Interrupted time series-xtitsa

[amera amery]

Hello,

I am using XTITSA command to study the intervention on the behavior of Covid-19 cases the code is as below:

xtitsa Avg_cases i.calender_week_from_opening    i._x0#c.less_20 i._x0#c.age_20_39 i._x0#c.age_40_59 i._x0#c.age_60_84 i._x0#c.age_over_85  i._x0##i.income_median  i._x0#c.non_hispanic_latin_percent i._x0#c.hispanic_latin_percent i._x0#c.others_percent i._x0#c.white_only_percent i._x0#c.blk_african_amr_percent  i._x0#c.other_races_percent , single  trperiod(0) vce(robust) posttrend figure replace

the problem is that the chi2 test return no result"." as shown below, the second issue that I got wide CI for _x0 coefficient.

Panel variable: fips (unbalanced)
 Time variable: weeks, -4 to 12
         Delta: 1 unit
note: 1._x0 omitted because of collinearity.

Iteration 1: tolerance = 8.2533026
Iteration 2: tolerance = .27798123
Iteration 3: tolerance = .00738007
Iteration 4: tolerance = .00021537
Iteration 5: tolerance = 6.315e-06
Iteration 6: tolerance = 1.646e-07

GEE population-averaged model                         Number of obs    = 1,905
Group variable: fips                                  Number of groups =   114
Family: Gaussian                                      Obs per group:  
Link:   Identity                                                   min =     5
Correlation: exchangeable                                          avg =  16.7
                                                                   max =    17
                                                      Wald chi2(70)    =     .
Scale parameter = 242.8472                            Prob > chi2      =     .

                                                       (Std. err. adjusted for clustering on fips)
--------------------------------------------------------------------------------------------------
                                 |               Robust
                      _Avg_cases | Coefficient  std. err.      z    P>|z|     [95% conf. interval]
---------------------------------+----------------------------------------------------------------
                              _t |   .7490542   .6613248     1.13   0.257    -.5471185    2.045227
                             _x0 |   1232.655   925.8409     1.33   0.183      -581.96     3047.27
                           _x_t0 |  -.7143659    .903106    -0.79   0.429    -2.484421    1.055689
                                 |
      calender_week_from_opening |
                             12  |    6.10015   1.583337     3.85   0.000     2.996867    9.203433
                             13  |   8.807565   1.867446     4.72   0.000     5.147438    12.46769
                             14  |   15.56154   2.614388     5.95   0.000     10.43744    20.68565
                             15  |     17.097   3.447891     4.96   0.000     10.33926    23.85474
                             16  |   11.66626   3.748177     3.11   0.002     4.319965    19.01255
                             17  |   12.30973   4.018168     3.06   0.002      4.43427     20.1852
                             18  |   10.47149   4.630941     2.26   0.024     1.395013    19.54797
                             19  |   8.610292   5.071487     1.70   0.090    -1.329639    18.55022
                             20  |   3.386195   5.573258     0.61   0.543    -7.537189    14.30958
                             21  |    2.98615   5.868291     0.51   0.611     -8.51549    14.48779
                             22  |   1.925049   6.277048     0.31   0.759    -10.37774    14.22784
                             23  |   1.453094   6.766818     0.21   0.830    -11.80963    14.71581
                             24  |  -.2908339   7.218285    -0.04   0.968    -14.43841    13.85675
                             25  |  -.1396369   7.853128    -0.02   0.986    -15.53148    15.25221
                             26  |   .2681749   8.381983     0.03   0.974    -16.16021    16.69656
                             27  |   .2163127   8.825185     0.02   0.980    -17.08073    17.51336
                             28  |   2.806489   9.143373     0.31   0.759    -15.11419    20.72717
                             29  |   3.590357   9.617246     0.37   0.709     -15.2591    22.43981
                             30  |   5.980221   10.16806     0.59   0.556    -13.94881    25.90925
                             31  |   9.937396   11.39831     0.87   0.383    -12.40289    32.27768
                             32  |   2.912618   11.14749     0.26   0.794    -18.93606     24.7613
                             41  |   61.63976   11.10608     5.55   0.000     39.87223    83.40728
                             42  |   56.33374   15.48673     3.64   0.000      25.9803    86.68717
                             43  |   72.50997   12.52376     5.79   0.000     47.96384    97.05609
                             44  |   77.22091   15.33212     5.04   0.000     47.17051    107.2713
                             45  |   93.09452   16.31863     5.70   0.000      61.1106    125.0784
                             46  |   114.6563   11.83311     9.69   0.000     91.46386    137.8488
                             47  |   137.2613    8.33408    16.47   0.000     120.9268    153.5958
                             48  |   136.2834   10.75055    12.68   0.000     115.2127    157.3541
                             49  |   100.6777   9.125954    11.03   0.000     82.79119    118.5643
                             50  |   76.45914   8.667602     8.82   0.000     59.47095    93.44733
                             51  |   60.05309    7.56156     7.94   0.000     45.23271    74.87348
                             52  |   56.65902   9.143364     6.20   0.000     38.73835    74.57968
                             53  |   40.00331   7.354229     5.44   0.000     25.58928    54.41733
                             54  |   42.67515   8.206872     5.20   0.000     26.58998    58.76032
                             55  |   56.11203   9.980569     5.62   0.000     36.55047    75.67358
                             56  |   43.82822   9.115273     4.81   0.000     25.96262    61.69383
                             57  |    26.8053   8.294249     3.23   0.001     10.54887    43.06173
                             58  |   20.33176   9.004612     2.26   0.024     2.683044    37.98047
                             59  |   12.25771   9.041326     1.36   0.175    -5.462961    29.97838
                             60  |   8.001382    8.86166     0.90   0.367    -9.367153    25.36992
                             61  |   10.28908   9.488877     1.08   0.278    -8.308783    28.88693
                             62  |   11.42351   10.50192     1.09   0.277    -9.159879    32.00691
                                 |
                   _x0#c.less_20 |
                              0  |    16.1055   10.37929     1.55   0.121    -4.237539    36.44853
                              1  |   2.694422   6.039775     0.45   0.656     -9.14332    14.53216
                                 |
                 _x0#c.age_20_39 |
                              0  |   16.95458    10.3578     1.64   0.102    -3.346334    37.25549
                              1  |   1.323924   5.836164     0.23   0.821    -10.11475     12.7626
                                 |
                 _x0#c.age_40_59 |
                              0  |   16.58113   10.46101     1.59   0.113    -3.922073    37.08434
                              1  |   1.690815     5.9803     0.28   0.777    -10.03036    13.41199
                                 |
                 _x0#c.age_60_84 |
                              0  |   15.41947   10.23532     1.51   0.132    -4.641387    35.48032
                              1  |   1.566354    5.88014     0.27   0.790    -9.958507    13.09122
                                 |
               _x0#c.age_over_85 |
                              0  |   22.33079   11.38901     1.96   0.050     .0087275    44.65284
                              1  |   3.104724   6.156173     0.50   0.614    -8.961154     15.1706
                                 |
                           1._x0 |          0  (omitted)
                                 |
                   income_median |
                              2  |   -9.03515   5.129031    -1.76   0.078    -19.08787    1.017567
                              3  |  -8.121615   6.938838    -1.17   0.242    -21.72149    5.478257
                                 |
               _x0#income_median |
                            1 2  |   10.50574   5.709952     1.84   0.066     -.685561    21.69704
                            1 3  |   6.923233   6.676728     1.04   0.300    -6.162913    20.00938
                                 |
_x0#c.non_hispanic_latin_percent |
                              0  |   -2.42185   2.934019    -0.83   0.409    -8.172422    3.328723
                              1  |   1.329811   1.955834     0.68   0.497    -2.503554    5.163175
                                 |
    _x0#c.hispanic_latin_percent |
                              0  |  -2.378546   2.924871    -0.81   0.416    -8.111187    3.354096
                              1  |   1.417435   1.949878     0.73   0.467    -2.404256    5.239126
                                 |
            _x0#c.others_percent |
                              0  |  -2.088937   2.911955    -0.72   0.473    -7.796263     3.61839
                              1  |   1.385719   1.935896     0.72   0.474    -2.408568    5.180005
                                 |
        _x0#c.white_only_percent |
                              0  |  -1.823048   2.527546    -0.72   0.471    -6.776948    3.130852
                              1  |  -3.348394   1.286855    -2.60   0.009    -5.870584   -.8262046
                                 |
   _x0#c.blk_african_amr_percent |
                              0  |  -2.158053   2.560706    -0.84   0.399    -7.176944    2.860838
                              1  |  -3.103864   1.281219    -2.42   0.015    -5.615007     -.59272
                                 |
       _x0#c.other_races_percent |
                              0  |  -1.857145   2.534325    -0.73   0.464    -6.824331    3.110041
                              1  |  -3.240256   1.283958    -2.52   0.012    -5.756767   -.7237451
                                 |
                           _cons |  -1221.086       1075    -1.14   0.256    -3328.046    885.8743
--------------------------------------------------------------------------------------------------


                    Postintervention Linear Trend: 0

Treated: _b[_t]+_b[_x_t0]
------------------------------------------------------------------------------
Linear Trend |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
     Treated |   .0346882   .6493813     0.05   0.957    -1.238076    1.307452
------------------------------------------------------------------------------

Any suggestions to resolve this issue?

[Jared Greathouse]

Why is your panel so unbalanced?

[amera amery]

I used two variables panel data county_fips and time : some counties will have 17 observations and others will have 16 observations. I do not think this is the problem because I am running same code with another group of counties and its work. I noticed when I remove i.week_calender variable the chi2 is estimated but CI still very wide.

[Jared Greathouse]

Oh I didn't think that was the problem, I was just curious about why it was so unbalanced.


Anyways, your real issue seems to be all of the one way interaction terms you include. Why do you include so many? And also, what policy are you evaluating anyways? Gimme details: how many units are treated versus untreated?

[amera amery]

I am studying the mask mandate policy, so want to see how all variables behave after and before the mask policy. for instance, the number of people with age>85 significantly increased the number of covid-19 (22.33 p-value 0.05) but after the policy was implemented this effectiveness is no more significant (3.11 with p_value .61) that is why I include many interaction terms.

[amera amery]

Even I removed all variables and try with :

 
 xtitsa Avg_cases i.calender_week_from_opening     i._x0##i.income_median   , single  trperiod(0) vce(robust) posttrend figure replace

I got better CI but still unable to estimate chi2 test.

[Jared Greathouse]

I have a much better suggestion for you, in this case. I think it'll take away some of the major issues there are with ITSA. May I suggest a slight addendum? amera amery

[amera amery]

Thank you Jared. Yes sure. I just have an update. I read in one forum we can get chi2 by removing
vce(robust) from the previous command. I tried it with having many variables and I got the chi2 results. What makes me uncomfortable with my result is the high value and CI for _x0

[Jared Greathouse]

So wait.... do you have only one treated unit? Is that all we're working with here?

[amera amery]

Yes. Mask Treatment in my model. Is this what you mean? But I have 3000 counties I want to check the impact of treatment over them