Introduction to tabxplor

library(tabxplor)
set_color_style(type = "text", theme = "light")

tabxplor tries to make it easy to deal with multiple cross-tables: to create and manipulate them, but also to read them, using color helpers to highlight important informations, and simplify your experience of data exploration. All functions are tidyverse propelled, pipe-friendly, and render tibble data frames which can be easily manipulated with dplyr. In the same time, time-taking operations are done with data.table to go faster with big dataframes. Tables can be exported to Excel and in html with formats and colors.

Base usage: cross-tables with color helpers for data exploration

The main functions are made to be user-friendly and time-saving in data analysis workflows.

tab makes a simple cross-table:

tab(forcats::gss_cat, marital, race)
#> # A tabxplor tab: 7 × 5
#>   marital       Other Black  White  Total
#>   <fct>           <n>   <n>    <n>    <n>
#> 1 No answer         2     2     13     17
#> 2 Never married   633 1 305  3 478  5 416
#> 3 Separated       110   196    437    743
#> 4 Divorced        212   495  2 676  3 383
#> 5 Widowed          70   262  1 475  1 807
#> 6 Married         932   869  8 316 10 117
#> 7 Total         1 959 3 129 16 395 21 483

When one of the row or column variables is numeric, tab calculates means by category of the other variable.

tab comes with options to weight the table, print percentages, manage totals, digits and missing values, add legends, gather rare categories in a “Others” level.

tab(forcats::gss_cat, marital, race, pct = "row", na = "drop", 
rare_to_other = TRUE, n_min = 1000, other_level = "Custom_other_level_name")

When a third variable is provided, tab makes a table with as many subtables as it has levels. With several tab_vars, it makes a subtable for each combination of their levels. The result is grouped: in dplyr, operations like sum() or all() are done within each subtable, and not for the whole dataframe.

Colors may be added to highlight over-represented and under-represented cells, and therefore help the user read the table. By default, with color = "diff", colors are based on the differences between a cell and it’s related total (which only works with means and row or col pct). When a percentage is superior to the average percentage of the line or column, it appears with shades of green (or blue). When it’s inferior, it appears with shades of red/orange. A color legend is added below the table. In RStudio colors are adapted to the theme, light or dark.

data <- forcats::gss_cat %>% 
  dplyr::filter(year %in% c(2000, 2006, 2012), !marital %in% c("No answer", "Widowed"))
gss  <- "Source: General social survey 2000-2014"
gss2 <- "Source: General social survey 2000, 2006 and 2012"
tab(data, race, marital, year, subtext = gss2, pct = "row", color = "diff")
race Never married Separated Divorced Married Total
Other 36% 5% 12% 47% 100%
Black 41% 11% 16% 32% 100%
White 25% 3% 18% 54% 100%
Total 2000 28% 4% 17% 50% 100%
Other 29% 7% 11% 53% 100%
Black 47% 6% 18% 30% 100%
White 22% 3% 19% 57% 100%
Total 2006 26% 4% 18% 52% 100%
Other 37% 6% 7% 51% 100%
Black 43% 5% 21% 31% 100%
White 25% 3% 18% 53% 100%
Total 2012 29% 4% 18% 50% 100%
Total Ensemble 27% 4% 18% 51% 100%
marital: x > tot +5%; +10%; +20%; +30%; x < tot -5%; -10%; -20%; -30%
Source: General social survey 2000, 2006 and 2012

The sup_cols argument adds supplementary column variables to the table. With numeric variables, it calculates the mean for each category or the row variable. With text variables, only the first level is kept (you can choose which one to use by placing it first with forcats::fct_relevel). Use tab_many to keep all levels.

tab(dplyr::storms, category, status, sup_cols = c("pressure", "wind"))
category disturbance extratropical hurricane other low subtropical depression subtropical storm tropical depression tropical storm tropical wave Total pressure wind
1 0 0 2 548 0 0 0 0 0 0 2 548 981 71
2 0 0 993 0 0 0 0 0 0 993 967 89
3 0 0 593 0 0 0 0 0 0 593 955 104
4 0 0 553 0 0 0 0 0 0 553 940 122
5 0 0 116 0 0 0 0 0 0 116 918 146
NA 171 2 151 0 1 453 151 298 3 569 6 830 111 14 734 1 002 38
Total 171 2 151 4 803 1 453 151 298 3 569 6 830 111 19 537 993 50

References and comparison levels for colors

By default, to calculate colors, each cell is compared to the subtable’s related total.

When a third variable or more are provided, it’s possible to compare with the general total line instead, by setting comp = "all". Here, only the last total row is highlighted (TOTAL ENSEMBLE appears in white but other total rows in grey).

tab(data, race, marital, year, subtext = gss2, pct = "row", color = "diff", comp = "all")
race Never married Separated Divorced Married Total
Other 36% 5% 12% 47% 100%
Black 41% 11% 16% 32% 100%
White 25% 3% 18% 54% 100%
Total 2000 28% 4% 17% 50% 100%
Other 29% 7% 11% 53% 100%
Black 47% 6% 18% 30% 100%
White 22% 3% 19% 57% 100%
Total 2006 26% 4% 18% 52% 100%
Other 37% 6% 7% 51% 100%
Black 43% 5% 21% 31% 100%
White 25% 3% 18% 53% 100%
Total 2012 29% 4% 18% 50% 100%
Total Ensemble 27% 4% 18% 51% 100%
marital: x > tot +5%; +10%; +20%; +30%; x < tot -5%; -10%; -20%; -30%
Source: General social survey 2000, 2006 and 2012

With diff = "first", each row (or column) is compared to the first row (or column), which is particularly helpful to highlight historical evolutions. The first rows then appears in white (while rows totals are themselves colored like normal lines).

data <- data %>% dplyr::mutate(year = as.factor(year))
tab(data, year, marital, race, pct = "row", color = "diff", diff = "first", tot = "col",
    totaltab = "table")
year Never married Separated Divorced Married Total
2000 36% 5% 12% 47% 100%
2006 29% 7% 11% 53% 100%
2012 37% 6% 7% 51% 100%
2000 41% 11% 16% 32% 100%
2006 47% 6% 18% 30% 100%
2012 43% 5% 21% 31% 100%
2000 25% 3% 18% 54% 100%
2006 22% 3% 19% 57% 100%
2012 25% 3% 18% 53% 100%
2000 28% 4% 17% 50% 100%
2006 26% 4% 18% 52% 100%
2012 29% 4% 18% 50% 100%
marital: x > x1 +5%; +10%; +20%; +30%; x < x1 -5%; -10%; -20%; -30%

When diff is a number, the nth row (or column) is used for comparison.

tab(data, year, marital, race, pct = "row", color = "diff", diff = 3)
year Never married Separated Divorced Married Total
2000 36% 5% 12% 47% 100%
2006 29% 7% 11% 53% 100%
2012 37% 6% 7% 51% 100%
Total Other 32% 6% 11% 51% 100%
2000 41% 11% 16% 32% 100%
2006 47% 6% 18% 30% 100%
2012 43% 5% 21% 31% 100%
Total Black 44% 7% 18% 31% 100%
2000 25% 3% 18% 54% 100%
2006 22% 3% 19% 57% 100%
2012 25% 3% 18% 53% 100%
Total White 23% 3% 19% 55% 100%
Total Ensemble 27% 4% 18% 51% 100%
marital: x > x3 +5%; +10%; +20%; +30%; x < x3 -5%; -10%; -20%; -30%

Finally, when diff is a string, it it used as a regular expression, to match with the names of the rows (or columns).

tab(data, year, marital, race, pct = "col", tot = "row", color = "diff", diff = "Married")
year Never married Separated Divorced Married
2000 20% 14% 20% 16%
2006 56% 68% 66% 63%
2012 24% 19% 13% 20%
Total Other 100% 100% 100% 100%
2000 29% 47% 27% 32%
2006 49% 38% 46% 46%
2012 22% 14% 27% 22%
Total Black 100% 100% 100% 100%
2000 34% 33% 31% 31%
2006 44% 44% 48% 49%
2012 23% 24% 21% 21%
Total White 100% 100% 100% 100%
Total Ensemble 100% 100% 100% 100%
marital: x > Married +5%; +10%; +20%; +30%; x < Married -5%; -10%; -20%; -30%

Confidence intervals

It it possible to print confidence intervals for each cell:

tab(forcats::gss_cat, race, marital, pct = "row", ci = "cell")
race No answer Never married Separated Divorced Widowed Married Total
Other 0%\(_{±0.3}\) 32%\(_{±2.1}\) 6%\(_{±1.1}\) 11%\(_{±1.5}\) 4%\(_{±0.9}\) 48%\(_{±2.2}\) 100%
Black 0%\(_{±0.2}\) 42%\(_{±1.7}\) 6%\(_{±0.9}\) 16%\(_{±1.3}\) 8%\(_{±1.0}\) 28%\(_{±1.6}\) 100%
White 0%\(_{±0.1}\) 21%\(_{±0.6}\) 3%\(_{±0.3}\) 16%\(_{±0.6}\) 9%\(_{±0.4}\) 51%\(_{±0.8}\) 100%
Total 0% 25% 3% 16% 8% 47% 100%

It is also possible to use confidence intervals to enhance colors helpers. With color = "diff_ci", the cells are only colored if the confidence interval of the difference between them and their reference cell (in total or first row/col) is superior to the difference itself. Otherwise, it means the cell is not significantly different from it’s reference in the total (or first) row: it turns grey, and the reader is not anymore tempted to over-interpret the difference.

tab(forcats::gss_cat, race, marital, pct = "row", color = "diff_ci")
race No answer Never married Separated Divorced Widowed Married Total
Other 0% 32% 6% 11% 4% 48% 100%
Black 0% 42% 6% 16% 8% 28% 100%
White 0% 21% 3% 16% 9% 51% 100%
Total 0% 25% 3% 16% 8% 47% 100%
marital: |x-tot|>ci & x > tot +5%; +10%; +20%; +30%; |x-tot|>ci & x < tot -5%; -10%; -20%; -30%

Finally, another calculation appears helpful: the difference between the cell and the total, minus the confidence interval of this difference (or in other word, what remains of that difference after having subtracted the confidence interval). ci = "after_ci" highligths all the cells whose value is significantly different from the relative total (or first cell). This is particularly useful when working on small populations: we can see at a glance which numbers we have right to read and interpret.

tab(forcats::gss_cat, race, marital, subtext = gss, pct = "row", color = "after_ci")
race No answer Never married Separated Divorced Widowed Married Total
Other 0% 32% 6% 11% 4% 48% 100%
Black 0% 42% 6% 16% 8% 28% 100%
White 0% 21% 3% 16% 9% 51% 100%
Total 0% 25% 3% 16% 8% 47% 100%
marital: |x-tot| > ci +0%; +5%; +15%; +25%; |x-tot| > ci -0%; -5%; -15%; -25%
Source: General social survey 2000-2014

Chi2 stats and contributions of cells to variance

chi2 = TRUE add summary statistics made in the chi2 metric: degrees of freedom (df), unweighted count, pvalue and (sub)table’s variance. Chi2 pvalue is colored in green when inferior to 5%, and in red when superior or equal to 5%, meaning that the table is not significantly different from the independent hypothesis (the two variables may be independent).

tab(forcats::gss_cat, race, marital, chi2 = TRUE)
race No answer Never married Separated Divorced Widowed Married Total
Other 2 633 110 212 70 932 1 959
Black 2 1 305 196 495 262 869 3 129
White 13 3 478 437 2 676 1 475 8 316 16 395
Total 17 5 416 743 3 383 1 807 10 117 21 483

Chi2 stats can also be used to color cells based on their contributions to the variance of the (sub)table, with color = "contrib". By default, only the cells whose contribution is superior to the mean contribution are colored. It highlights the cells which would stand out in a correspondence analysis (the two related categories would be located at the edges of the first axes ; here, being black is associated with never married and being separated).

tab(forcats::gss_cat, race, marital, color = "contrib")
race No answer Never married Separated Divorced Widowed Married Total
Other 2 633 110 212 70 932 1 959
Black 2 1 305 196 495 262 869 3 129
White 13 3 478 437 2 676 1 475 8 316 16 395
Total 17 5 416 743 3 383 1 807 10 117 21 483
marital: contrib > mean_ctr ×1; ×2; ×5; ×10; contrib > mean_ctr ×1; ×2; ×5; ×10

Combine tabxplor and dplyr

The result of tab is a tibble::tibble dataframe with class tab. It gets it’s own printing methods but, in the same time, can be transformed using most dplyr verbs, like a normal tibble.

library(dplyr)
tab(storms, category, status, sup_cols = c("pressure", "wind")) %>%
  filter(category != "-1") %>%
dplyr::select(-`tropical depression`)
  arrange(is_totrow(.), desc(category)) # use is_totrow to keep total rows

With dplyr::arrange, don’t forget to keep the order of tab variables and total rows:

tab(data, race, marital, year, pct = "row") %>%
  arrange(year, is_totrow(.), desc(Married))

Draw more complex tables with tab_many

tab is a wrapper around the more powerful function tab_many, which can be used to customize your tables.

It’s possible, for example, to make a summary table of as many columns variables as you want (showing all levels, or showing only one specific level like here):

library(dplyr)
first_lvs <- c("Married", "$25000 or more", "Strong republican", "Protestant")
data <- forcats::gss_cat %>% mutate(across(
  where(is.factor),
  ~ forcats::fct_relevel(., first_lvs[first_lvs %in% levels(.)])
))
tabs <- tab_many(data, race, c(marital, rincome, partyid, relig, age, tvhours),
         levels = "first", pct = "row", chi2 = TRUE, color = "auto")
tabs
#> chi2 stats      marital    rincome    partyid      relig
#> df                    4          4          4          4
#> variance         0.0259     0.0033     0.0214     0.0630
#> pvalue               0%         0%         0%         0%
#> count            21 483     21 483     21 483     21 483
#> 
#> # A tabxplor tab: 4 × 8
#>   race   Married `$25000 or more` `Strong republican` Protestant  Total    age
#>   <fct>   <row%>           <row%>              <row%>     <row%> <row%> <mean>
#> 1 Other      48%              32%                  4%        20%   100%     39
#> 2 Black      28%              28%                  2%        73%   100%     44
#> 3 White      51%              36%                 13%        50%   100%     49
#> 4 Total      47%              34%                 11%        50%   100%     47
#> # ℹ 1 more variable: tvhours <mean>
#> # marital, rincome, partyid: x > tot +5%; +10%; +20%; +30%; x < tot -5%; -10%; -20%; -30%
#> # age, tvhours             : x > tot ×1.15; ×1.5; ×2; ×4; x < tot /1.15; /1.5; /2; /4

Using tab or tab_many with purrr::map and tibble::tribble, you can program several tables with different parameters all at once, in a readable way:

tabs <-
  purrr::pmap(
    tibble::tribble(
      ~row_var, ~col_vars       , ~pct , ~filter              , ~subtext               ,
      "race"  , "marital"       , "no" , NULL                 , "Source: GSS 2000-2014",
      "race"  , "marital"       , "row", NULL                 , "Source: GSS 2000-2014",
      "race"  , "marital"       , "col", NULL                 , "Source: GSS 2000-2014",
      "relig" , c("race", "age"), "row", "year %in% 2000:2010", "Source: GSS 2000-2010",
      "relig" , c("race", "age"), "row", "year %in% 2010:2014", "Source: GSS 2010-2014",
      NA_character_, "race"     , "no" , NULL                 , "Source: GSS 2000-2014",
    ),
    .f = tab_many,
    data = forcats::gss_cat, color = "auto", chi2 = TRUE)

Export to html or Excel

To export a table to html with colors, like most of them in the current vignette, tabxplor uses knitr::kable and kableExtra. In this format differences from totals, confidence intervals, contribution to variance, and unweighted counts, are available in a tooltip at cells hover.

tabs %>% tab_kable()
race Married $25000 or more Strong republican Protestant Total age tvhours
Other 48% 32% 4% 20% 100% 39 3
Black 28% 28% 2% 73% 100% 44 4
White 51% 36% 13% 50% 100% 49 3
Total 47% 34% 11% 50% 100% 47 3
marital, rincome, partyid: x > tot +5%; +10%; +20%; +30%; x < tot -5%; -10%; -20%; -30%
age, tvhours : x > tot ×1.15; ×1.5; ×2; ×4; x < tot /1.15; /1.5; /2; /4

To print an html table by default (for example, in RStudio viewer), use tabxplor options:

options(tabxplor.print = "kable") # default to options(tabxplor.print = "console")

tab_xl exports any table or list of tables to Excel, with all colors, chi2 stats and formatting. On Excel, it is still possible to do calculations on raw numbers.

tabs %>% tab_xl(replace = TRUE, sheets = "unique")

Programming with tabxplor

When not doing data analysis but writing functions, you can use the sub-functions of tab_many step by step to attain more flexibility or speed. That way, it’s possible to write new functions to customize your tables even more.

data <- dplyr::starwars %>%
tab_prepare(sex, hair_color, gender, rare_to_other = TRUE,
n_min = 5, na_drop_all = sex)

data %>%
tab_plain(sex, hair_color, gender, tot = c("row", "col"), pct = "row", comp = "all") %>%
tab_ci("diff", color = "after_ci")  %>%
tab_chi2(calc = "p")

The whole architecture of tabxplor is powered by a special vector class, named tabxplor_fmt for formatted numbers. As a vctrs::record, it stores behind the scenes all the data necessary to calculate printed results, formats and colors. A set of functions are available to access or transform this data. ?fmt to get more information.

The simple way to recover the underlying numbers as numeric vectors is get_num, which extract the currently displayed field whatever it is :

tabs <- tab(data, race, marital, year, pct = "row")
tabs %>% mutate(across(where(is_fmt), get_num))
#> # A tabxplor tab: 33 × 9
#> # Groups:         year [9]
#>    year    race           Married `No answer` `Never married` Separated Divorced
#>    <fct>   <fct>            <dbl>       <dbl>           <dbl>     <dbl>    <dbl>
#>  1 2000    Other            0.446    0.00571            0.343    0.0457   0.114 
#>  2 2000    Black            0.282    0                  0.366    0.100    0.140 
#>  3 2000    White            0.488    0                  0.224    0.0276   0.163 
#>  4 2000    Total 2000       0.454    0.000355           0.253    0.0398   0.157 
#> 
#>  5 2002    Other            0.497    0                  0.281    0.0539   0.156 
#>  6 2002    Black            0.283    0                  0.398    0.0634   0.161 
#>  7 2002    White            0.489    0                  0.228    0.0279   0.161 
#>  8 2002    Total 2002       0.459    0                  0.256    0.0347   0.161 
#> 
#>  9 2004    Other            0.557    0                  0.289    0.0398   0.0846
#> 10 2004    Black            0.332    0                  0.403    0.0716   0.133 
#> 11 2004    White            0.556    0                  0.183    0.0269   0.156 
#> 12 2004    Total 2004       0.526    0                  0.220    0.0338   0.148 
#> 
#> 13 2006    Other            0.510    0.00169            0.279    0.0676   0.110 
#> 14 2006    Black            0.273    0                  0.426    0.0552   0.161 
#> 15 2006    White            0.516    0.00152            0.196    0.0247   0.172 
#> 16 2006    Total 2006       0.481    0.00133            0.239    0.0346   0.162 
#> 
#> 17 2008    Other            0.437    0                  0.372    0.0601   0.0765
#> 18 2008    Black            0.302    0                  0.416    0.0890   0.128 
#> 19 2008    White            0.518    0.00321            0.222    0.0218   0.148 
#> 20 2008    Total 2008       0.480    0.00247            0.262    0.0346   0.139 
#> 
#> 21 2010    Other            0.377    0                  0.404    0.0601   0.109 
#> 22 2010    Black            0.215    0                  0.511    0.0257   0.174 
#> 23 2010    White            0.487    0.000645           0.214    0.0297   0.172 
#> 24 2010    Total 2010       0.436    0.000489           0.276    0.0318   0.167 
#> 
#> 25 2012    Other            0.490    0                  0.357    0.0561   0.0663
#> 26 2012    Black            0.282    0                  0.399    0.0432   0.196 
#> 27 2012    White            0.487    0                  0.227    0.0298   0.166 
#> 28 2012    Total 2012       0.456    0                  0.266    0.0344   0.161 
#> 
#> 29 2014    Other            0.427    0                  0.347    0.0458   0.141 
#> 30 2014    Black            0.251    0.00518            0.433    0.0492   0.176 
#> 31 2014    White            0.502    0.00106            0.221    0.0265   0.162 
#> 32 2014    Total 2014       0.456    0.00158            0.266    0.0319   0.162 
#> 
#> 33 Ensemb… Total Ensemble   0.471    0.000791           0.252    0.0346   0.157 
#> # ℹ 2 more variables: Widowed <dbl>, Total <dbl>

To render character vectors (without colors), use format:

tabs %>% mutate(across(where(is_fmt), format))

The following fields compose any fmt column (though many can be NA if not calculated) :

vctrs::vec_data(tabs$Married)
#>        n display digits wn       pct mean         diff ctr var ci in_totrow
#> 1     78     pct      0 NA 0.4457143   NA -0.007959836  NA  NA NA     FALSE
#> 2    121     pct      0 NA 0.2820513   NA -0.171622839  NA  NA NA     FALSE
#> 3   1079     pct      0 NA 0.4875734   NA  0.033899308  NA  NA NA     FALSE
#> 4   1278     pct      0 NA 0.4536741   NA  0.000000000  NA  NA NA      TRUE
#> 5     83     pct      0 NA 0.4970060   NA  0.038054813  NA  NA NA     FALSE
#> 6    116     pct      0 NA 0.2829268   NA -0.176024346  NA  NA NA     FALSE
#> 7   1070     pct      0 NA 0.4890311   NA  0.030079903  NA  NA NA     FALSE
#> 8   1269     pct      0 NA 0.4589512   NA  0.000000000  NA  NA NA      TRUE
#> 9    112     pct      0 NA 0.5572139   NA  0.031253760  NA  NA NA     FALSE
#> 10   125     pct      0 NA 0.3315650   NA -0.194395184  NA  NA NA     FALSE
#> 11  1242     pct      0 NA 0.5559534   NA  0.029993276  NA  NA NA     FALSE
#> 12  1479     pct      0 NA 0.5259602   NA  0.000000000  NA  NA NA      TRUE
#> 13   302     pct      0 NA 0.5101351   NA  0.028982142  NA  NA NA     FALSE
#> 14   173     pct      0 NA 0.2728707   NA -0.208282331  NA  NA NA     FALSE
#> 15  1695     pct      0 NA 0.5161389   NA  0.034985862  NA  NA NA     FALSE
#> 16  2170     pct      0 NA 0.4811530   NA  0.000000000  NA  NA NA      TRUE
#> 17    80     pct      0 NA 0.4371585   NA -0.043316073  NA  NA NA     FALSE
#> 18    85     pct      0 NA 0.3024911   NA -0.177983440  NA  NA NA     FALSE
#> 19   807     pct      0 NA 0.5176395   NA  0.037164970  NA  NA NA     FALSE
#> 20   972     pct      0 NA 0.4804745   NA  0.000000000  NA  NA NA      TRUE
#> 21    69     pct      0 NA 0.3770492   NA -0.058860800  NA  NA NA     FALSE
#> 22    67     pct      0 NA 0.2154341   NA -0.220475897  NA  NA NA     FALSE
#> 23   755     pct      0 NA 0.4870968   NA  0.051186794  NA  NA NA     FALSE
#> 24   891     pct      0 NA 0.4359100   NA  0.000000000  NA  NA NA      TRUE
#> 25    96     pct      0 NA 0.4897959   NA  0.033868867  NA  NA NA     FALSE
#> 26    85     pct      0 NA 0.2823920   NA -0.173535025  NA  NA NA     FALSE
#> 27   719     pct      0 NA 0.4867976   NA  0.030870511  NA  NA NA     FALSE
#> 28   900     pct      0 NA 0.4559271   NA  0.000000000  NA  NA NA      TRUE
#> 29   112     pct      0 NA 0.4274809   NA -0.028783859  NA  NA NA     FALSE
#> 30    97     pct      0 NA 0.2512953   NA -0.204969439  NA  NA NA     FALSE
#> 31   949     pct      0 NA 0.5021164   NA  0.045851627  NA  NA NA     FALSE
#> 32  1158     pct      0 NA 0.4562648   NA  0.000000000  NA  NA NA      TRUE
#> 33 10117     pct      0 NA 0.4709305   NA  0.000000000  NA  NA NA      TRUE
#>    in_tottab in_refrow
#> 1      FALSE     FALSE
#> 2      FALSE     FALSE
#> 3      FALSE     FALSE
#> 4      FALSE     FALSE
#> 5      FALSE     FALSE
#> 6      FALSE     FALSE
#> 7      FALSE     FALSE
#> 8      FALSE     FALSE
#> 9      FALSE     FALSE
#> 10     FALSE     FALSE
#> 11     FALSE     FALSE
#> 12     FALSE     FALSE
#> 13     FALSE     FALSE
#> 14     FALSE     FALSE
#> 15     FALSE     FALSE
#> 16     FALSE     FALSE
#> 17     FALSE     FALSE
#> 18     FALSE     FALSE
#> 19     FALSE     FALSE
#> 20     FALSE     FALSE
#> 21     FALSE     FALSE
#> 22     FALSE     FALSE
#> 23     FALSE     FALSE
#> 24     FALSE     FALSE
#> 25     FALSE     FALSE
#> 26     FALSE     FALSE
#> 27     FALSE     FALSE
#> 28     FALSE     FALSE
#> 29     FALSE     FALSE
#> 30     FALSE     FALSE
#> 31     FALSE     FALSE
#> 32     FALSE     FALSE
#> 33      TRUE     FALSE

To get those underlying fields you can either use vctrs::fields or, more simply, $ :

tabs %>% mutate(across(where(is_fmt), ~ vctrs::field(., "pct") ))

tabs$Married$pct
tabs$Married$n
tabs %>% mutate(across(where(is_fmt), ~ .$n))

To modify a field, you can use vctrs field<-. For example, to change the displayed field :

tab(data, race, marital, year, pct = "row") %>%
mutate(across(where(is_fmt), ~ vctrs::`field<-`(., "display", rep("diff", length(.)))))

Faster to write and easier to read, you can also use dplyr::mutate() on an fmt vector. For example, to create a new column with standards deviations and display it with decimals :

tab_num(data, race, c(age, tvhours), marital, digits = 1L, comp = "all") |>
  dplyr::mutate(dplyr::across( #Mutate over the whole table.
    c(age, tvhours),
    ~ dplyr::mutate(., #Mutate over each fmt vector's underlying data.frame.
                    var     = sqrt(var), 
                    display = "var", 
                    digits  = 2L) |> 
      set_color("no"),
    .names = "{.col}_sd"
  ))
#> # A tabxplor tab: 25 × 6
#> # Groups:         marital [7]
#>    marital       race                   age tvhours     age_sd tvhours_sd
#>    <fct>         <fct>               <mean>  <mean> <mean-var> <mean-var>
#>  1 Married       Other                 42.2     2.5      13.01       1.88
#>  2 Married       Black                 46.4     3.8      13.40       3.06
#>  3 Married       White                 49.7     2.6      15.24       1.98
#>  4 Married       Total Married         48.7     2.7      15.06       2.11
#> 
#>  5 No answer     Other                 34.0     2.0       8.49           
#>  6 No answer     Black                 64.0                              
#>  7 No answer     White                 56.0     2.6      15.71       1.19
#>  8 No answer     Total No answer       52.4     2.6      16.51       1.13
#> 
#>  9 Never married Other                 30.2     2.8      10.60       2.67
#> 10 Never married Black                 34.5     4.2      12.14       3.39
#> 11 Never married White                 34.4     2.8      14.29       2.56
#> 12 Never married Total Never married   33.9     3.1      13.47       2.86
#> 
#> 13 Separated     Other                 42.5     3.3      12.97       3.26
#> 14 Separated     Black                 46.2     5.1      13.36       4.73
#> 15 Separated     White                 45.6     2.9      13.52       2.77
#> 16 Separated     Total Separated       45.3     3.5      13.43       3.60
#> 
#> 17 Divorced      Other                 45.5     3.0      11.82       2.71
#> 18 Divorced      Black                 51.0     4.3      12.67       3.74
#> 19 Divorced      White                 51.6     2.9      13.22       2.43
#> 20 Divorced      Total Divorced        51.1     3.1      13.14       2.73
#> 
#> 21 Widowed       Other                 64.5     4.2      14.84       2.79
#> 22 Widowed       Black                 67.5     4.7      13.89       3.70
#> 23 Widowed       White                 72.8     3.7      12.48       2.70
#> 24 Widowed       Total Widowed         71.7     3.9      13.00       2.90
#> 
#> 25 Ensemble      Total Ensemble        47.2     3.0      17.29       2.59
#> # age, tvhours: x > tot ×1.15; ×1.5; ×2; ×4; x < tot /1.15; /1.5; /2; /4

Some helper functions exists for total rows, total tables and reference rows (is_totrow() / as_totrow(), is_tottab() / as_tottab(), is_refrow() / as_refrow()) :

tab(data, race, marital, year, pct = "row") %>%
  dplyr::mutate(across( 
    where(is_fmt),
    ~ dplyr::if_else(is_totrow(.), 
                true  = mutate(., digits = 1L), 
                false = mutate(., digits = 2L))
  ))
#> # A tabxplor tab: 33 × 9
#> # Groups:         year [9]
#>    year    race           Married `No answer` `Never married` Separated Divorced
#>    <fct>   <fct>           <row%>      <row%>          <row%>    <row%>   <row%>
#>  1 2000    Other           44.57%       0.57%          34.29%     4.57%   11.43%
#>  2 2000    Black           28.21%          0%          36.60%    10.02%   13.99%
#>  3 2000    White           48.76%          0%          22.37%     2.76%   16.31%
#>  4 2000    Total 2000       45.4%          0%           25.3%      4.0%    15.7%
#> 
#>  5 2002    Other           49.70%          0%          28.14%     5.39%   15.57%
#>  6 2002    Black           28.29%          0%          39.76%     6.34%   16.10%
#>  7 2002    White           48.90%          0%          22.76%     2.79%   16.13%
#>  8 2002    Total 2002       45.9%          0%           25.6%      3.5%    16.1%
#> 
#>  9 2004    Other           55.72%          0%          28.86%     3.98%    8.46%
#> 10 2004    Black           33.16%          0%          40.32%     7.16%   13.26%
#> 11 2004    White           55.60%          0%          18.31%     2.69%   15.58%
#> 12 2004    Total 2004       52.6%          0%           22.0%      3.4%    14.8%
#> 
#> 13 2006    Other           51.01%       0.17%          27.87%     6.76%   10.98%
#> 14 2006    Black           27.29%          0%          42.59%     5.52%   16.09%
#> 15 2006    White           51.61%       0.15%          19.64%     2.47%   17.20%
#> 16 2006    Total 2006       48.1%        0.1%           23.9%      3.5%    16.2%
#> 
#> 17 2008    Other           43.72%          0%          37.16%     6.01%    7.65%
#> 18 2008    Black           30.25%          0%          41.64%     8.90%   12.81%
#> 19 2008    White           51.76%       0.32%          22.19%     2.18%   14.82%
#> 20 2008    Total 2008       48.0%        0.2%           26.2%      3.5%    13.9%
#> 
#> 21 2010    Other           37.70%          0%          40.44%     6.01%   10.93%
#> 22 2010    Black           21.54%          0%          51.13%     2.57%   17.36%
#> 23 2010    White           48.71%       0.06%          21.42%     2.97%   17.23%
#> 24 2010    Total 2010       43.6%          0%           27.6%      3.2%    16.7%
#> 
#> 25 2012    Other           48.98%          0%          35.71%     5.61%    6.63%
#> 26 2012    Black           28.24%          0%          39.87%     4.32%   19.60%
#> 27 2012    White           48.68%          0%          22.75%     2.98%   16.59%
#> 28 2012    Total 2012       45.6%          0%           26.6%      3.4%    16.1%
#> 
#> 29 2014    Other           42.75%          0%          34.73%     4.58%   14.12%
#> 30 2014    Black           25.13%       0.52%          43.26%     4.92%   17.62%
#> 31 2014    White           50.21%       0.11%          22.06%     2.65%   16.19%
#> 32 2014    Total 2014       45.6%        0.2%           26.6%      3.2%    16.2%
#> 
#> 33 Ensemb… Total Ensemble   47.1%        0.1%           25.2%      3.5%    15.7%
#> # ℹ 2 more variables: Widowed <row%>, Total <row%>

Each fmt column have attributes, which you can access or modify with get_ and set_ functions :

For example, to print the number of observations of the total column :

tab(data, race, marital, year, pct = "row") %>%
  mutate(across(where(is_totcol), ~ mutate(., display = "n") ))
#> # A tabxplor tab: 33 × 9
#> # Groups:         year [9]
#>    year    race           Married `No answer` `Never married` Separated Divorced
#>    <fct>   <fct>           <row%>      <row%>          <row%>    <row%>   <row%>
#>  1 2000    Other              45%          1%             34%        5%      11%
#>  2 2000    Black              28%          0%             37%       10%      14%
#>  3 2000    White              49%          0%             22%        3%      16%
#>  4 2000    Total 2000         45%          0%             25%        4%      16%
#> 
#>  5 2002    Other              50%          0%             28%        5%      16%
#>  6 2002    Black              28%          0%             40%        6%      16%
#>  7 2002    White              49%          0%             23%        3%      16%
#>  8 2002    Total 2002         46%          0%             26%        3%      16%
#> 
#>  9 2004    Other              56%          0%             29%        4%       8%
#> 10 2004    Black              33%          0%             40%        7%      13%
#> 11 2004    White              56%          0%             18%        3%      16%
#> 12 2004    Total 2004         53%          0%             22%        3%      15%
#> 
#> 13 2006    Other              51%          0%             28%        7%      11%
#> 14 2006    Black              27%          0%             43%        6%      16%
#> 15 2006    White              52%          0%             20%        2%      17%
#> 16 2006    Total 2006         48%          0%             24%        3%      16%
#> 
#> 17 2008    Other              44%          0%             37%        6%       8%
#> 18 2008    Black              30%          0%             42%        9%      13%
#> 19 2008    White              52%          0%             22%        2%      15%
#> 20 2008    Total 2008         48%          0%             26%        3%      14%
#> 
#> 21 2010    Other              38%          0%             40%        6%      11%
#> 22 2010    Black              22%          0%             51%        3%      17%
#> 23 2010    White              49%          0%             21%        3%      17%
#> 24 2010    Total 2010         44%          0%             28%        3%      17%
#> 
#> 25 2012    Other              49%          0%             36%        6%       7%
#> 26 2012    Black              28%          0%             40%        4%      20%
#> 27 2012    White              49%          0%             23%        3%      17%
#> 28 2012    Total 2012         46%          0%             27%        3%      16%
#> 
#> 29 2014    Other              43%          0%             35%        5%      14%
#> 30 2014    Black              25%          1%             43%        5%      18%
#> 31 2014    White              50%          0%             22%        3%      16%
#> 32 2014    Total 2014         46%          0%             27%        3%      16%
#> 
#> 33 Ensemb… Total Ensemble     47%          0%             25%        3%      16%
#> # ℹ 2 more variables: Widowed <row%>, Total <row%-n>

Note that, if tab_vars are provided, the table is grouped and all operations are made within groups. To remove grouping (for example when it gives errors), use dplyr::ungroup().

If you only need the simplest table, with only numeric counts (no fmt), or even a base data.frame (not a tibble) :

tab_plain(data, race, marital, num = TRUE) # counts as a numeric vector
tab_plain(data, race, marital, df = TRUE)  # same, with unique class = "data.frame"