Chapter Contents |
Previous |
Next |

INTERVALS Statement |

The following entries provide detailed descriptions of
*options* in the INTERVALS statement.

**ALPHA=***value-list*-
specifies values of , the probability or confidence
associated with the interval. For example, the following
statements tabulate the default
intervals at probability or confidence levels of
, , , and
:
proc capability data=steel; intervals width / alpha = 0.05 0.10 0.15 0.20; run;

Note that some references use to denote probability or confidence levels. Values for the ALPHA= option must be between 0.00001 to 0.99999. By default, values of 0.01, 0.05, and 0.10 are used. **K=***value-list*-
lists values of
*k*for prediction intervals. Default*values*of 1, 2, and 3 are used for the prediction interval for*k*future observations and for the prediction interval for the mean of*k*future observations. Default*values*of 2 and 3 are used for the prediction interval for the standard deviation of*k*future observations. The*values*must be integers. **METHODS=***indices***METHOD=***indices*-
specifies which intervals are computed. The
*indices*can range from 1 to 6, and they correspond to the intervals described in Table 6.2.**Table 6.2:**Intervals Computed for METHOD=*Index**Index***Statistical Interval**1 approximate simultaneous prediction interval for *k*future observations2 prediction interval for the mean of *k*future observations3 approximate statistical tolerance interval that contains at least proportion *p*of the population4 confidence interval for the population mean 5 prediction interval for the standard deviation of *k*future observations6 confidence interval for the population standard deviation

For example, the following statements tabulate confidence limits for the population mean (METHOD=4) and confidence limits for the population standard deviation (METHOD=6):proc capability data=steel; intervals width / methods=4 6; run;

Formulas for the intervals are given in "Methods for Computing Statistical Intervals". By default, the procedure computes all six intervals. **NOPRINT**-
suppresses the tables produced by default. This option
is useful when you only want to save the interval
information in an OUTINTERVALS= data set.
**OUTINTERVALS=***SAS-data-set***OUTINTERVAL=***SAS-data-set***OUTINT=***SAS-data-set*-
specifies an output SAS data set containing the
intervals and related information. For example,
the following statements create a data set named
INTS containing intervals for the variable WIDTH:
proc capability data=steel; intervals width / outintervals=ints; run;

See "OUTINTERVALS= Data Set" for details. **P=***value-list*-
lists values of
*p*for the tolerance intervals. These values must be between 0.00001 to 0.99999. Note that the P= option applies only to the tolerance intervals (METHODS=3). By default, values of 0.90, 0.95, and 0.99 are used. **TYPE=LOWER | UPPER | TWOSIDED**-
determines whether the intervals computed are one-sided
lower, one-sided upper, or two-sided intervals, respectively.
See "Computing One-Sided Lower Prediction Limits" for an example.
The default interval type is TWOSIDED.

Chapter Contents |
Previous |
Next |
Top |

Copyright © 1999 by SAS Institute Inc., Cary, NC, USA. All rights reserved.