Statscraft 0.15.0, release v0.15.0
POWER PROPORTIONS ONESAMPLE
Power or sample size of a test of one proportion against a hypothesized value.
Syntax
POWER PROPORTIONS ONESAMPLE /PARAMETERS [TEST=NONDIRECTIONAL | DIRECTIONAL] [SIGNIFICANCE=0.05]
N=n | POWER=powers [NULL=0.5] ALTERNATIVE=p
[ESTIMATE=NORMAL | BINOMIAL] [CONTINUITY=FALSE | TRUE] [TIMER=minutes]
[/PLOT [N[(low high)]] [NULL] [ALTERNATIVE] [DIFF] [N_XAXIS[(low high)] DIFF_YAXIS]
[N_YAXIS[(low high)] DIFF_XAXIS] [NULL_XAXIS ALTERNATIVE_YAXIS]
[NULL_YAXIS ALTERNATIVE_XAXIS] [VROTATE(10)] [HROTATE(325)]].
Needs no data. NULL and ALTERNATIVE lie between 0 and 1, and differ when POWER is given.
NORMAL uses the normal approximation, with a continuity correction of 1 / (2 sqrt(n)) when
CONTINUITY=TRUE. BINOMIAL counts the exact binomial rejection region at NULL and its
probability at ALTERNATIVE, up to 500 trials; above that the normal approximation is used
with a warning. Because exact power rises in steps, the sample size search checks smaller
sizes and then counts upwards until the power is reached. TIMER stops that search after so
many minutes, with no result. A one-sided test looks in the direction of ALTERNATIVE. PLOT
charts power over N, over NULL or ALTERNATIVE (the other kept), over DIFF = ALTERNATIVE -
NULL with NULL kept, and surfaces over N and DIFF or NULL and ALTERNATIVE; proportion axes
run from 0.01 to 0.99.
Example
POWER PROPORTIONS ONESAMPLE /PARAMETERS NULL=0.3 ALTERNATIVE=0.42 POWER=0.8 0.9
ESTIMATE=BINOMIAL.
statscraft help POWER PROPORTIONS ONESAMPLE prints the same text in the terminal. All commands