PRINT Subcommand (PROBIT command)

Use PRINT to control the statistics calculated by PROBIT.

• PROBIT always displays the plot (for a single-predictor model) and the parameter estimates and covariances for the probit model.
• If PRINT is used, the requested statistics are calculated and displayed in addition to the parameter estimates and plot.
• If PRINT is not specified or is specified without any keyword, FREQ, CI, and RMP are calculated and displayed in addition to the parameter estimates and plot.

DEFAULT. FREQ, CI, and RMP. This is the default if PRINT is not specified or is specified by itself.

FREQ. Frequencies. Display a table of observed and predicted frequencies with their residual values. If observations are entered on a case-by-case basis, this listing can be quite lengthy.

CI. Fiducial confidence intervals. Print fiducial confidence intervals ¹ for the levels of the predictor needed to produce each proportion of responses. PROBIT displays this default output for single-predictor models only. If a categorical grouping variable is specified, PROBIT produces a table of confidence intervals for each group. If the Pearson chi-square goodness-of-fit test is significant (p < 0.15 by default), PROBIT uses a heterogeneity factor to calculate the limits.

RMP. Relative median potency. Display the relative median potency (RMP) of each pair of groups defined by the grouping variable. PROBIT displays this default output for single-predictor models only. For any pair of groups, the RMP is the ratio of the stimulus tolerances in those groups. Stimulus tolerance is the value of the predictor necessary to produce a 50% response rate. If the derived model for one predictor and two groups estimates that a predictor value of 21 produces a 50% response rate in the first group, and that a predictor value of 15 produces a 50% response rate in the second group, the relative median potency would be 21/15 = 1.40. In biological assay analyses, RMP measures the comparative strength of preparations.

PARALL. Parallelism test. Produce a test of the parallelism of regression lines for different levels of the grouping variable. This test displays a chi-square value and its associated probability. It requires an additional pass through the data and, thus, additional processing time.

NONE. Display only the unconditional output. This option can be used to override any other specification on the PRINT subcommand for PROBIT.

ALL. All available output. This is the same as requesting FREQ, CI, RMP, and PARALL.