I understand the question is vague so I hope the following details would make it clearer.
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input str8 id float(round healthshock) str2 cluster_id int score double treatment_intensity float months_exposure "IN011001" 1 . "01" . 0 0 "IN011001" 2 0 "01" 115 0 0 "IN011001" 3 0 "01" 172 . 22.454794 "IN011001" 4 0 "01" . 644 70.55342 "IN011001" 5 0 "01" . 1052 105.79726 "IN011002" 1 . "01" . 0 0 "IN011002" 2 0 "01" 107 0 0 "IN011002" 3 0 "01" 141 . 22.52055 "IN011002" 4 0 "01" . 644 74.03835 "IN011002" 5 0 "01" . 1052 106.81644 "IN011003" 1 . "01" . 0 0 "IN011003" 2 0 "01" 113 0 0 "IN011003" 3 0 "01" 167 . 22.38904 "IN011003" 4 0 "01" . 644 70.25754 "IN011003" 5 0 "01" . 1052 105.79726 "IN011004" 1 . "01" . 0 0 "IN011004" 2 0 "01" 115 0 0 "IN011004" 3 0 "01" 142 . 22.52055 "IN011004" 4 0 "01" . 644 70.5863 "IN011004" 5 1 "01" . 1052 105.86301 "IN011005" 1 1 "01" . 0 0 "IN011005" 2 0 "01" 124 0 0 "IN011005" 3 1 "01" 140 . 22.48767 "IN011005" 4 0 "01" . 644 70.520546 "IN011005" 5 0 "01" . 1052 105.76438 "IN011006" 1 . "01" . 0 0 "IN011006" 2 0 "01" 125 0 0 "IN011006" 3 0 "01" 153 . 22.454794 "IN011006" 4 0 "01" . 644 71.83562 "IN011006" 5 0 "01" . 1052 106.5863 "IN011007" 1 . "01" . 0 0 "IN011007" 2 1 "01" 111 0 0 "IN011007" 3 1 "01" 149 . 22.652056 "IN011007" 4 0 "01" . 644 70.25754 "IN011007" 5 0 "01" . 1052 110.5315 "IN011008" 1 . "01" . 0 0 "IN011008" 2 1 "01" 123 0 0 "IN011008" 3 1 "01" 152 . 22.38904 "IN011008" 4 0 "01" . 644 70.158905 "IN011008" 5 0 "01" . 1052 105.76438 "IN011009" 1 . "01" . 0 0 "IN011009" 2 0 "01" 47 0 0 "IN011009" 3 0 "01" 39 . 22.224657 "IN011009" 4 0 "01" . 644 70.84931 "IN011009" 5 0 "01" . 1052 106.75069 "IN011010" 1 1 "01" . 0 0 "IN011010" 2 0 "01" 118 0 0 "IN011010" 3 0 "01" 158 . 22.68493 "IN011010" 4 0 "01" . 644 70.78356 "IN011010" 5 1 "01" . 1052 106.91507 "IN011011" 1 . "01" . 0 0 "IN011011" 2 0 "01" 109 0 0 "IN011011" 3 0 "01" 129 . 21.961643 "IN011011" 4 0 "01" . 644 70.454796 "IN011011" 5 0 "01" . 1052 105.8959 "IN011012" 1 1 "01" . 0 0 "IN011012" 2 0 "01" 116 0 0 "IN011012" 3 0 "01" 111 . 22.52055 "IN011012" 4 0 "01" . 644 70.78356 "IN011012" 5 0 "01" . 1052 105.8959 "IN011013" 1 . "01" . 0 0 "IN011013" 2 0 "01" 116 0 0 "IN011013" 3 0 "01" 185 . 22.356165 "IN011013" 4 1 "01" . 644 70.520546 "IN011013" 5 1 "01" . 1052 105.76438 "IN011014" 1 . "01" . 0 0 "IN011014" 2 1 "01" 90 0 0 "IN011014" 3 1 "20" 148 . 15.156164 "IN011014" 4 0 "01" . 644 71.178085 "IN011014" 5 0 "01" . 1052 106.71781 "IN011015" 1 . "01" . 0 0 "IN011015" 2 0 "01" 110 0 0 "IN011015" 3 0 "01" 132 . 22.060274 "IN011015" 4 0 "01" . 644 71.93425 "IN011015" 5 0 "01" . 1052 106.0274 "IN011016" 1 . "01" . 0 0 "IN011016" 2 0 "01" 95 0 0 "IN011016" 3 0 "01" 114 . 22.19178 "IN011016" 4 0 "01" . 644 70.224655 "IN011016" 5 1 "01" . 1052 105.76438 "IN011017" 1 . "01" . 0 0 "IN011017" 2 1 "01" 121 0 0 "IN011018" 1 . "01" . 0 0 "IN011018" 2 0 "01" 113 0 0 "IN011018" 3 0 "01" 106 . 22.0274 "IN011018" 4 0 "01" . 644 70.520546 "IN011018" 5 0 "01" . 1052 105.79726 "IN011019" 1 . "01" . 0 0 "IN011019" 2 0 "01" 119 0 0 "IN011019" 3 0 "01" 176 . 23.375343 "IN011019" 4 0 "01" . 644 71.63836 "IN011019" 5 0 "01" . 1052 106.55342 "IN011020" 1 . "01" . 0 0 "IN011021" 1 1 "01" . 0 0 "IN011021" 2 0 "01" 120 0 0 "IN011021" 3 0 "01" 140 . 22.158905 "IN011021" 4 0 "01" . 644 70.454796 "IN011021" 5 0 "01" . 1052 105.83014 "IN011022" 1 . "01" . 0 0 "IN011022" 2 0 "01" 121 0 0 end
My panel data has rounds 1-5. The variable score is available for rounds 2 and 3. health shock,Treatment intensity and months_exposure is available for rounds 1-5.
While performing regressions I noticed something peculiar in the way results are being displayed and my query relates to that. For illustration purpose, please consider the following results:
(1)
Code:
areg score healthshock i.round,a( id )
Linear regression, absorbing indicators Number of obs = 1,912
F( 2, 919) = 699.53
Prob > F = 0.0000
R-squared = 0.8230
Adj R-squared = 0.6320
Root MSE = 23.5217
------------------------------------------------------------------------------
score | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
healthshock | -7.327439 2.164948 -3.38 0.001 -11.57626 -3.078622
3.round | 40.5999 1.098226 36.97 0.000 38.44457 42.75522
_cons | 91.42977 .8498429 107.58 0.000 89.76191 93.09762
-------------+----------------------------------------------------------------
id | F(990, 919) = 2.865 0.000 (991 categories)(2)
Code:
areg score treatment_intensity i.round,a( cluster_id ) cluster( cluster_id )
Linear regression, absorbing indicators Number of obs = 1,550
F( 2, 19) = 104.52
Prob > F = 0.0000
R-squared = 0.4112
Adj R-squared = 0.4031
Root MSE = 28.7665
(Std. Err. adjusted for 20 clusters in cluster_id)
-------------------------------------------------------------------------------------
| Robust
score | Coef. Std. Err. t P>|t| [95% Conf. Interval]
--------------------+----------------------------------------------------------------
treatment_intensity | .1362871 .082724 1.65 0.116 -.0368563 .3094305
3.round | 36.34813 6.293192 5.78 0.000 23.17633 49.51993
_cons | 89.9027 1.233783 72.87 0.000 87.32037 92.48504
--------------------+----------------------------------------------------------------
cluster_id | absorbed (20 categories)(3)
Code:
areg score healthshock##c.treatment_intensity i.round,a(id) cluster( cluster_id )
Linear regression, absorbing indicators Number of obs = 1,548
F( 4, 19) = 30.59
Prob > F = 0.0000
R-squared = 0.8643
Adj R-squared = 0.6252
Root MSE = 22.7989
(Std. Err. adjusted for 20 clusters in cluster_id)
---------------------------------------------------------------------------------------------------
| Robust
score | Coef. Std. Err. t P>|t| [95% Conf. Interval]
----------------------------------+----------------------------------------------------------------
1.healthshock | -6.939664 5.988231 -1.16 0.261 -19.47317 5.593847
treatment_intensity | .1486624 .1219357 1.22 0.238 -.106552 .4038768
|
healthshock#c.treatment_intensity |
1 | -.0561772 .1138478 -0.49 0.627 -.2944634 .182109
|
round |
1 | 0 (empty)
2 | -35.80429 9.806665 -3.65 0.002 -56.32987 -15.2787
3 | 0 (omitted)
4 | 0 (empty)
5 | 0 (empty)
|
_cons | 126.8508 8.48484 14.95 0.000 109.0918 144.6098
----------------------------------+----------------------------------------------------------------
id | absorbed (984 categories)Moreover, these empty rounds are no longer shown if don't cluster the standard errors, as can be seen below
(4)
Code:
areg score healthshock##c.treatment_intensity i.round,a(id)
Linear regression, absorbing indicators Number of obs = 1,548
F( 4, 560) = 271.37
Prob > F = 0.0000
R-squared = 0.8643
Adj R-squared = 0.6252
Root MSE = 22.7989
---------------------------------------------------------------------------------------------------
score | Coef. Std. Err. t P>|t| [95% Conf. Interval]
----------------------------------+----------------------------------------------------------------
1.healthshock | -6.939664 2.917619 -2.38 0.018 -12.67048 -1.208849
treatment_intensity | .1486624 .0314199 4.73 0.000 .0869472 .2103776
|
healthshock#c.treatment_intensity |
1 | -.0561772 .0967086 -0.58 0.562 -.2461331 .1337787
|
3.round | 35.80429 2.203127 16.25 0.000 31.47688 40.13169
_cons | 91.04651 .9460822 96.24 0.000 89.18821 92.90481
----------------------------------+----------------------------------------------------------------
id | F(983, 560) = 2.260 0.000 (984 categories)Also, I'm not sure if this is only a display issue or this has implications for my results as well.
Would appreciate any help in this regard.
Thanks,
0 Response to Why is stata reporting rounds not used in the analysis?
Post a Comment