Skip to contents

Extract surveys by periodicity from a rotating panel

Source: R/PanelSurvey.R
extract_surveys.Rd

Extracts subsets of surveys from a RotativePanelSurvey object based on temporal criteria. Allows obtaining surveys for different types of analysis (monthly, quarterly, annual) respecting the rotating panel's temporal structure.

Usage

extract_surveys(
  RotativePanelSurvey,
  index = NULL,
  monthly = NULL,
  annual = NULL,
  quarterly = NULL,
  biannual = NULL,
  use.parallel = FALSE
)

Arguments

RotativePanelSurvey

A RotativePanelSurvey object containing the rotating panel surveys organized temporally

index

Integer vector specifying survey indices to extract. If a single value, returns that survey; if a vector, returns a list

monthly

Integer vector specifying which months to extract for monthly analysis (1-12)

annual

Integer vector specifying which years to extract for annual analysis

quarterly

Integer vector specifying which quarters to extract for quarterly analysis (1-4)

biannual

Integer vector specifying which semesters to extract for biannual analysis (1-2)

use.parallel

Logical indicating whether to use parallel processing for intensive operations. Default FALSE

Value

A list of Survey objects matching the specified criteria, or a single Survey object if a single index is specified

Details

This function is essential for working with rotating panels because:

  • Enables periodicity-based analysis: Extract data for different types of temporal estimations

  • Preserves temporal structure: Respects temporal relationships between different panel waves

  • Optimizes memory: Only loads surveys needed for the analysis

  • Facilitates comparisons: Extract specific periods for comparative analysis

  • Supports parallelization: For operations with large data volumes

Extraction criteria are interpreted according to survey frequency:

  • For monthly ECH: monthly=c(1,3,6) extracts January, March and June

  • For annual analysis: annual=1 typically extracts the first available year

  • For quarterly analysis: quarterly=c(1,4) extracts Q1 and Q4

If no criteria are specified, the function returns the implantation survey with a warning.

See also

load_panel_survey for loading rotating panels get_implantation for obtaining implantation data get_follow_up for obtaining follow-up data workflow for using extracted surveys in analysis

Other panel-surveys: PoolSurvey, RotativePanelSurvey, get_follow_up(), get_implantation()

Examples

# Panels usually come from load_panel_survey(); here we build a
# small one in memory
mk <- function(edition) {
  Survey$new(
    data = data.table::data.table(id = 1:5, w = 1),
    edition = edition, type = "ech", psu = NULL,
    engine = "data.table", weight = add_weight(monthly = "w")
  )
}
panel <- RotativePanelSurvey$new(
  implantation = mk("2023-01-01"),
  follow_up = list(mk("2023-01-01"), mk("2023-02-01"), mk("2023-03-01")),
  type = "ech", default_engine = "data.table",
  steps = list(), recipes = list(), workflows = list(), design = NULL
)

# Extract follow-up surveys by index
ech_first <- extract_surveys(panel, index = 1)
ech_first_two <- extract_surveys(panel, index = c(1, 2))

if (FALSE) { # \dontrun{
# Not run: interval extraction needs a panel loaded from the real
# microdata files (editions must carry the survey time pattern)
ech_q1 <- extract_surveys(panel_ech, monthly = c(1, 2, 3))
ech_annual <- extract_surveys(panel_ech, annual = 1)
results <- workflow(
  survey = extract_surveys(panel_ech, quarterly = c(1, 2)),
  svymean(~unemployed, na.rm = TRUE),
  estimation_type = "quarterly"
)
} # }