A dialogue lets the user assign variables as doses, counts of successes and
totals. If desired; a logarithmic transformation may be applied to the dose
variable. The script calculates and adds all columns neeeded to perform a
nonlinear regression for the parameters of a probit regression for dose. The
user dialogue asks for #Responses, #Total and Dose. The script adds columns for
relative frequency of the "successes", the probit function and the loss
function. After estimating the regression parameters the report with a set of
inverse predictions and a graph of the fit is displayed. The prediction column
is appended to the data table.
Application: Dose response tests are routine in
lab experiments to assess efficcacy or toxicity of agents. The results are
reported as totals and counts of events. The probit regression estimates the
probability to observe an event as a function of the dose applied. Knowing the
regression function alows for the calculation of frequently used indicators like
ED50, ED90 or so on. A set of these estimators along with their confdence
intervals is part of the report.