Tabmiss sort

[joan marc]

Hi all statalizers,

I am performing descriptive analysis. I want to list variables with missing values, sorting them by the amount or frequency of missings. Can you think of anything I might use? I have tried mdesc(SSC), misstable, nmissing, missing and tabmiss and checked around the internet, but none of them seems to have options in order to sort.

Thank you in advance.

[Marcos Almeida]

Hello Joan,

Please check if this is what you wish:

. sysuse auto.dta
(1978 Automobile Data)

. codebook rep78

---------------------------------------------------------------------------------------------------------------------------------------
rep78                                                                                                                Repair Record 1978
---------------------------------------------------------------------------------------------------------------------------------------

                  type:  numeric (int)

                 range:  [1,5]                        units:  1
         unique values:  5                        missing .:  5/74

            tabulation:  Freq.  Value
                             2  1
                             8  2
                            30  3
                            18  4
                            11  5
                             5  .

. replace rep78 = .a in 3
(1 real change made, 1 to missing)

. replace rep78 = .c in 7
(1 real change made, 1 to missing)

. replace rep78 = .c in 45
(1 real change made, 1 to missing)

. replace rep78 = .a in 51
(1 real change made, 1 to missing)

. list rep78 if missing(rep78)

     +-------+
     | rep78 |
     |-------|
  3. |    .a |
  7. |    .c |
 45. |    .c |
 51. |    .a |
 64. |     . |
     +-------+

. gsort rep78, mfirst

. list rep78 if missing(rep78)

     +-------+
     | rep78 |
     |-------|
 70. |     . |
 71. |    .a |
 72. |    .a |
 73. |    .c |
 74. |    .c |
     +-------+

Best,

Marcos

[Nick Cox]

mdesc is from SSC (Rose Medeiros)

misstable is official.

nmissing is from Stata Journal (myself),

missing: do you mean missings (myself, Stata Journal)?

tabmiss is from SSC (Marcelo Coca-Perraillon).

You're likely to be right. But it's programmable.

webuse nlswork, clear

unab all : _all

gen nmissing = .

gen varname = ""

local i = 1
qui foreach v of local all {
    count if missing(`v')
    replace nmissing = r(N) in `i'
    replace varname = "`v'" in `i'
    local ++i
}

gsort -nmissing

list varname nmissing if inrange(nmissing, 1, .) , noobs sep(0)

  +---------------------+
  |  varname   nmissing |
  |---------------------|
  |    union       9296 |
  |   wks_ue       5704 |
  | wks_work        703 |
  |   tenure        433 |
  | ind_code        341 |
  | occ_code        121 |
  |    hours         67 |
  |      age         24 |
  |      msp         16 |
  |  nev_mar         16 |
  |   c_city          8 |
  | not_smsa          8 |
  |    south          8 |
  |    grade          2 |
  +---------------------+

[joan marc]

How can I add column percentages to this list?

[Carlo Lazzaro]

Joan:
shamelessly taking advantage of Nick's code, you may want to try:

g perc_nmissing= nmissing/_N
format perc_nmissing %12.4f
list varname nmissing perc_nmissing if inrange(nmissing, 1, .) , noobs sep(0)