PARAMETERS Subcommand (POWER PROPORTIONS INDEPENDENT command)

PARAMETERS controls the test assumptions. The keywords specify the sample size, total number of trials for groups, proportion parameters for groups, significance level, statistical tests, estimation method, and test direction.

POWER = value(s)
Optional keyword that specifies the desired power level(s). There is no default. Multiple values can be specified with each value separated by a blank space (or blank spaces). A grid of power values from value1 to value2 with the increment of value3, connected by the keywords TO and BY, can also be specified. When a grid of power values is specified, only one valid set of [value1 TO value2 BY value3] is allowed. Value1 and value2 must satisfy the condition 0 < value1 ≤ value2 < 1. The case where value1 = value2 is equivalent to specifying a single value1, regardless of value3. In order to restrict the number of power values, value3 must satisfy the condition 0.01 ≤ value3 < 1. The grid can be mixed with other power values, and can be placed anywhere.
Notes:
  • All specified values must be in the range [0,1].
  • In cases where duplicate values are specified, only unique values are used, and a warning message is generated to indicate that all duplicate values are excluded from the analysis.
TEST = {NONDIRECTIONAL}{DIRECTIONAL}
An optional keyword that determines whether the test is one or two-sided. NONDIRECTIONAL, or a two-sided test, is the default setting. When DIRECTIONAL is specified, power is computed for a one-sided test.
SIGNIFICANCE = value
An optional keyword that specifies the significance level of the Type I error rate for the test. The value must be a single double value ∈(0,1). The default level is 0.05.
N = (integer1 integer2)
A required keyword that specifies the total number of trials for two groups. Both integer values must be ≥ 1. When METHOD = T, integer1 + integer2 > 2.
NRATIO = value
An optional keyword that specifies the sample size ratios. Must be a single value ∈(0.01,100) with 1 as the default setting. The keyword takes effect when POWER is specified and is ignored when N is specified.
PROPORTIONS = (value1 value2)
An optional keyword that specifies the proportion parameters for two groups. Both values must be ∈(0,1). value1 and value2 cannot be the same when POWER is specified.
METHOD = {CHISQ} {T} {LRT} {FISHER}
An optional keyword that specifies the statistical test. CHISQ estimates the power based on the Pearson's chi-squared test (the default setting). T, LRT, and FISHER estimate the power based on the Student's t-test, likelihood ratio test, and Fisher's exact test, respectively. Note that the combination of METHOD = FISHER and ESTIMATE = BINOMIAL takes effect when the sum of integer1 and integer2 specified for /PARAMETERS N does not exceed 500 (otherwise CHISQ is used).
Note: In some cases, METHOD=FISHER may take an extended amount of time to complete.
ESTIMATE = {NORMAL} {BINOMIAL}
An optional keyword that determines the power estimation method, if the total sample size specified for /PARAMETERS N does not exceed 500. NORMAL calls for the normal approximation (the default setting). BINOMIAL calls for the binomial enumeration method. The keyword has no effect when the specified /PARAMETERS N value exceeds 500 (normal approximation is used).
TIMER = {value}
An optional keyword to specify the maximum number of minutes allowed to estimate the sample size. If specified, the value must be a single positive integer to denote the number of minutes. The keyword takes effect when ESTIMATE = BINOMIAL and POWER is specified. The keyword is not pasted by default, which turns the timer off completely. When the time limit is reached, the analysis is terminated and a warning message is displayed (no results are provided).
CONTINUITY = boolean
An optional keyword that controls whether or not continuity correction is used. FALSE is the default setting. Continuity correction is applied when TRUE is specified. The keyword takes effect when METHOD = CHISQ.
POOLED = boolean
An optional keyword that controls whether the estimation of the standard deviation is pooled or unpooled. TRUE is the default setting. When FALSE is specified, the unpooled standard deviation is calculated. The keyword takes effect when METHOD = CHISQ or METHOD = T.