## Probit analysis

### Probit analysis

Hi all,

I am doing some dose-response experiments and want to use probit
analysis to calculate the 50th percentile values. When I use the probit
analysis routine of SPSS 9, I get weird results. Here is a sample set of
data:

Concentration    number of subjects    observed responses
0.00                    78                                    2
0.06                    49                                    4
0.13                    61                                    14
0.18                    114                                  68
0.20                    99                                    78
0.23                    105                                  79
0.25                    82                                    77
0.50                    89                                    89

The output from the probit analysis gives an intercept of -2.26864 and a
slope of 14.22993. The graph of the regression line shows all probit
transformed values to lie between -2 and 1.8. This line and the data
points on the graph do not make any sense to me. My data should convert
to probit values between 3 and 7+. What am I doing wrong? How do I get a
printout of the actual probit values used by SPSS?

Thanks,

Tony Haigh

### Probit analysis

< ... >

Quote:>                           ... The graph of the regression line shows all probit
> transformed values to lie between -2 and 1.8. This line and the data
> points on the graph do not make any sense to me. My data should convert
> to probit values between 3 and 7+. What am I doing wrong? How do I get a
> printout of the actual probit values used by SPSS?

Tony,
It sure looks like, " ... add 5.0".  That is a guess, but a pretty
solid one.

Since that original probit add-on of 5.0 has not been needed for the
convenience since the time we gave up hand-computing, it should
transparent enough if someone has decided to do without -- though I
agree that they ought to mention it and label it, if they did not.

Since you don't mention that you have tried that obvious adjustment,
you probably ought to have someone on call for advice who would have
told you at least that much...  and who you would discuss things with,
regularly.  When you don't have the basic background and vocabulary,
there is no telling how much nonsense may be getting past.

--

http://www.pitt.edu/~wpilib/index.html

Hello,

I want to perform a multivariate probit analysis (say for two dimensions)
with a random effect. Somethere I read that this should be possible in PROC
NLMIXED. Does someone have an example code to see how this can be performed,