[R-sig-eco] Zero infalted model- T50

[Bob O'Hara]

On 08/07/15 09:56, Mehdi Abedi wrote:
> Hi,
> I want to calculate time to 50 percent reduction of seed viability. Mainly
> researcher calculate P50 using probit model for calculation T50.
>
> In some treatments seed viability decline quit fast therefore i have only
> value 100 and 0 values . Can i calculate T50 for this data using probit?!
> Could we apply Zero inflated model for this conditions with 0?
>
> All the best,
> Mehdi
>
>    Time(weed) Seed viability  0 100  8 0  16 0  24 0  32 0  40 0
>
You can use a probit, but it'll be unhappy. The problem is that the 
curve is steep compared to the times, so you don't see it. The only real 
solution is to change the assays, so you observe more times. But I 
appreciate this might not be practical.

I don't think zero inflation will help. There isn't a unique maximum 
likelihood estimate for this, but I think a sensible estimate would be 
mid way between the times where you observe 100 and the first 0. If you 
want to be Bayesian, this would be the posterior mean. The best 
confidence interval is probably the central 95% of the interval that the 
T50 is in (i.e. assuming the likelihood is flat). It's probably a little 
bit wider than it should be, but I doubt that will be a big problem.

I used to get data where the data would flip between 0 and 100 over a 
range of doses. In the end I gave up with those problems and became an 
ecologist

Bob

-- 

Bob O'Hara

Biodiversity and Climate Research Centre
Senckenberganlage 25
D-60325 Frankfurt am Main,
Germany

Tel: +49 69 7542 1863
Mobile: +49 1515 888 5440
WWW:   http://www.bik-f.de/root/index.php?page_id=219
Blog: http://blogs.nature.com/boboh
Journal of Negative Results - EEB: www.jnr-eeb.org