Build a master banner (cross-tab) table

Produces the market-research "banner" table: one master table with a stack of question variables down the side (rows) and one or more grouping variables across the top (cols), plus an Overall column for the whole sample. Categorical questions become column-percentage blocks (each banner column sums to ~100 within the block); numeric questions become a mean / median / sd / quartile block. It generalises crosstab() from a single pair to a whole table, and picks the cell content per question automatically.

Usage

crosstab_banner(
  data = NULL,
  rows,
  cols,
  cell = c("auto", "pct", "count", "mean", "diff"),
  stats = c("mean", "median", "sd", "p25", "p75"),
  total = TRUE,
  digits = NULL,
  na_rm = TRUE,
  drop = NULL,
  weights = NULL,
  max_levels = 20,
  long = FALSE,
  flextable = FALSE,
  checkpoint = NULL,
  confirm = NULL
)

Arguments

data

A data frame. If omitted, the session default (use_dataset()) is used.

rows

Stub variables, down the side (tidyselect, e.g. c(satis_return, demo_edu) or starts_with("ratings_")). If omitted, every eligible variable is used (see max_levels and Details).

cols

Banner / grouping variables, across the top (tidyselect). Pass several to get several spanning column groups. If omitted, every eligible variable is used, giving a full variable-by-variable banner.

cell

What each body cell holds: "auto" (default) uses column percentages for categorical questions and a statistics block for numeric ones; "pct" forces column percentages; "count" uses (weighted) counts; "mean" forces a numeric mean (errors on a categorical question); "diff" shows each cell's difference from the Overall column (percentage points, or mean difference for numeric questions).

stats

For numeric questions, which statistics form the block. Any of "mean", "median", "sd", "p25", "p75". Default all five.

total

Include the whole-sample Overall column. Default TRUE.

digits

Decimal places. NULL (default) uses 0 for percentages / counts and 2 for numeric statistics; a value overrides both.

na_rm

Drop blanks / non-answers in the questions and banner variables. Default TRUE.

drop

Answer values to remove before tabulating (see drop_items()), passed to the underlying crosstab() calls. Defaults to the drop_answers option.

weights

Survey weighting: NULL (default) uses the session scheme from set_weights() if set; FALSE forces unweighted; or pass an ad-hoc scheme. When weighting is active the cells are weighted (weighted percentages, counts, means and quartiles).

max_levels

When rows / cols are omitted, the largest number of distinct answers a categorical variable may have to be used automatically (default 20). Bigger categorical variables, identifier columns and free-text are skipped; numeric scales are always kept as stub rows.

long

If TRUE, return the tidy long form (variable, item, group, group_item, value) instead of the wide master table. Default FALSE.

flextable

If TRUE, return a flextable with the two-row banner header ready for a slide or Word report. Default FALSE. Requires the suggested flextable package.

checkpoint

Save each question as it finishes, so re-running the same call picks up where an interrupted run stopped. TRUE keeps the file under tools::R_user_dir() and deletes it once the run completes; a file path puts it where you choose and leaves it there for you to manage. NULL (default) or FALSE writes nothing. See Details and clear_checkpoints().

confirm

When rows / cols are left to the automatic selection, show which variables were chosen and which were skipped, and wait for a yes before the run starts. NULL (default) follows the confirm option, which asks in an interactive session and never asks in a script; TRUE or FALSE force it either way. Answering no returns NULL and computes nothing.

Value

By default a wide tibble: variable, item, Overall, then one column per banner item. The banner grouping (which column belongs to which top-row variable) is stored in the "banner_spanners" attribute, which flextable = TRUE turns into a spanning two-row header. With long = TRUE, the tidy long form.

Details

A tibble has a single header row, so the two-row banner header (each grouping variable's name spanning its items) cannot be stored in the data itself: the wide return carries the grouping in the "banner_spanners" attribute and flextable = TRUE renders the real spanning header. Column percentages are computed within each banner column, so every column (Overall included) sums to about 100 down each question block. Registered orders (register_order()) set the item and banner-column ordering automatically. Passing the same selection to rows and cols gives the full every-question-by-every-question matrix (a variable is never crossed with itself). Supplying only the data frame does this automatically: every variable with at most max_levels distinct answers becomes both a stub and a banner group, numeric scales are added as stub statistics blocks, and identifier / free-text columns are skipped (and named in a message). That selection decides the whole table, so an interactive session prints it and waits for a yes before starting the run (see confirm). Check-all-that-apply blocks (columns sharing a prefix that each hold one option or blank, such as motivations_*) are recognised as a single multi-select stub question and tabulated with calc_percentage_multi(), whose base is the respondents who picked any option (so those rows can sum past 100). Multi-select blocks appear as stubs only, not banner groups, and are unweighted.

A full every-variable banner is one crosstab() per question per grouping variable, so a wide survey takes minutes rather than seconds. In an interactive session it reports which question it is on, with an estimate of the time left; ezrsurvey_options(progress = FALSE) turns that off, and scripts are silent by default. checkpoint = TRUE additionally saves each question as it completes, so re-running the identical call after an interruption resumes instead of starting again. The managed file is named after the run's own fingerprint, so a call finds its own interrupted run and a call with different data or arguments never sees it, and it is deleted as soon as the table is built. Pass a path instead if you would rather choose the location; that file is yours, so the package neither deletes it nor counts it as one of its own. clear_checkpoints() empties the managed folder if a run was interrupted and never repeated.

See also

crosstab() for a single pair, calc_percentage_batch() for stacked one-variable percentages, export_summary_xlsx(), register_order().

Other summaries: calc_percentage(), calc_percentage_batch(), calc_percentage_multi(), calc_summary(), clear_checkpoints(), crosstab()

Examples

# gender and region banner over two questions (column percentages)
crosstab_banner(podracing_survey,
                rows = c(satis_return, demo_edu),
                cols = c(demo_gender, region))
#> # A tibble: 12 × 11
#>    variable item  Overall Female  Male `Non-binary`  Asia Europe `Latin America`
#>    <chr>    <chr>   <dbl>  <dbl> <dbl>        <dbl> <dbl>  <dbl>           <dbl>
#>  1 demo_edu Bach…      25     26    25           22    23     28              15
#>  2 demo_edu Doct…       4      3     4            4     4      3               5
#>  3 demo_edu Lowe…      11     12    11           15    12     10              10
#>  4 demo_edu Mast…      13     12    13           19    11     15               9
#>  5 demo_edu Prim…       4      3     5            7     7      6               4
#>  6 demo_edu Shor…      13     14    12            4    19     11              17
#>  7 demo_edu Uppe…      30     30    31           30    25     26              41
#>  8 satis_r… Like…      27     28    27           15    24     28              30
#>  9 satis_r… Not …      24     25    22           33    25     25              21
#> 10 satis_r… Unli…      16     14    17           26    12     16              11
#> 11 satis_r… Very…      26     24    26           26    34     25              28
#> 12 satis_r… Very…       8      9     7            0     5      6              10
#> # ℹ 2 more variables: `North America` <dbl>, Oceania <dbl>

# numeric questions become a mean / median / sd / quartile block
crosstab_banner(podracing_survey,
                rows = c(nps_value, demo_age),
                cols = demo_gender)
#> # A tibble: 10 × 6
#>    variable  item   Overall Female  Male `Non-binary`
#>    <chr>     <chr>    <dbl>  <dbl> <dbl>        <dbl>
#>  1 demo_age  mean     32.6   32.9  32.3         31.2 
#>  2 demo_age  median   32     33    32           31   
#>  3 demo_age  sd       11.2   11.4  11.1         11.1 
#>  4 demo_age  p25      24     25    24           23.5 
#>  5 demo_age  p75      40     40    40           37   
#>  6 nps_value mean      7.56   7.51  7.61         7.56
#>  7 nps_value median    8      8     8            8   
#>  8 nps_value sd        1.76   1.81  1.72         1.69
#>  9 nps_value p25       6      6     7            7   
#> 10 nps_value p75       9      9     9            8   

# each cell as its difference from the Overall column
crosstab_banner(podracing_survey, rows = satis_return,
                cols = region, cell = "diff")
#> # A tibble: 5 × 8
#>   variable    item  Overall  Asia Europe `Latin America` `North America` Oceania
#>   <chr>       <chr>   <dbl> <dbl>  <dbl>           <dbl>           <dbl>   <dbl>
#> 1 satis_retu… Like…      27    -3      1               3              -1       4
#> 2 satis_retu… Not …      24     1      1              -3               0     -12
#> 3 satis_retu… Unli…      16    -4      0              -5               1       8
#> 4 satis_retu… Very…      26     8     -1               2              -2      -2
#> 5 satis_retu… Very…       8    -3     -2               2               0       2

# \donttest{
# full variable-by-variable banner: just pass the data frame
crosstab_banner(podracing_survey)
#> crosstab_banner: skipped 3 identifier / free-text / high-cardinality column(s): respondent_id, nps_com, show_com. Raise max_levels or name them in rows/cols to include them.
#> # A tibble: 129 × 126
#>    variable   item  Overall email in_app panel socials Female  Male `Non-binary`
#>    <chr>      <chr>   <dbl> <dbl>  <dbl> <dbl>   <dbl>  <dbl> <dbl>        <dbl>
#>  1 collector  email    28    NA     NA    NA      NA     29    29           33  
#>  2 collector  in_a…    16    NA     NA    NA      NA     15    17           26  
#>  3 collector  panel    35    NA     NA    NA      NA     35    34           33  
#>  4 collector  soci…    20    NA     NA    NA      NA     21    20            7  
#>  5 demo_age   mean     32.6  32.3   32.5  32.5    33.2   32.9  32.3         31.2
#>  6 demo_age   medi…    32    32     33    32      32     33    32           31  
#>  7 demo_age   sd       11.2  11.5   10.5  11.4    11.0   11.4  11.1         11.1
#>  8 demo_age   p25      24    24     25    23.5    25     25    24           23.5
#>  9 demo_age   p75      40    40     39    40      41     40    40           37  
#> 10 demo_gend… Fema…    42    42     39    43      44     NA    NA           NA  
#> # ℹ 119 more rows
#> # ℹ 116 more variables: `Bachelor or equivalent` <dbl>,
#> #   `Doctoral or equivalent` <dbl>, `Lower secondary` <dbl>,
#> #   `Master or equivalent` <dbl>, `Primary or less` <dbl>,
#> #   `Short-cycle tertiary` <dbl>, `Upper secondary` <dbl>, Australia <dbl>,
#> #   Brazil <dbl>, Canada <dbl>, France <dbl>, Germany <dbl>, Japan <dbl>,
#> #   Mexico <dbl>, Sweden <dbl>, `United Kingdom` <dbl>, …
# }