The PSMOOTH Procedure

PROC PSMOOTH Statement

PROC PSMOOTH <options> ;

You can specify the following options in the PROC PSMOOTH statement.

ADJUST=NONE ADJUST=BON | BONFERRONI ADJUST=FDR ADJUST=SIDAK

indicates which adjustment for multiple testing to apply to the set(s) of $p$ -values in the output data set. This adjustment is applied after any smoothing has occurred. ADJUST=NONE is the default.

BANDWIDTH=number-list BW=number-list

gives the values for the bandwidths to use in combining $p$ -values. A bandwidth of $w$ indicates that $w$ $p$ -values on each side of the original $p$ -value are included in the combining method to create a sliding window of size $2w+1$ . The number list can contain any combination of the following forms, with the forms separated by commas:

$w_1,w_2,\ldots ,w_ n$: a list of several values
$w_1$ to $w_2$: a sequence where $w_1$ is the starting value, $w_2$ is the ending value, and the increment is 1.
$w_1$ to $w_2$ by $i$: a sequence where $w_1$ is the starting value, $w_2$ is the ending value, and the increment is $i$ .

All numbers in the number list must be integers, and any negative numbers are ignored. An example of a valid number list is

   bandwidth = 1,2, 5 to 15 by 5, 18

which would perform the combining of $p$ -values by using bandwidths 1, 2, 5, 10, 15, and 18, which create sliding windows of size 3, 5, 11, 21, 31, and 37, respectively.

DATA=SAS-data-set

names the input SAS data set to be used by PROC PSMOOTH. The default is to use the most recently created data set.

FISHER

requests that Fisher’s method for combining $p$ -values from multiple hypotheses be applied to the original $p$ -values.

NEGLOG

requests that all $p$ -values, original and combined, be transformed to their negative log (base $e$ ) in the output data set; that is, for each $p$ -value, $-\log (p$ -value) is reported in the OUT= data set. This option is useful for graphing purposes.

NEGLOG10

requests that all $p$ -values, original and combined, be transformed to their negative log (base 10) in the output data set; that is, for each $p$ -value, $-\log _{10}(p$ -value) is reported in the OUT= data set. This option is useful for graphing purposes.

OUT=SAS-data-set

names the output SAS data set containing the original $p$ -values and the new combined $p$ -values. When this option is omitted, an output data set is created by default and named according to the DATAn convention.

SIMES

requests that Simes’ method for combining $p$ -values from multiple hypotheses be applied to the original $p$ -values.

TAU=number

indicates the value of $\tau$ to be used in the TPM. The significance level for the tests can be used as the value for number, although this is not the only possibility. The value of number must be greater than 0 and less than or equal to 1. By default, number is set to 0.05. This option is ignored if the TPM option is not specified.

TPM

requests that the TPM for combining $p$ -values from multiple hypotheses be applied to the original $p$ -values.