csdid output varying with the addition of a single variable

[Scott Rick]

Hi. I am trying to use the csdid command and am struggling to understand the workings of it. My results appear to vary wildly on adding an additional control variable that doesn't have any missing values. My time variable has 48 periods (number of weeks) and my gvar ranges from 3-9.

In model 1 below, I get very few failed 2x2 estimations. However, when I add just one more variable in model 2 - rate_hospital_beds_total1k which has no missing observations - the number of failed 2x2 estimations increase significantly with all of g9 omitted. I would greatly appreciate understanding:

1. why the addition of a single variable with no missing values would cause so much fluctuation? My gvar=9 group has 46 treated observations. However, gvar=7 group has 47 treated observations and appears to run fine for that group. I do see that the number of observations has changed with the new variable being added, but fail to understand why it is so.
2. In the output below, for t_2_46, t_2_47 I understand that the 46 and corresponds to the week number/time variable. However, I am failing to understand what the preceding 2 corresponds to.

FernandoRios I'm adding you on here given your expertise on this.

Model 1:

 csdid rate_con_avg7day pop_perkm2_2019 povertyperct deathrate_2019 infantmortrate_2019 birthr
> ate_2019 naturalgrwthrate_2019  rate_hospital_total1k, ivar(statenum) time(weeknum) gvar(gvar
> )
Panel is not balanced
Will use observations with Pair balanced (observed at t0 and t1)
...............................................x..
..................................................
.........................................x........
......................................xxxxx.......
...................................xx.............
................................
Difference-in-difference with Multiple Time Periods

                                                Number of obs     =      1,539
Outcome model  : least squares
Treatment model: inverse probability
------------------------------------------------------------------------------
             |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
g3           |
       t_1_2 |   6.33e-08   1.44e-07     0.44   0.660    -2.19e-07    3.46e-07
       t_2_3 |   6.18e-06   4.04e-06     1.53   0.127    -1.75e-06    .0000141
       t_2_4 |   .0000146   .0000108     1.34   0.179    -6.66e-06    .0000358
..................................    
      t_2_46 |   .0033591   .0128041     0.26   0.793    -.0217365    .0284547
      t_2_47 |   .0032703   .0130755     0.25   0.803    -.0223572    .0288979
      t_2_48 |   .0009822   .0038111     0.26   0.797    -.0064875    .0084518
-------------+----------------------------------------------------------------
g4           |
       t_1_2 |          0  (omitted)
       t_2_3 |   6.61e-06   2.72e-06     2.43   0.015     1.27e-06    .0000119
       t_3_4 |  -3.00e-06   8.29e-06    -0.36   0.718    -.0000193    .0000133
       t_3_5 |  -2.14e-06   9.60e-06    -0.22   0.823     -.000021    .0000167
       t_3_6 |   2.69e-08   .0000124     0.00   0.998    -.0000243    .0000244
   ..................................    
      t_3_48 |  -.0010156   .0014222    -0.71   0.475     -.003803    .0017718
-------------+----------------------------------------------------------------
g5           |
       t_1_2 |   1.62e-08   9.78e-07     0.02   0.987    -1.90e-06    1.93e-06
       t_2_3 |  -8.39e-07   4.56e-06    -0.18   0.854    -9.78e-06    8.10e-06
       t_3_4 |   2.16e-06   .0000205     0.11   0.916    -.0000381    .0000424
       t_4_5 |   5.01e-06   6.49e-06     0.77   0.440    -7.72e-06    .0000177
       t_4_6 |   .0000153   .0000151     1.01   0.311    -.0000143     .000045
     ..................................    
      t_4_46 |    .011131   .0118398     0.94   0.347    -.0120746    .0343366
      t_4_47 |   .0110686   .0111938     0.99   0.323    -.0108709    .0330081
      t_4_48 |   .0031596   .0031183     1.01   0.311    -.0029522    .0092714
-------------+----------------------------------------------------------------
g7           |
       t_1_2 |          0  (omitted)
       t_2_3 |   2.15e-07   7.30e-07     0.29   0.769    -1.22e-06    1.65e-06
       t_3_4 |   8.76e-06   8.37e-06     1.05   0.295    -7.64e-06    .0000252
       t_4_5 |   1.43e-06   2.52e-06     0.57   0.571    -3.51e-06    6.37e-06
       t_5_6 |   7.47e-07   1.99e-06     0.38   0.707    -3.15e-06    4.64e-06
       t_6_7 |  -2.21e-06   2.39e-06    -0.92   0.356    -6.90e-06    2.48e-06
       ..................................    
      t_6_47 |  -.0050742   .0030715    -1.65   0.099    -.0110943    .0009459
      t_6_48 |  -.0015361   .0008884    -1.73   0.084    -.0032774    .0002052
-------------+----------------------------------------------------------------
g8           |
       t_1_2 |          0  (omitted)
       t_2_3 |          0  (omitted)
       t_3_4 |          0  (omitted)
       t_4_5 |          0  (omitted)
       t_5_6 |          0  (omitted)
       t_6_7 |   4.55e-06   3.46e-06     1.32   0.188    -2.23e-06    .0000113
       t_7_8 |   6.05e-06   5.52e-06     1.10   0.273    -4.76e-06    .0000169
       t_7_9 |   7.19e-07   7.95e-06     0.09   0.928    -.0000149    .0000163
      t_7_10 |  -8.68e-06   .0000171    -0.51   0.612    -.0000422    .0000249
      t_7_11 |  -3.15e-07   .0000265    -0.01   0.991    -.0000523    .0000517
     ..................................    
      t_7_47 |   .0009435   .0025036     0.38   0.706    -.0039635    .0058505
      t_7_48 |   .0002654   .0007203     0.37   0.712    -.0011463    .0016772
-------------+----------------------------------------------------------------
g9           |
       t_1_2 |          0  (omitted)
       t_2_3 |          0  (omitted)
       t_3_4 |   1.02e-07   3.67e-06     0.03   0.978    -7.09e-06    7.30e-06
       t_4_5 |   7.45e-07   1.17e-06     0.63   0.526    -1.56e-06    3.05e-06
       t_5_6 |   1.72e-06   8.83e-07     1.95   0.051    -7.42e-09    3.45e-06
       t_6_7 |   3.74e-06   1.71e-06     2.19   0.029     3.90e-07    7.10e-06
       t_7_8 |   8.07e-06   2.39e-06     3.37   0.001     3.37e-06    .0000128
       t_8_9 |   4.16e-06   3.03e-06     1.37   0.170    -1.78e-06    .0000101
      t_8_10 |   .0000124   6.79e-06     1.83   0.067    -8.61e-07    .0000258
      t_8_11 |   .0000224   .0000106     2.13   0.034     1.75e-06    .0000431
      t_8_12 |    .000037   .0000142     2.61   0.009     9.23e-06    .0000649
    ..................................    
      t_8_47 |   .0039992   .0010031     3.99   0.000     .0020332    .0059652
      t_8_48 |   .0011212   .0002918     3.84   0.000     .0005493    .0016931
------------------------------------------------------------------------------



Model 2: 

 csdid rate_con_avg7day pop_perkm2_2019 povertyperct deathrate_2019 infantmortrate_2019 birthr
> ate_2019 naturalgrwthrate_2019  rate_hospital_total1k rate_hospital_beds_total1k, ivar(staten
> um) time(weeknum) gvar(gvar)
Panel is not balanced
Will use observations with Pair balanced (observed at t0 and t1)
...............................................x..
..................................................
.........................................x........
......................................xxxxx.......
...................................xxxxxxxxxxxxxxx
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Difference-in-difference with Multiple Time Periods

                                                Number of obs     =      1,493
Outcome model  : least squares
Treatment model: inverse probability
------------------------------------------------------------------------------
             |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
g3           |
       t_1_2 |  -6.62e-07   5.50e-20 -1.2e+13   0.000    -6.62e-07   -6.62e-07
       t_2_3 |  -1.97e-06   2.67e-06    -0.74   0.460    -7.21e-06    3.26e-06
       t_2_4 |  -6.96e-06   8.59e-06    -0.81   0.418    -.0000238    9.87e-06
         ..................................    
      t_2_48 |  -.0026097   .0037858    -0.69   0.491    -.0100298    .0048104
-------------+----------------------------------------------------------------
g4           |
       t_1_2 |          0  (omitted)
       t_2_3 |   5.47e-06   2.27e-06     2.41   0.016     1.02e-06    9.91e-06
       t_3_4 |  -2.66e-06   8.64e-06    -0.31   0.758    -.0000196    .0000143
       t_3_5 |  -1.78e-06   8.34e-06    -0.21   0.831    -.0000181    .0000146
       ..................................    
      t_3_48 |    -.00091   .0015944    -0.57   0.568     -.004035     .002215
-------------+----------------------------------------------------------------
g5           |
       t_1_2 |   3.81e-06   2.60e-06     1.46   0.143    -1.29e-06    8.90e-06
       t_2_3 |  -2.80e-12   3.70e-07    -0.00   1.000    -7.25e-07    7.25e-07
       t_3_4 |   6.47e-06    .000034     0.19   0.849    -.0000603    .0000732
       t_4_5 |   7.55e-06   7.60e-06     0.99   0.321    -7.36e-06    .0000224
       t_4_6 |   .0000215    .000018     1.20   0.232    -.0000138    .0000569
         ..................................    
      t_4_48 |   .0028758   .0058455     0.49   0.623    -.0085811    .0143326
-------------+----------------------------------------------------------------
g7           |
       t_1_2 |          0  (omitted)
       t_2_3 |  -2.21e-08   8.03e-07    -0.03   0.978    -1.60e-06    1.55e-06
       t_3_4 |   .0000107   8.11e-06     1.32   0.186    -5.18e-06    .0000266
       t_4_5 |   1.68e-06   2.57e-06     0.65   0.515    -3.36e-06    6.71e-06
       t_5_6 |  -1.74e-06   2.38e-06    -0.73   0.464    -6.41e-06    2.92e-06
       t_6_7 |  -6.96e-06   4.41e-06    -1.58   0.114    -.0000156    1.67e-06
       t_6_8 |  -.0000214   .0000104    -2.05   0.040    -.0000418   -9.58e-07
          ..................................    
      t_6_48 |  -.0025231    .001737    -1.45   0.146    -.0059275    .0008812
-------------+----------------------------------------------------------------
g8           |
       t_1_2 |          0  (omitted)
       t_2_3 |          0  (omitted)
       t_3_4 |          0  (omitted)
       t_4_5 |          0  (omitted)
       t_5_6 |          0  (omitted)
       t_6_7 |   2.34e-06   3.73e-06     0.63   0.530    -4.97e-06    9.66e-06
       t_7_8 |   6.16e-06   6.59e-06     0.93   0.350    -6.76e-06    .0000191
       t_7_9 |   -.000014   9.88e-06    -1.41   0.158    -.0000333    5.40e-06
        ..................................    
      t_7_48 |   .0016771   .0011288     1.49   0.137    -.0005353    .0038895
-------------+----------------------------------------------------------------
g9           |
       t_1_2 |          0  (omitted)
       t_2_3 |          0  (omitted)
       t_3_4 |          0  (omitted)
       t_4_5 |          0  (omitted)
       t_5_6 |          0  (omitted)
       t_6_7 |          0  (omitted)
       t_7_8 |          0  (omitted)
       t_8_9 |          0  (omitted)
      t_8_10 |          0  (omitted)
      t_8_11 |          0  (omitted)
      t_8_12 |          0  (omitted)
      t_8_13 |          0  (omitted)
      t_8_14 |          0  (omitted)
      t_8_15 |          0  (omitted)
      t_8_16 |          0  (omitted)
      t_8_17 |          0  (omitted)
      t_8_18 |          0  (omitted)
      t_8_19 |          0  (omitted)
      t_8_20 |          0  (omitted)
      t_8_21 |          0  (omitted)
      t_8_22 |          0  (omitted)
      t_8_23 |          0  (omitted)
      t_8_24 |          0  (omitted)
      t_8_25 |          0  (omitted)
      t_8_26 |          0  (omitted)
      t_8_27 |          0  (omitted)
      t_8_28 |          0  (omitted)
      t_8_29 |          0  (omitted)
      t_8_30 |          0  (omitted)
      t_8_31 |          0  (omitted)
      t_8_32 |          0  (omitted)
      t_8_33 |          0  (omitted)
      t_8_34 |          0  (omitted)
      t_8_35 |          0  (omitted)
      t_8_36 |          0  (omitted)
      t_8_37 |          0  (omitted)
      t_8_38 |          0  (omitted)
      t_8_39 |          0  (omitted)
      t_8_40 |          0  (omitted)
      t_8_41 |          0  (omitted)
      t_8_42 |          0  (omitted)
      t_8_43 |          0  (omitted)
      t_8_44 |          0  (omitted)
      t_8_45 |          0  (omitted)
      t_8_46 |          0  (omitted)
      t_8_47 |          0  (omitted)
      t_8_48 |          0  (omitted)
------------------------------------------------------------------------------

[FernandoRios]

Can you tab your time variable va gvar?
I suspect your effective sample is too small for the number of controls you are using

[Scott Rick]

FernandoRios Here's what I get:

tab weeknum gvar

           |                               gvar
   weeknum |         0          3          4          5          7          8 |     Total
-----------+------------------------------------------------------------------+----------
         1 |         8          1          0          3          0          0 |        12 
         2 |        13          2          3          3          1          0 |        22 
         3 |        16          2          3          3          1          1 |        27 
         4 |        17          2          3          5          1          1 |        30 
         5 |        19          2          4          5          1          1 |        33 
         6 |        19          2          4          5          1          2 |        34 
         7 |        19          2          4          5          1          2 |        34 
         8 |        19          2          4          5          1          2 |        34 
         9 |        20          2          4          5          1          2 |        35 
        10 |        20          2          4          5          1          2 |        35 
        11 |        20          2          4          5          1          2 |        35 
        12 |        20          2          4          5          1          2 |        35 
        13 |        20          2          4          5          1          2 |        35 
        14 |        20          2          4          5          1          2 |        35 
        15 |        20          2          4          5          1          2 |        35 
        16 |        20          2          4          5          1          2 |        35 
        17 |        20          2          4          5          1          2 |        35 
        18 |        20          2          4          5          1          2 |        35 
        19 |        20          2          4          5          1          2 |        35 
        20 |        20          2          4          5          1          2 |        35 
        21 |        20          2          4          5          1          2 |        35 
        22 |        20          2          4          5          1          2 |        35 
        23 |        20          2          4          5          1          2 |        35 
        24 |        20          2          4          5          1          2 |        35 
        25 |        20          2          4          5          1          2 |        35 
        26 |        20          2          4          5          1          2 |        35 
        27 |        20          2          4          5          1          2 |        35 
        28 |        20          2          4          5          1          2 |        35 
        29 |        20          2          4          5          1          2 |        35 
        30 |        20          2          4          5          1          2 |        35 
        31 |        20          2          4          5          1          2 |        35 
        32 |        20          2          4          5          1          2 |        35 
        33 |        20          2          4          5          1          2 |        35 
        34 |        20          2          4          5          1          2 |        35 
        35 |        20          2          4          5          1          2 |        35 
        36 |        20          2          4          5          1          2 |        35 
        37 |        20          2          4          5          1          2 |        35 
        38 |        20          2          4          5          1          2 |        35 
        39 |        20          2          4          5          1          2 |        35 
        40 |        20          2          4          5          1          2 |        35 
        41 |        20          2          4          5          1          2 |        35 
        42 |        20          2          4          5          1          2 |        35 
        43 |        20          2          4          5          1          2 |        35 
        44 |        20          2          4          5          1          2 |        35 
        45 |        20          2          4          5          1          2 |        35 
        46 |        20          2          4          5          1          2 |        35 
        47 |        20          2          4          5          1          2 |        35 
        48 |        20          2          4          5          1          2 |        35 
-----------+------------------------------------------------------------------+----------
     Total |       930         95        185        234         47         89 |     1,626 


           |    gvar
   weeknum |         9 |     Total
-----------+-----------+----------
         1 |         0 |        12 
         2 |         0 |        22 
         3 |         1 |        27 
         4 |         1 |        30 
         5 |         1 |        33 
         6 |         1 |        34 
         7 |         1 |        34 
         8 |         1 |        34 
         9 |         1 |        35 
        10 |         1 |        35 
        11 |         1 |        35 
        12 |         1 |        35 
        13 |         1 |        35 
        14 |         1 |        35 
        15 |         1 |        35 
        16 |         1 |        35 
        17 |         1 |        35 
        18 |         1 |        35 
        19 |         1 |        35 
        20 |         1 |        35 
        21 |         1 |        35 
        22 |         1 |        35 
        23 |         1 |        35 
        24 |         1 |        35 
        25 |         1 |        35 
        26 |         1 |        35 
        27 |         1 |        35 
        28 |         1 |        35 
        29 |         1 |        35 
        30 |         1 |        35 
        31 |         1 |        35 
        32 |         1 |        35 
        33 |         1 |        35 
        34 |         1 |        35 
        35 |         1 |        35 
        36 |         1 |        35 
        37 |         1 |        35 
        38 |         1 |        35 
        39 |         1 |        35 
        40 |         1 |        35 
        41 |         1 |        35 
        42 |         1 |        35 
        43 |         1 |        35 
        44 |         1 |        35 
        45 |         1 |        35 
        46 |         1 |        35 
        47 |         1 |        35 
        48 |         1 |        35 
-----------+-----------+----------
     Total |        46 |     1,626

[FernandoRios]

That explains it
you see how your sample is very small? 1 2 4 & 5
so based on the smallest cell size you can’t add any controls
Adding controls creates problems of perfect fitting with drimp dripw or ipw
reg will work but the extrapolation may be really bad
hope it helps

[Scott Rick]

Thanks FernandoRios. Does that mean that this model does not work at all given my dataset? If so, is there any other method that you recommend?

[FernandoRios]

It works
you simply can’t use controls on it
am alternative could be using did2s or did_impute which use more data for the pretreatment identification and extrapolation.
however even with that you have only minimal room for controls.
the other option is sdid and use synthetic control did for this
hyh

[Scott Rick]

FernandoRios I apologize, I got sidetracked by another project and just returned to this. I looked up sdid (thank you, I was unaware of this), but since my data is an unbalanced panel it appears I cannot use it.

I returned to the csdid command to see if changing the time time variable from week to month would help, and get the result below which still has significant omissions. As seen in the tab of time variable va gvar, the previous problem of the sample being very small and the smallest cell size not allowing for any controls still stands. Is there any other approach that you recommend for an unbalanced panel with multiple time periods and different intervention start dates across units?

Output:

sdid rate_con_avg7day pop_perkm2_2019 povertyperct rate_test_avg7day per60_2011 ///
>litrate_2011 hospital_beds_total1k hospital_total1k htnanddm20181k commoncancers20181k ///
>hepcases_20181k malariacase2018 mentalillness_20111k  pneumoniatot_case2018    ///
> stroke20181k tb2018 exphealth_percap1516 typoidtot_case2018, ivar(statenum) time(month) gvar(gvar) 
Units always treated found. These will be ignored
Panel is not balanced
Will use observations with Pair balanced (observed at t0 and t1)
....................xxxxxxxxxxx.........xxxxxxxxxx

Difference-in-difference with Multiple Time Periods

                                                           Number of obs = 290
Outcome model  : least squares
Treatment model: inverse probability
------------------------------------------------------------------------------
             | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
g4           |
       t_3_4 |  -4.44e-15   2.06e-15    -2.16   0.031    -8.47e-15   -4.14e-16
       t_3_5 |  -6.66e-14   3.12e-14    -2.14   0.033    -1.28e-13   -5.48e-15
       t_3_6 |   2.68e-14   1.65e-14     1.63   0.103    -5.44e-15    5.91e-14
       t_3_7 |   4.39e-13   2.15e-13     2.04   0.041     1.78e-14    8.61e-13
       t_3_8 |   1.80e-12   8.41e-13     2.13   0.033     1.47e-13    3.44e-12
       t_3_9 |   3.26e-12   1.56e-12     2.09   0.037     2.04e-13    6.32e-12
      t_3_10 |   5.85e-12   2.80e-12     2.09   0.037     3.64e-13    1.13e-11
      t_3_11 |   5.02e-12   2.38e-12     2.11   0.035     3.46e-13    9.69e-12
      t_3_12 |   5.01e-12   2.39e-12     2.09   0.036     3.22e-13    9.70e-12
      t_3_13 |   5.30e-12   2.47e-12     2.14   0.032     4.55e-13    1.01e-11
-------------+----------------------------------------------------------------
g5           |
       t_3_4 |   6.53e-15   1.82e-15     3.59   0.000     2.97e-15    1.01e-14
       t_4_5 |  -.7811111   .8156601    -0.96   0.338    -2.379775    .8175533
       t_4_6 |   3.679993   3.234755     1.14   0.255    -2.660009       10.02
       t_4_7 |   14.33042   12.76803     1.12   0.262    -10.69446     39.3553
       t_4_8 |   50.88598    38.0864     1.34   0.182    -23.76198    125.5339
       t_4_9 |   106.4795   76.54976     1.39   0.164    -43.55531    256.5143
      t_4_10 |   201.3841   143.6211     1.40   0.161    -80.10814    482.8763
      t_4_11 |   185.7676   133.4797     1.39   0.164     -75.8479     447.383
      t_4_12 |   193.0112   138.0383     1.40   0.162    -77.53896    463.5613
      t_4_13 |   205.1946   145.8356     1.41   0.159    -80.63789    491.0271
-------------+----------------------------------------------------------------
g7           |
       t_3_4 |          0  (omitted)
       t_4_5 |          0  (omitted)
       t_5_6 |          0  (omitted)
       t_6_7 |          0  (omitted)
       t_6_8 |          0  (omitted)
       t_6_9 |          0  (omitted)
      t_6_10 |          0  (omitted)
      t_6_11 |          0  (omitted)
      t_6_12 |          0  (omitted)
      t_6_13 |          0  (omitted)
-------------+----------------------------------------------------------------
g8           |
       t_3_4 |          0  (omitted)
       t_4_5 |  -1.31e-13   3.82e-14    -3.43   0.001    -2.06e-13   -5.62e-14
       t_5_6 |  -4.861025    2.64531    -1.84   0.066    -10.04574    .3236876
       t_6_7 |  -7.99e-12   3.33e-12    -2.40   0.016    -1.45e-11   -1.47e-12
       t_7_8 |   1.75e-11   6.23e-12     2.81   0.005     5.28e-12    2.97e-11
       t_7_9 |   8.12e-11   2.88e-11     2.83   0.005     2.49e-11    1.38e-10
      t_7_10 |   1.27e-10   4.48e-11     2.84   0.005     3.92e-11    2.15e-10
      t_7_11 |   9.62e-11   3.47e-11     2.77   0.006     2.82e-11    1.64e-10
      t_7_12 |   9.26e-11   3.27e-11     2.83   0.005     2.84e-11    1.57e-10
      t_7_13 |   8.85e-11   3.17e-11     2.79   0.005     2.64e-11    1.51e-10
-------------+----------------------------------------------------------------
g9           |
       t_3_4 |          0  (omitted)
       t_4_5 |          0  (omitted)
       t_5_6 |          0  (omitted)
       t_6_7 |          0  (omitted)
       t_7_8 |          0  (omitted)
       t_8_9 |          0  (omitted)
      t_8_10 |          0  (omitted)
      t_8_11 |          0  (omitted)
      t_8_12 |          0  (omitted)
      t_8_13 |          0  (omitted)
------------------------------------------------------------------------------
Control: Never Treated

See Callaway and Sant'Anna (2021) for details


tab month gvar

           |                          gvar
     month |         0          3          4          5          7 |     Total
-----------+-------------------------------------------------------+----------
         3 |        14          2          3          4          1 |        26 
         4 |        16          2          4          5          1 |        31 
         5 |        17          2          4          5          1 |        32 
         6 |        17          2          4          5          1 |        32 
         7 |        17          2          4          5          1 |        32 
         8 |        17          2          4          5          1 |        32 
         9 |        17          2          4          5          1 |        32 
        10 |        17          2          4          5          1 |        32 
        11 |        17          2          4          5          1 |        32 
        12 |        17          2          4          5          1 |        32 
        13 |        17          2          4          5          1 |        32 
-----------+-------------------------------------------------------+----------
     Total |       183         22         43         54         11 |       345 


           |         gvar
     month |         8          9 |     Total
-----------+----------------------+----------
         3 |         1          1 |        26 
         4 |         2          1 |        31 
         5 |         2          1 |        32 
         6 |         2          1 |        32 
         7 |         2          1 |        32 
         8 |         2          1 |        32 
         9 |         2          1 |        32 
        10 |         2          1 |        32 
        11 |         2          1 |        32 
        12 |         2          1 |        32 
        13 |         2          1 |        32 
-----------+----------------------+----------
     Total |        21         11 |       345 

.

[FernandoRios]

One that is more flexible in terms of model specification for repeated crossection would be did_imputation.
It basically models your outcome using only not-treated data. and uses this to extrapolate into the treated units (during treatment).
Because of this, you can be more flexible adding controls, but only to certain extent.
HTH

[Scott Rick]

FernandoRios Thanks, Fernando. I did try out did_imputation and would greatly appreciate your guidance on interpreting the output. I've run three specifications of the command:

1. In #1 below, A very basic command that allows for lags and leads, but with no controls. I've used the minn(0) option to reduce the threshold for calculating the estimates based on to treated groups to zero. However, since my my pre-treatment parallel trends are conditional on covariates, this basic model with no covariates doesn't work. (Note: my understanding here is that like in csdid, including covariates allows for the “parallel trends assumption to hold potentially only after conditioning on observed covariates". Please do let me know if I have misinterpreted this)
2. In #2 below, I included all necessary covariates, but run into omitted output as seen below. Ideally, this would be the specification I would like to run, and so am not sure what to do.
3. In #3 below, I included only 2 covariates, but get omitted coefficients even for these two, with the tau coefficient being the same if I ran the specification without the controls.

did_imputation is very new to me and I'm having difficulty understanding what exactly is happening here. Does this mean that my sample size is just too small even for this model to include all the covariates? If so, how is it that even paring the specification down to include just two covariates is running into problems. A pared down version with csdid appeared to work fine, and my understanding is that this is a more flexible model.

Model 1:
did_imputation rate_con_avg7day state weeknum gvar_wk, horizons(0/10) pretrend(
> 10) minn(0)

                                                         Number of obs = 1,152
------------------------------------------------------------------------------
rate_con_a~y | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
        tau0 |  -.1664607   .1097231    -1.52   0.129     -.381514    .0485926
        tau1 |  -.1952425   .1403175    -1.39   0.164    -.4702598    .0797748
        tau2 |  -.1981032   .1835093    -1.08   0.280    -.5577749    .1615684
        tau3 |  -.1282566   .2306716    -0.56   0.578    -.5803647    .3238514
        tau4 |   -.043209   .2815442    -0.15   0.878    -.5950256    .5086076
        tau5 |   .0676714   .3273894     0.21   0.836        -.574    .7093429
        tau6 |   .1803828   .3717323     0.49   0.627    -.5481992    .9089647
        tau7 |   .2567773   .4053272     0.63   0.526    -.5376495    1.051204
        tau8 |     .32404   .4352558     0.74   0.457    -.5290457    1.177126
        tau9 |   .3963354   .4695793     0.84   0.399    -.5240232    1.316694
       tau10 |    .482609   .5089584     0.95   0.343    -.5149311    1.480149
        pre1 |  -.6511666   .3964881    -1.64   0.101    -1.428269    .1259357
        pre2 |  -.5971879   .3531467    -1.69   0.091    -1.289343    .0949668
        pre3 |  -.5550397   .3075969    -1.80   0.071    -1.157919    .0478391
        pre4 |  -.4942177   .2646653    -1.87   0.062    -1.012952    .0245168
        pre5 |  -.4695266   .2365316    -1.99   0.047      -.93312   -.0059332
        pre6 |  -.4570495   .2252221    -2.03   0.042    -.8984767   -.0156223
        pre7 |   -.328927   .1606409    -2.05   0.041    -.6437773   -.0140767
        pre8 |  -.3326807   .1495892    -2.22   0.026    -.6258701   -.0394913
        pre9 |  -.2803194   .1602777    -1.75   0.080    -.5944579    .0338191
       pre10 |  -.1949567   .1109105    -1.76   0.079    -.4123373    .0224238
------------------------------------------------------------------------------


Model 2:
did_imputation rate_con_avg7day state weeknum gvar_wk, ///
> controls (pop_perkm2_2019 povertyperct rate_test_avg7day per60_2011 litrate_201
> 1 ///
> hospital_beds_total1k hospital_total1k htnanddm20181k commoncancers20181k      
>          ///
> hepcases_20181k malariacase2018 mentalillness_20111k stroke20181k ///
> pneumoniatot_case2018 tb2018 exphealth_percap1516 typoidtot_case2018), ///

end of do-file

. did_imputation rate_con_avg7day state weeknum gvar_wk, controls (pop_perkm2_201
> 9 povertyperct rate_test_avg7day per60_2011 litrate_2011 hospital_beds_total1k 
> hospital_total1k htnanddm20181k commoncancers20181k hepcases_20181k malariacase
> 2018 mentalillness_20111k stroke20181k pneumoniatot_case2018 tb2018 exphealth_p
> ercap1516 typoidtot_case2018)

                                                         Number of obs = 1,533
--------------------------------------------------------------------------------
rate_con_avg~y | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
---------------+----------------------------------------------------------------
           tau |   .1436655   .8182949     0.18   0.861    -1.460163    1.747494
pop_perkm~2019 |          0  (omitted)
  povertyperct |          0  (omitted)
rate_test_av~y |   .0676723   .0253258     2.67   0.008     .0180347    .1173099
    per60_2011 |          0  (omitted)
  litrate_2011 |          0  (omitted)
hospital_be~1k |          0  (omitted)
hospital_to~1k |          0  (omitted)
htnanddm20181k |          0  (omitted)
commonc~20181k |          0  (omitted)
hepcase~20181k |          0  (omitted)
malariaca~2018 |          0  (omitted)
mentali~20111k |          0  (omitted)
  stroke20181k |          0  (omitted)
pne~t_case2018 |          0  (omitted)
        tb2018 |          0  (omitted)
expheal~ap1516 |          0  (omitted)
typoidtot_ca~8 |          0  (omitted)
--------------------------------------------------------------------------------

Model 3

did_imputation rate_con_avg7day state weeknum gvar_wk, controls (pop_perkm2_201
> 9 povertyperct)

                                                         Number of obs = 1,533
--------------------------------------------------------------------------------
rate_con_avg~y | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
---------------+----------------------------------------------------------------
           tau |   .2935677   1.057376     0.28   0.781    -1.778851    2.365987
pop_perkm~2019 |          0  (omitted)
  povertyperct |          0  (omitted)
--------------------------------------------------------------------------------

. did_imputation rate_con_avg7day state weeknum gvar_wk

                                                         Number of obs = 1,533
------------------------------------------------------------------------------
rate_con_a~y | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         tau |   .2935677   1.057376     0.28   0.781    -1.778851    2.365987
------------------------------------------------------------------------------

[FernandoRios]

Correct
you can’t add many covariants because your sample is just small
it may be easier to see that if you implement the methods manually

[Scott Rick]

FernandoRios Thank you. But it appears to be failing even when I add just two covariates as seen in model 3 in #9. The sample should be large enough for two covariates, I think. Would you also know if there is a guide on how to implement the methods manually?

Also, if I may continue to bother you - I've been trying to explore the flexpaneldid command, which appears to work when I use the nearest neighbor matching method. However, I appear to run into trouble on using the CEM matching method within flexpaneldid. Would you have any suggestions on this? I'd posted about it here a while back, but did not find a resolution: https://www.statalist.org/forums/for...o-flexpaneldid

[FernandoRios]

If you see the cross tab you show me earlier you will see some cells have 1 or 2 observations
that means that you can’t add controls with csdid or just very minimal with did_imputation (1? 2? At most)
don’t know flexdid but you can refer to the original papers for manual implementation
Asjad Navqi has a blog where he describes implementation for few of the strategies

[Scott Rick]

Thanks FernandoRios. I appreciate all the help!