Skip to main content
. 2005 Oct;40(5 Pt 1):1279–1296. doi: 10.1111/j.1475-6773.2005.00432.x

Table 5.

Difference-in-Differences Model of Physician Spending for Gainers and Nevers

Fixed-Effects (within) Regression
Group Variable (i): baseid
R2: within=0.0066
between=0.0242
overall=0.0098
Corr(u_i, Xb)=-0.3113
Number of obs=45,384
Number of groups=4,054
Obsper group: min=8
avg=11.0
max=16
F(14,41316)=19.68
Prob>F=0.0000
Total Physician Coefficent SE t p>∣t∣ [95% Confidence Interval]
y1996 9.100585 3.430918 2.65 0.008 2.375913 15.82526
y1997 14.25092 4.949754 2.88 0.004 4.549295 23.95254
y1998 22.28622 6.533179 3.41 0.001 9.481051 35.09139
y1999 29.73888 8.153666 3.65 0.000 13.75752 45.72024
y2000 33.8768 9.628832 3.52 0.000 15.00408 52.74951
Age −20.6048 20.90089 −0.99 0.324 −61.57099 20.36139
Age2 .122362 .1327581 0.92 0.357 −.1378466 .3825706
Married −17.61857 12.10421 −1.46 0.146 −41.34308 6.105949
Income −.0014219 .0547972 −0.03 0.979 −.1088255 .1059817
Metro −22.75821 37.74338 −0.60 0.547 −96.73604 51.21962
HCC Index 29.52116 3.312933 8.91 0.000 23.02774 36.01458
Postperiod 19.03616 2.345403 8.12 0.000 14.43912 23.6332
Gain .8919971 7.274963 0.12 0.902 −13.36709 15.15108
Post × Gain −2.738953 4.122824 −0.66 0.506 −10.81978 5.341871
_cons 900.9817 827.3328 1.09 0.276 −720.6082 2,522.572
sigma_u 94.073779
sigma_e 168.21478
rho .23824516 (fraction of variance because of u_i)