NATRES Subcommand (PROBIT command)

You can use NATRES either to supply a known natural response rate to be used in the solution or to instruct PROBIT to estimate the natural (or threshold) response rate of the model.

  • To supply a known natural response rate as a constraint on the model solution, specify a value less than 1 on NATRES.
  • To instruct PROBIT to estimate the natural response rate of the model, you can indicate a control group by giving a 0 value to any of the predictor variables. PROBIT displays the estimate of the natural response rate and the standard error and includes the estimate in the covariance/correlation matrix as NAT RESP.
  • If no control group is indicated and NATRES is specified without a given value, PROBIT estimates the natural response rate from the entire data and informs you that no control group has been provided. The estimate of the natural response rate and the standard error are displayed and NAT RESP is included in the covariance/correlation matrix.
  • If you have a control group in your data and specify NONE on the LOG subcommand, the control group is included in the analysis.

Example

DATA LIST FREE / SOLUTION DOSE NOBSN NRESP.
BEGIN DATA
1  5 100 20
1 10  80 30
1  0 100 10
...
END DATA.

PROBIT NRESP OF NOBSN BY SOLUTION(1,4) WITH DOSE
  /NATRES.
  • This example reads four variables and requests a default analysis with an estimate of the natural response rate.
  • The predictor variable, DOSE, has a value of 0 for the third case.
  • The response count (10) and the observation count (100) for this case establish the initial estimate of the natural response rate.
  • Because the default log transformation is performed, the control group is not included in the analysis.

Example

DATA LIST FREE / SOLUTION DOSE NOBSN NRESP.
BEGIN DATA
1  5 100 20
1 10  80 30
1  0 100 10
  ...
END DATA.

PROBIT NRESP OF NOBSN BY SOLUTION(1,4) WITH DOSE
  /NATRES = 0.10.
  • This example reads four variables and requests an analysis in which the natural response rate is set to 0.10. The values of the control group are ignored.
  • The control group is excluded from the analysis because the default log transformation is performed.