Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Multiple imputation problem

    This is a specific problem emerged in my previous post "Reshape wide".

    The data are the following:

    Code:
    * Example generated by -dataex-. To install: ssc install dataex
    clear
    input double(idfirm Year) float(inflow_ratio outflow_ratio avsales) double tot_sales
     1 2008          .           .  94701.45 211591.55354704778
     1 2009          0           0  83603.08  308746.0587639413
     1 2010        .25           0  91092.75  407237.1453954251
     1 2011         .4           0 118792.94  598549.0481277383
     1 2012   .2857143           0  104542.8  664061.6305459269
     1 2013   .5555556           0 73541.766    608140.67505112
     1 2014   .3571429           0  48557.22  715163.0758130773
     1 2015          0   .05263158  52369.71  768412.7578732732
     2 2004          .           . 2039558.8  12237352.38962967
     2 2005          0           0   2109991 11513177.507807251
     2 2006          0           0 2050668.5 12304011.532418355
     2 2007          0           0   6330808 11406777.403632266
     2 2008  .16666667           0 2734569.5 12355367.684264863
     2 2009          0           0 1841746.4 12892224.398947667
     2 2010          0   .14285715   5212612 10897763.444727022
     2 2011          0           0   5262992  10580189.86100467
     2 2012          0           0 2037856.5 10967562.283051893
     2 2013          0           0 1823603.5  9124157.813647183
     2 2014          0    .3333333   1785749  7133930.997182059
     2 2015          0           0   1356981  5427924.711631365
     3 2004          .           .  53702.81 2464566.2638161103
     3 2005  .23913044           0  81622.65   4406664.46751826
     3 2006          0   .03508772  67388.88  1843225.499704768
     3 2007          0   .14545454  11123.49  476002.8259390887
     3 2008          0   .21276596  7850.393  160865.5648153848
     3 2009          0    .1081081  6033.619   82650.9899064962
     3 2010   .3030303           0 4410.1416 126687.58501765266
     3 2011          0     .372093 2412.5247  35362.49685707843
     3 2012          0    .4074074  1075.685 16035.562916347304
     3 2013      .1875           0   3456.24 22272.424034809985
     3 2014          0    .2105263 1496.9843  21090.64112449751
     3 2015          0   .26666668  4020.239  23837.82400447101
     4 2004          .           .  56691.86  8398246.263306491
     4 2005  .03947368           0  78010.12  8251045.135936161
     4 2006   .0949367           0  94482.98  9847010.316997783
     4 2007   .2485549           0 126469.87 13503137.133153718
     4 2008   .1712963           0  78780.68 13768126.595234673
     4 2009  .07509881           0 73256.164  13834060.86932386
     4 2010 .025735294           0  62012.55 12531817.194234656
     4 2011 .014336918           0   70322.5 12433251.178176137
     4 2012  .03180212           0  64410.49 11206517.411851741
     4 2013  .05821918           0     79104 14594676.742348071
     4 2014  .05825243           0 119126.19 26481277.906628065
     4 2015          0 .0091743115  176931.8  29946995.67406721
     5 2007          .           . 157.76617   39.4415442669969
     6 2004          .           .  28066686  28066686.19154623
     6 2005          0           0   5704382  5704381.338495776
     6 2006          0           0  496491.9 496491.87314146105
     6 2007          0           0  9421.785  9421.785319502253
     6 2008          0           0  1577.368  394.3420099042942
     6 2009          0           0  5520.829 1380.2071011962253
     7 2007          .           .  35459.79  35459.78814655074
     7 2008          0           0  42076.49  42076.49109553711
     7 2009          0           0  381988.5   381988.536163157
     7 2010          0           0 1804061.4  451015.3511721381
     7 2011          0           0  270585.6 270585.60116440145
     7 2012          0           0   1318229 329557.26253078564
     7 2013          0           0 1220001.8  305000.4474947418
     7 2014          0           0  693862.8 173465.68589452517
     7 2015          0           0 31625.557  31625.55690904084
     8 2004          .           .  11277370  1499855855.865098
     8 2005          0   .03007519  18729456 1689540069.7170644
     8 2006          0    .1627907  18472182  1598121054.119387
     8 2007          0    .0462963  32214094  1656992400.409758
     8 2008          0   .03883495  36879372   2142910992.52344
     8 2009          0    .1010101  34457748 2201020370.9059067
     8 2010          0    .0786517  35879268  2181284539.324025
     8 2011 .024390243           0  32816464 2029940306.3960855
     8 2012  .11904762           0  27333772 2225894374.1270456
     8 2013  .05319149           0  46580236  2727305730.021949
     8 2014          0           0  33949468 2359259573.3842134
     8 2015          0   .05050505  29031148  2370530389.480409
     9 2004          .           . 300541632   9917874207.75519
     9 2005          0           0 398063520 10607521709.816193
     9 2006          0   .03030303 711789248 11125455014.939121
     9 2007          0           0 485468992 10999908825.891459
     9 2008      .0625           0 471612384 11020835162.138296
     9 2009          0           0 619878528 11052308521.725805
     9 2010          0           0 775569536 11844314656.874342
     9 2011          0           0 471647968  12248782762.78322
     9 2012          0  .029411765 423355936 12198970579.756456
     9 2013          0    .0909091 670351680 11978041077.652853
     9 2014         .1           0 574183808 13100594403.908373
     9 2015  .03030303           0 532608064 16695333854.076557
    10 2004          .           . 10196.007 20392.013641854945
    11 2008          .           .  830263.1 1020956.3147491836
    11 2009          0           0 1215518.4  2431036.790986024
    11 2010          0           0 302000.13  514778.3255435499
    11 2011          0           0 133911.64  66955.82008718849
    11 2012          0          .5 1857.2793 1857.2792628005145
    11 2013          0           0 229202.33  57300.58370722538
    11 2014          0           0 124664.38  31166.09542619379
    12 2012          .           .  5597.375 1399.3437130821144
    12 2013          0           0  46875.85 46875.852646417305
    12 2014          0           0  92413.03  92413.03425338621
    12 2015          0           0 120786.42 120786.41920184325
    13 2004          .           .   3551802 3551802.0606248565
    13 2005          0           0   5632205  5632204.502071029
    13 2006          1           0   3682722 6445605.7851592805
    13 2007          0           0  11139598  5607216.736637451
    end

    What I am trying to do is multiple imputation after having reshaped wide. Having discovered the need to declare xtset, clear before mi-setting in the forum I produced the following code:

    Code:
    xtset, clear
    mi set mlong
    drop data_lancio_molecola inflow outflow internationalproduct  molecule
    mi reshape wide inflow_ratio outflow_ratio nprod avsales avsales_existing avsales_new agefirm tot_sales recalled_dummy_byfirm newmolfirm newmolmarket, i(idfirm) j(Year)
     
    mi register imputed inflow_ratio* outflow_ratio* agefirm*
    mi impute chained (logit) inflow_ratio* (regress) outflow_ratio* (regress) agefirm*(regress) = avsales* tot_sales* newmolfirm* newmolmarket*, add(10) rseed (091107)
    which however produces the following error:

    Code:
    at least one imputation variable required: (method ...) ivar(s)
     -- above applies to specification (regress )
    
    mi impute chained: invalid specification;
        see above error messages
    Can anyone help me please?

    Federico
    Last edited by Federico Nutarelli; 10 Jan 2019, 09:52.

  • #2
    I'm not quite sure what you are trying to accomplish here, but it looks to me like you want to reshape the data before you do the imputation using just the standard reshape command. After reshaping the data your MI commands appear to be intended to work on the resulting file. Using mi reshape assumes you have already imputed the data and want to reshape the resulting files. See help for mi reshape.
    Richard T. Campbell
    Emeritus Professor of Biostatistics and Sociology
    University of Illinois at Chicago

    Comment


    • #3
      Many thanks for the reply Professor Campbell,

      actually if I use just reshape, STATA ask me to use reshape:

      Code:
      no; data are mi set
          Use mi reshape to perform reshape on these data.  mi reshape has the same syntax as reshape.
      
          Perhaps you did not type reshape.  In that case, the command you typed calls reshape and it is
          not appropriate for use with mi data.  Use mi extract or mi xeq to select the data on which you
          want to run the command, which is probably m=0.

      Comment


      • #4
        You are trying to reshape after declaring the data as mi. If I understand what you are trying to accomplish you should do the reshape before doing anything at all about mi. It appear to me that you wish to reshape a pre-exiting wide format data set as indicated by your mi register statement and then do multiple imputation. If that is the case, as I said, you want to reshape the data before starting the MI process.

        xtset, clear
        drop .....
        reshape ....
        mi set mlong
        mi register ......
        mi impute ......

        Richard T. Campbell
        Emeritus Professor of Biostatistics and Sociology
        University of Illinois at Chicago

        Comment


        • #5
          So I have a dataset which is in long format. I am reshaping it into wide and then trying to perform multiple imputation.
          Also with the code you kindly posted, the error:
          Code:
          at least one imputation variable required: (method ...) ivar(s)
           -- above applies to specification (regress )
          
          mi impute chained: invalid specification;
              see above error messages
          appears. Is it possible that mi do not accept "*"?

          Comment

          Working...
          X