Generate Methods Section for a Causal Inference Study

This function generates a comprehensive methods section for a research paper, including details about the sample, variables, causal interventions, identification assumptions, target population, eligibility criteria, confounding control, missing data handling, statistical estimators, and additional sections like sensitivity analysis and scope of interventions.

Usage

boilerplate_report_methods(
  exposure_var,
  outcome_vars,
  n_total,
  baseline_wave,
  exposure_wave,
  outcome_wave,
  baseline_missing_data_proportion,
  sections_to_include = "all",
  appendices_measures = NULL,
  ...
)

Arguments

exposure_var

Character string specifying the primary exposure variable.

outcome_vars

A named list of character vectors specifying the outcome variables by domain.

n_total

Numeric value indicating the total number of participants in the study.

baseline_wave

Character string specifying the baseline wave of the study.

exposure_wave

Character string specifying the exposure wave of the study.

outcome_wave

Character string specifying the outcome wave of the study.

baseline_missing_data_proportion

Numeric value indicating the proportion of missing data at baseline.

sections_to_include

Character vector specifying which sections to include in the output. Default is 'all', which includes all available sections. Use 'list' to see available sections.

appendices_measures

Character string specifying the appendix where detailed variable descriptions can be found.

...

Additional arguments to be passed to helper functions.

measure_data

A list containing information about each measure.

Value

If sections_to_include is not 'list', returns a character string containing the generated methods section in markdown format. If sections_to_include is 'list', returns a character vector of available sections.

Examples

if (FALSE) { # \dontrun{
# Define outcomes by domain
outcomes_health <- c("smoker_binary", "hlth_bmi", "log_hours_exercise")
outcomes_psychological <- c("hlth_fatigue", "kessler_latent_anxiety")
outcomes_social <- c("belong", "neighbourhood_community")

# Combine all outcomes into a single list
all_outcomes <- list(
  health = outcomes_health,
  psychological = outcomes_psychological,
  social = outcomes_social
)

# Define the exposure variable
exposure_var <- "political_conservative"

# Load your measure_data
measure_data <- readRDS(here::here("boilerplate", "data", "measure_data.rds"))

# Generate methods text
methods_text <- boilerplate_report_methods(
  exposure_var = exposure_var,
  outcome_vars = all_outcomes,
  measure_data = measure_data,
  n_total = 47000,
  baseline_wave = "NZAVS time 10, years 2018-2019",
  exposure_wave = "NZAVS time 11, years 2019-2020",
  outcome_wave = "NZAVS time 12, years 2020-2021",
  baseline_missing_data_proportion = 0.15,
  appendices_measures = "C",
  causal_interventions = list(interventions = c("Increase exposure_var", "Do not change exposure_var")),
  contrasts = "null",
  null_intervention = "Do not change exposure_var",
  sample = list(appendices = "A-C"),
  statistical_estimator = list(estimators = c("lmtp", "grf")),
  inclusion_criteria = c(
    "Enrolled in the 2018 wave of the New Zealand Attitudes and Values Study (NZAVS time 10).",
    "Missing covariate data at baseline was permitted, and the data was subjected to imputation methods to reduce bias."
  ),
  exclusion_criteria = c(
    "Did not answer the political conservative question at NZAVS time 10 and time 11."
  ),
  n_participants = 32451,
  confounding_control = list(appendix_ref = "B", protocol_url = "https://osf.io/ce4t9/"),
  additional_sections = list(
    sensitivity_analysis = list(
      description = "We use the E-value method to assess sensitivity to unmeasured confounding."
    ),
    scope_interventions = list(figure_ref = "@fig-custom-hist"),
    evidence_change = list(table_ref = "@tbl-custom-transition")
  )
)

# Print the result
cat(methods_text)
} # }