median lethal dose - is the probit slope the "log(load)" value under parameter estimates (i.e. 3.62)? Any help/advice would be greatly appreciated. Thanks!

Re: Probit Slope Determination

cwillden — Thu, 01 Mar 2018 23:01:04 GMT

It looks like you applied a log transformation to Load in the dialog, but yes that is the slope.

To get the LD50, you can just play with the profiler value for Load until you get 0.5, or you can use inverse prediction from the red-arrow menu next to "Generalized Linear Model Fit." Just put whatever confidence in there and specify 0.5 for your quantile. The predicted load value will show up in the results and you can just ignore the confidence interval stuff.

Re: Probit Slope Determination

tox — Fri, 02 Mar 2018 14:34:05 GMT

Great - thanks for the help!