Unified sample-size and power interface for creel surveys

A single tidy entry point for pre-survey sample-size planning that wraps creel_n_effort(), creel_n_cpue(), and creel_power() and returns a consistent tibble.

Usage

power_creel(
  mode = c("effort_n", "cpue_n", "power"),
  target_rse = NULL,
  strata = NULL,
  N_h = NULL,
  ybar_h = NULL,
  s2_h = NULL,
  cv_catch = NULL,
  cv_effort = NULL,
  rho = 0,
  n = NULL,
  cv_historical = NULL,
  delta_pct = NULL,
  alpha = 0.05,
  alternative = c("two.sided", "one.sided")
)

Arguments

mode: Character scalar. One of "effort_n", "cpue_n", or "power". Selects the planning formula.
target_rse: Numeric scalar in (0, 1]. Target relative standard error (= target CV). Required for mode %in% c("effort_n", "cpue_n").
strata: Character vector of stratum names. Required for mode = "effort_n". Length must match N_h, ybar_h, and s2_h.
N_h: Numeric vector. Total sampling days available per stratum. Required for mode = "effort_n".
ybar_h: Numeric vector. Pilot mean effort per day per stratum. Required for mode = "effort_n".
s2_h: Numeric vector. Pilot variance of effort per day per stratum. Required for mode = "effort_n".
cv_catch: Numeric scalar. Pilot CV of catch per interview. Required for mode %in% c("cpue_n", "power").
cv_effort: Numeric scalar. Pilot CV of effort per interview. Required for mode = "cpue_n".
rho: Numeric scalar in [-1, 1]. Pilot correlation between catch and effort. Default 0 (conservative). Used for mode %in% c("cpue_n", "power").
n: Integerish scalar. Sample size (interviews) for mode = "power".
cv_historical: Numeric scalar. Historical CV of CPUE for mode = "power". If NULL, cv_catch is used as a proxy.
delta_pct: Numeric scalar (> 0). Fractional change to detect. Required for mode = "power".
alpha: Numeric scalar in (0, 0.5]. Type I error rate. Default 0.05. Used for mode = "power".
alternative: Character. "two.sided" (default) or "one.sided". Used for mode = "power".

Value

A tibble (data frame) with columns varying by mode:

mode = "effort_n" (one row per stratum plus a "total" row):

stratum: Stratum name.
n_required: Sampling days required.
target_rse: The requested target RSE.

mode = "cpue_n" (one row):

n_required: Interviews required.
target_rse: The requested target RSE.
cv_catch: Input CV of catch.
cv_effort: Input CV of effort.
rho: Input correlation.

mode = "power" (one row):

power: Estimated statistical power.
n: Input sample size.
delta_pct: Input fractional change.
cv_historical: Historical CV used.
alpha: Input significance level.
alternative: Input test direction.

Details

Three mode values are supported:

"effort_n": Required sampling days per stratum to achieve target_rse on the effort estimate (calls creel_n_effort()).
"cpue_n": Required interviews to achieve target_rse on the CPUE estimate (calls creel_n_cpue()).
"power": Statistical power to detect a fractional change in CPUE at a given sample size (calls creel_power()).

Examples

# Effort: sampling days needed for 20 percent RSE
power_creel(
  mode       = "effort_n",
  target_rse = 0.20,
  strata     = c("weekday", "weekend"),
  N_h        = c(65, 28),
  ybar_h     = c(50, 60),
  s2_h       = c(400, 500)
)
#>   stratum n_required target_rse
#> 1 weekday          3        0.2
#> 2 weekend          2        0.2
#> 3   total          4        0.2

# CPUE: interviews needed for 20 percent RSE
power_creel(
  mode       = "cpue_n",
  target_rse = 0.20,
  cv_catch   = 0.8,
  cv_effort  = 0.5
)
#>   n_required target_rse cv_catch cv_effort rho
#> 1         23        0.2      0.8       0.5   0

# Power: detect a 20 percent change with n = 80 interviews
power_creel(
  mode          = "power",
  n             = 80L,
  cv_historical = 0.5,
  delta_pct     = 0.20
)
#>       power  n delta_pct cv_historical alpha alternative
#> 1 0.7156166 80       0.2           0.5  0.05   two.sided