In codebook: Automatic Codebooks from Metadata Encoded in Dataset Attributes

if (exists("testing")) {
    indent = '#' # ugly hack so _regression_summary can be "spun" (variables included via `r ` have to be available)
    results = data.frame()
    survey_repetition = 'single'
    reliabilities = list()
}

r indent# Survey overview

r ended(results) completed rows, r started who entered any information, r only_viewed only viewed the first page. There are r expired(results) expired rows (people who did not finish filling out in the requested time frame). In total, there are r nrow(results) rows including unfinished and expired rows.

There were r users unique participants, of which r finished_users finished filling out at least one survey.

r ifelse(survey_repetition == 'single', "This survey was not repeated.", ifelse(survey_repetition == 'repeated_once', paste0("This survey was repeated once by ", sum(rows_by_user == 2), " out of ", sum(rows_by_user <= 2), " users."), paste0("This survey was repeated many times, on average ", round(rows_per_user,2), " times per user.")))

r ifelse(survey_repetition != "single" && round(rows_per_user) == 1, "Oddly there seem to be almost no users with several rows. Maybe something went wrong and a few rows are duplicated by accident?", "")

if (survey_repetition != "single") {
    overview = results %>% dplyr::group_by(session) %>% 
        dplyr::summarise(
            n = sum(!is.na(session)),
            expired = sum(!is.na(expired)),
            ended = sum(!is.na(ended))
        ) %>% 
        tidyr::gather(key, value, -session)
    if (length(unique(dplyr::filter(overview, key == "expired")$value)) == 1) {
        overview = dplyr::filter(overview, key != "expired")
    }
    print(
        ggplot2::ggplot(overview, ggplot2::aes(value, ..count..)) + ggplot2::geom_bar() + ggplot2::facet_wrap(~ key, nrow = 1)
    )
}

The first session started on r min(results$created), the last session on r max(results$created).

ggplot2::qplot(results$created) + ggplot2::scale_x_datetime("Date/time when survey was started")

People took on average r round(mean(duration$duration),2) minutes (median r round(median(duration$duration),2)) to answer the survey.

ggplot2::qplot(duration$duration, binwidth = 0.5) + ggplot2::scale_x_continuous(paste("Duration (in minutes), excluding", high_vals, "values above median + 4*MAD and ", low_vals, "values below 0."), limits = c(lower_limit, upper_limit))