CODEWITHSUNDEEP
rplotcontingency
Antoni Parellada
Antoni Parellada

Reputation: 4801

How to extract the calculated expected frequencies in a higher-dimensional mosaicplot in R

After producing a table with three-variables as in:

salary <- array(
  c(38, 12, 102, 141, 12, 9, 136, 383),
  dim=c(2, 2, 2),
  dimnames=list(exposure = c("exposed", "not"),
                disease = c("case", "control"),
                salary = c("<1000", ">=1000"))
) 

salary
, , salary = <1000

         disease
exposure  case control
  exposed   38     102
  not       12     141

, , salary = >=1000

         disease
exposure  case control
  exposed   12     136
  not        9     383

the vcd package allows the creation of a mosaic plot like this:

mosaic(salary, shade = T)

enter image description here

in which the surface area is proportional to the counts in the cell, and the color corresponds to the standardized Pearson's residuals.

The question is how to calculate and extract the expected counts?

I know that I can produce the margins like this:

addmargins(salary)
, , salary = <1000

         disease
exposure  case control Sum
  exposed   38     102 140
  not       12     141 153
  Sum       50     243 293

, , salary = >=1000

         disease
exposure  case control Sum
  exposed   12     136 148
  not        9     383 392
  Sum       21     519 540

, , salary = Sum

         disease
exposure  case control Sum
  exposed   50     238 288
  not       21     524 545
  Sum       71     762 833

and if it were a 2 x 2 contingency table I would use chisq.test(x)$expected to see the expected values, but I don't know how to go about it in a 3-way contingency table.

Upvotes: 0

Views: 413

Answers (1)

Johan Larsson
Johan Larsson

Reputation: 3694

There is a designated argument to compute (and visualize) expected values using the type = "expected" option.

mos_exp <- mosaic(salary, type = "expected")

mos_exp

#>                 disease       case    control
#> exposure salary                              
#> exposed  <1000            8.634326  92.666994
#>          >=1000          15.913093 170.785587
#> not      <1000           16.339263 175.359416
#>          >=1000          30.113318 323.188003

Upvotes: 2

Related Questions