Boxplot/quantiles for frequency

[Harina Peternella]

Hi everyone, I'm struggling with my dataset and I don't know how I can fix it. Here you see a sample of the data:

* Example generated by -dataex-. To install: ssc install dataex
clear
input byte Inkomensgroep int Allehuishoudens

 1  48
 3  57
 5  58
 7  65
 9  71
11 102
13 206
end

The first column is income while the second one shows the frequency. I would like to calculate the mean, first and third quantile (and after that a box plot).
However, how should I let STATA know it is a frequency so it doesn't not only plot income without taking the different frequencies into account /or how can I make one dataset with 48 times 1 as value, and 57 times 3 etc ?

Thank you very much!

[Sergiy Radyakin]

Hi everyone, I'm struggling with my dataset ... how can I make one dataset with 48 times 1 as value, and 57 times 3 etc ?

expand Allehuishoudens

To verify:

tabulate Inkomensgroep 

Inkomensgro |
         ep |      Freq.     Percent        Cum.
------------+-----------------------------------
          1 |         48        7.91        7.91
          3 |         57        9.39       17.30
          5 |         58        9.56       26.85
          7 |         65       10.71       37.56
          9 |         71       11.70       49.26
         11 |        102       16.80       66.06
         13 |        206       33.94      100.00
------------+-----------------------------------
      Total |        607      100.00

[Harina Peternella]

That was the command I was looking for, thank you!!!

[Nick Cox]

I guess quantile here means quartile. Note that box plots can look very odd with discrete variables. Here your example probably is weirder than your full dataset, but with stripplot from SSC I get this:

clear
input byte Inkomensgroep int Allehuishoudens
 1  48
 3  57
 5  58
 7  65
 9  71
11 102
13 206
end

expand Alle
stripplot Ink , vertical box cumul refline centre yla(, ang(h))

This has to be decoded: more than 25% are in the top group: therefore the same value is reported as the upper (third) quartile and the maximum, and there is no whisker.

The refline option in stripplot defaults to showing the mean as an extra horizontal line (in this case).

A histogram might be clearer.