The R Book

Tables of Counts

Here are simulated data from a trial in which red blood cells were counted on 10 000 slides. The mean number of cells per slide was (μ)1.2 and the distribution had an aggregation parameter k = 0.63 (known in R as size). The probability for a negative binomial distribution (prob in R) is given by k/(μ+k) = 0.63/1.83 so

cells<-rnbinom(10000,size=0.63,prob=0.63/1.83)

We want to count how many times we got no red blood cells on the slide, and how often we got 1, 2, 3, . . . cells. The R function for this is table:

table(cells)

cells
   0    1    2   3   4   5  6   7   8   9   10    11    12   13   14   15
5149 2103 1136 629 364 226 158 81  52  33   22    11    11    6    9    5
  16   17   24
   3    1    1

That's all there is to it. You will get slightly different values because of the randomization. We found 5149 slides with no red blood cells and one slide with a massive 24 red blood cells.

We often want to count separately for each level of a factor. Here we know that the first 5000 samples came from male patients and the second 5000 from females:

gender<-rep(c("male","female"),c(5000,5000))

To tabulate the counts separately for the two sexes we just write

table(cells,gender)

gender
cells    female    male
    0      2646    2503
    1      1039    1064
    2       537     599
    3       298     331
    4       165     199
    5       114     112
    6       82       76
    7       43       38
    8       30       22
    9       16       17
   10        7       15
   11        5        6
   12        5        6
   13        6        0
   14        3        6
   15        3        2
   16        0        3
   17        1        0
   24        0        1

Evidently there are no major differences between the sexes in these red blood counts. A statistical comparison of the two sets of counts involves the use of log-linear ...