survey: Survey Simulations
In peterdonati/hhss: Hunter Harvest Survey Simulations

Description Usage Arguments Details Value Examples

The survey functions take an output from pop and simulate survey responses based on the method specified.

census() creates a simulation where all hunters report, successful or not. This can mimic both mandatory reporting for all hunters as well as voluntary reporting. Follow up surveys can be completed on the non-responding portion of the population by simple random sampling.
mand() creates a simulation where only successful hunters report and a follow up sample of non-respondents can be taken through a simple random sample.
simple() creates a simulation where a population of hunters are surveyed using a simple random sample. Follow up survey is completed by following up with all non-respondents from the initial sample.

census(
  x,
  resp,
  bias,
  fol = FALSE,
  fol_sample = NULL,
  fol_scale = NULL,
  times = 1
)

mand(
  x,
  resp,
  fol = FALSE,
  bias = NULL,
  fol_sample = NULL,
  fol_scale = NULL,
  times = 1
)

simple(x, sample, resp, bias, fol = FALSE, fol_scale = NULL, times = 1)

`x`	An output from `pop`. It is the population to simulate response for.
`resp`	Probability/probabilities of response. In `simple()` and `census()` it defines response probabilities for unsuccessful hunters. In `mand()` it defines response probabilities for initial reporting, and then response probabilities for unsuccessful hunters in follow up samples.
`bias`	Scales the value(s) supplied to `resp` to create response probabilities for successful hunters. Introduces response bias for any value not equal to 1.
`fol`	Logical. If `TRUE`, a single follow up survey will be simulated.
`fol_sample`	Probability that a non-respondent is sampled for a follow up survey.
`fol_scale`	Scales initial response probabilities, creating new probabilities of response for follow up surveys.
`times`	The number of times to repeat the simulation.
`sample`	Probability a hunter is sampled for a survey

More than one value can be supplied to resp and bias. These functions automatically create a full factorial design on these two arguments.

If any scaling arguments scale probabilities to be > 1, the probabilities will silently be limited to 1.

A list of class survsim_census, survsim_mand, or survsim_simple where the length is equal to the integer supplied to times. The ultimate elements are data frames that will contain some, but not all, of these variables:

N: The population size.
true_harvest: The sum of harvests from the population.
resp_bias: The response bias simulated.
init_rate: The response probability for hunters to initially report harvest. Only reported in mand() outputs.
init_uns_rate: The probability at which a hunter responded to an initial survey if they were unsuccessful in harvesting.
init_suc_rate: The probability at which a hunter responded to an initial survey if they were successful in harvesting.
init_sample: The sum of hunters sampled in initial survey.
init_resp: The sum of responses to intial survey.
init_yes: The sum of initial responses that were from hunters who harvested.
init_no: The sum of initial responses that were from hunters who did not harvest.
fol_sample: The sum of hunters sampled for follow up.
fol_uns_rate: The probability at which a hunter responded to a follow up survey if they did not harvest.
fol_suc_rate: The probability at which a hunter responded to a follow up survey if they harvested.
fol_resp: The sum of hunters that responded to the follow up.
fol_yes: The sum of follow up responses that were from hunters who harvested.
fol_no: The sum of follow up responses that were from hunters who did not harvest.

# First, create a population:
my_pop <- pop(N = 1000, split = 0.7, success1 = 0.25, success2 = 0.6)

# Simulate a simple random sample from that population:
simple(
  my_pop,
  sample = 0.4,
  resp = 0.3,
  bias = 1,
  times = 10
)

# Multiple values can be passed to 'resp' and 'bias' arguments to create
# simulations for each unique pairing of the two:
census(
  my_pop,
  resp = seq(0.3, 0.8, 0.1),
  bias = c(1, 1.1, 1.2),
  fol = TRUE,
  fol_sample = 0.4,
  fol_scale = 0.7,
  times = 10
)