pweight
is est_wt
strata
is strata
.
svyprop eo
------------------------------------------------------------------------------
pweight: est_wt Number of
obs = 2184
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
2184
Population
size = 2159.0837
------------------------------------------------------------------------------
Survey
proportions estimation
+------------------------------------+
| eo
Obs Est. Prop. Std. Err. |
|------------------------------------|
|
0 412 0.324695
0.018071 |
|
1 1772 0.675305
0.018071 |
+------------------------------------+
.
/*--------------------------------------------------------------
>
now weighted table of proportions woth EOP by size of workplace
>
------------------------------------------------------------*/
.
svytab eo nempsize, column
pweight: est_wt Number of
obs = 2184
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
2184
Population size = 2159.0837
-------------------------------------------------------------------------
equal |
opps |
policy |
recoded | nemp size
to
0 1 | 1
2 3 4
5 6 7
8 Total
----------+--------------------------------------------------------------
0 | .3627 .343
.2934 .1847 .1294
.1126 .0196 .0227
.3247
1 | .6373 .657
.7066 .8153 .8706
.8874 .9804 .9773
.6753
|
Total |
1 1 1
1 1 1
1 1 1
-------------------------------------------------------------------------
Key:
column proportions
Pearson:
Uncorrected chi2(7) =
44.5838
Design-based F(2.33, 4945.51)= 6.6138
P = 0.0007
.
/*--------------------------------------------------------------
>
now unweighted linear models to get effect of factors on EOP
>
------------------------------------------------------------*/
.
svyset [pweight=const]
pweight
is const
strata
is strata
.
svyregress female eo
Survey
linear regression
pweight: const Number of
obs = 2169
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
2169
Population size = 2169
F( 1, 2103)
= 57.87
Prob > F =
0.0000
R-squared = 0.0232
------------------------------------------------------------------------------
female | Coef.
Std. Err. t P>|t|
[95% Conf. Interval]
-------------+----------------------------------------------------------------
eo |
11.12066 1.461883 7.61
0.000 8.253773 13.98755
_cons |
40.65796 1.278403 31.80
0.000 38.15089 43.16503
------------------------------------------------------------------------------
.
svyregress disab eo
Survey
linear regression
pweight: const Number of
obs = 2074
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
2074
Population size = 2074
F( 1,
2008) = 4.29
Prob > F = 0.0385
R-squared = 0.0010
------------------------------------------------------------------------------
disab | Coef.
Std. Err. t P>|t|
[95% Conf. Interval]
-------------+----------------------------------------------------------------
eo |
.2120406 .1023986 2.07
0.039 .0112221 .4128592
_cons |
.7256245 .0757176 9.58
0.000 .5771312 .8741178
------------------------------------------------------------------------------
.
svyregress ethnic eo
Survey
linear regression
pweight: const Number of
obs = 2063
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
2063
Population
size = 2063
F( 1, 1997)
= 17.52
Prob > F = 0.0000
R-squared =
0.0062
------------------------------------------------------------------------------
ethnic | Coef.
Std. Err. t P>|t|
[95% Conf. Interval]
-------------+----------------------------------------------------------------
eo |
2.104745 .5027954 4.19
0.000 1.118686 3.090803
_cons |
3.190257 .4276565 7.46
0.000 2.351557 4.028956
------------------------------------------------------------------------------
.
/*--------------------------------------------------------------
>
now weighted linear models
>
------------------------------------------------------------*/
.
svyset [pweight=est_wt]
pweight
is est_wt
strata
is strata
.
svyregress female eo
Survey
linear regression
pweight: est_wt Number of
obs = 2169
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
2169
Population size = 2153.8422
F( 1, 2103)
= 37.86
Prob > F =
0.0000
R-squared = 0.0704
------------------------------------------------------------------------------
female | Coef.
Std. Err. t P>|t|
[95% Conf. Interval]
-------------+----------------------------------------------------------------
eo |
17.63143 2.865355 6.15
0.000 12.0122 23.25065
_cons |
42.40965 2.474206 17.14
0.000 37.55751
47.2618
------------------------------------------------------------------------------
.
svyregress disab eo
Survey
linear regression
pweight: est_wt Number of
obs = 2074
Strata: strata
Number
of strata = 66
PSU: <observations> Number of PSUs =
2074
Population size = 2131.5835
F(
1, 2008) =
0.22
Prob > F = 0.6424
R-squared = 0.0004
------------------------------------------------------------------------------
disab | Coef.
Std. Err. t P>|t|
[95% Conf. Interval]
-------------+----------------------------------------------------------------
eo |
.1026639 .2210432 0.46
0.642 -.3308341 .5361618
_cons |
.7534789 .1829337 4.12
0.000 .3947191 1.112239
------------------------------------------------------------------------------
.
svyregress ethnic eo
Survey
linear regression
pweight: est_wt Number of obs =
2063
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
2063
Population size = 2126.2916
F( 1, 1997)
= 14.14
Prob > F = 0.0002
R-squared = 0.0166
------------------------------------------------------------------------------
ethnic | Coef.
Std. Err. t P>|t|
[95% Conf. Interval]
-------------+----------------------------------------------------------------
eo | 2.74197
.72911 3.76 0.000
1.312074 4.171866
_cons |
2.373978 .4754587 4.99
0.000 1.441531 3.306425
------------------------------------------------------------------------------
.
/*--------------------------------------------------------------
>
now unweighted multivariate logistic models
>
------------------------------------------------------------*/
.
svyset [pweight=const]
pweight
is const
strata
is strata
.
svylogit eo female disab ethnic
Survey
logistic regression
pweight: const Number of
obs = 2024
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs = 2024
Population size = 2024
F( 3, 1956)
= 21.89
Prob > F = 0.0000
------------------------------------------------------------------------------
eo | Coef.
Std. Err. t P>|t|
[95% Conf. Interval]
-------------+----------------------------------------------------------------
female |
.0139578 .0019384 7.20
0.000 .0101562 .0177593
disab | .032551
.0316652 1.03 0.304
-.02955 .0946519
ethnic |
.0245636 .0097939 2.51
0.012 .005356 .0437713
_cons |
.6657324 .1027319 6.48
0.000 .4642571
.8672078
------------------------------------------------------------------------------
.
/*--------------------------------------------------------------
>
now weighted multivariate logistic models
>
------------------------------------------------------------*/
.
svyset [pweight=est_wt]
pweight
is est_wt
strata
is strata
.
svylogit eo female disab ethnic
Survey
logistic regression
pweight: est_wt Number of
obs = 2024
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
2024
Population size = 2119.4079
F( 3,
1956) = 13.83
Prob > F = 0.0000
------------------------------------------------------------------------------
eo | Coef.
Std. Err. t P>|t|
[95% Conf. Interval]
-------------+----------------------------------------------------------------
female |
.0190382 .0032781 5.81
0.000 .0126093 .0254671
disab |
-.0047271 .0358489 -0.13
0.895 -.0750331
.0655789
ethnic |
.0372057 .0143067 2.60
0.009 .0091477 .0652637
_cons |
-.3738791 .1798482 -2.08
0.038 -.726593 -.0211651
------------------------------------------------------------------------------
.
/*--------------------------------------------------------------
>
now unweighted multivariate logistic model with NEMPSIZE
>
need to use xi: to code up dummy variables then run analyses
>
------------------------------------------------------------*/
.
svyset [pweight=const]
pweight
is const
strata
is strata
.
xi: svylogit eo female disab ethnic i.nempsize
i.nempsize _Inempsize_1-8 (naturally coded; _Inempsize_1 omitted)
note:
_Inempsize_7 != 0 predicts success perfectly
_Inempsize_7 dropped and 55 obs not used
note:
_Inempsize_8 != 0 predicts success perfectly
_Inempsize_8 dropped and 29 obs not used
Survey
logistic regression
pweight: const Number of
obs = 1940
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
1940
Population size = 1940
F( 8,
1867) = 19.28
Prob > F = 0.0000
------------------------------------------------------------------------------
eo | Coef.
Std. Err. t P>|t|
[95% Conf. Interval]
-------------+----------------------------------------------------------------
female |
.0176531 .0020094 8.79
0.000 .0137121 .021594
disab | .019558
.021002 0.93 0.352
-.0216319 .0607478
ethnic |
.0170366 .0088342 1.93
0.054 -.0002892 .0343625
_Inempsize_2
| .1417863 .179564
0.79 0.430 -.2103801
.4939528
_Inempsize_3
| .585323 .1878167
3.12 0.002 .216971
.9536749
_Inempsize_4
| 1.155253 .2091719
5.52 0.000 .7450186
1.565487
_Inempsize_5
| 1.668426 .2202926
7.57 0.000 1.236381 2.10047
_Inempsize_6
| 1.415234 .2846374
4.97 0.000 .8569947
1.973474
_cons |
-.2367999 .1762974 -1.34
0.179 -.5825597 .1089599
------------------------------------------------------------------------------
.
/*--------------------------------------------------------------
>
same thing weighted
>
------------------------------------------------------------*/
.
svyset [pweight=est_wt]
pweight
is est_wt
strata
is strata
.
xi: svylogit eo female disab ethnic i.nempsize
i.nempsize _Inempsize_1-8 (naturally coded; _Inempsize_1 omitted)
note:
_Inempsize_7 != 0 predicts success perfectly
_Inempsize_7 dropped and 55 obs not used
note:
_Inempsize_8 != 0 predicts success perfectly
_Inempsize_8 dropped and 29 obs not used
Survey
logistic regression
pweight: est_wt Number of obs =
1940
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
1940
Population size = 2111.6086
F( 8, 1867)
= 12.51
Prob > F = 0.0000
------------------------------------------------------------------------------
eo |
Coef. Std. Err. t
P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
female |
.0209044 .0033645 6.21
0.000 .0143058 .0275031
disab |
-.0116614 .03635 -0.32
0.748 -.0829521 .0596293
ethnic | .03552
.0139514 2.55 0.011
.008158 .0628819
_Inempsize_2
| .1070193 .2177569
0.49 0.623 -.3200522
.5340909
_Inempsize_3
| .5146863 .2121884
2.43 0.015 .098536
.9308367
_Inempsize_4
| 1.161438 .2323457
5.00 0.000 .7057545
1.617121
_Inempsize_5
| 1.671919 .2585829
6.47 0.000 1.164779 2.17906
_Inempsize_6
| 1.709645 .3174224
5.39 0.000
1.087106 2.332184
_cons |
-.6780242 .2313491 -2.93
0.003 -1.131753 -.2242953
------------------------------------------------------------------------------
.
/*--------------------------------------------------------------
>
now mean of eo and by size group allowing for finite population
>
correction
>
Stata requires two things to get the finite population correct
>
1. A variable with either the sampling farction or the number of PSUs in the
stratum
>
2. Weights that add to the population size
>
------------------------------------------------------------*/
.
svyset [pweight=est_wt], strata(strata) clear(fpc)
pweight
is est_wt
strata
is strata
.
svymean eo ,deff deft
Survey
mean estimation
pweight: est_wt Number of
obs = 2184
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
2184
Population size = 2159.0837
------------------------------------------------------------------------------
Mean |
Estimate Std. Err. Deff Deft
---------+--------------------------------------------------------------------
eo |
.675305 .0180714 3.251329
1.803144
------------------------------------------------------------------------------
.
gen gweight=199952/2191
.
svyset [pweight=gweight] ,fpc(sampfrac) strata(strata)
pweight
is gweight
strata
is strata
fpc
is sampfrac
.
svymean eo, deff deft
Survey
mean estimation
pweight: gweight Number of
obs = 2184
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
2184
FPC: sampfrac Population
size = 199313.18
------------------------------------------------------------------------------
Mean |
Estimate Std. Err. Deff Deft
---------+--------------------------------------------------------------------
eo |
.8113553 .0076138 .8359598
.9092853
------------------------------------------------------------------------------
Finite
population correction (FPC) assumes simple random sampling without
replacement
of PSUs within each stratum with no subsampling within PSUs.
Weights
must represent population totals for deff to be correct when
using
an FPC. Note: deft is invariant to the
scale of weights.
/*--------------------------------------------------------------------
bit
that follows illustrates the different definitions of the design effect for
subgroups
----------------------------------------------------------------------------*/
.
svymean eo, deff deft by(nempsize)
Survey
mean estimation
pweight: gweight Number of
obs = 2184
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
2184
FPC: sampfrac Population
size = 199313.18
------------------------------------------------------------------------------
Mean Subpop. |
Estimate Std. Err. Deff Deft
---------------+--------------------------------------------------------------
eo |
nempsize==1 | .6911197
.027551 .9307145 .9594353
nempsize==2 | .7171717
.0218021 .9378447 .9631034
nempsize==3 | .774359
.0202909 .9287394 .9584167
nempsize==4 | .8501292
.0175549 .9460063 .967285
nempsize==5 | .8903509
.0140018 .925454 .95672
nempsize==6 | .8852459
.0216589 .8540405 .919066
nempsize==7 | .9722222
.0173717 .8130946 .8967636
nempsize==8 | .9756098
.022264 .863143 .9239507
------------------------------------------------------------------------------
Finite
population correction (FPC) assumes simple random sampling without
replacement
of PSUs within each stratum with no subsampling within PSUs.
Weights
must represent population totals for deff to be correct when
using
an FPC. Note: deft is invariant to the
scale of weights.
.
svymean eo, deff deft
by(nempsize)srssubpop
Survey
mean estimation
pweight: gweight Number of
obs = 2184
Strata: strata Number of
strata = 66
PSU: <observations> Number of PSUs =
2184
FPC: sampfrac Population
size = 199313.18
------------------------------------------------------------------------------
Mean Subpop. |
Estimate Std. Err. Deff Deft
---------------+--------------------------------------------------------------
eo |
nempsize==1 | .6911197
.027551 .9275457 .9578006
nempsize==2 | .7171717
.0218021 .935905 .9621069
nempsize==3 | .774359
.0202909 .9267824 .9574064
nempsize==4 | .8501292
.0175549 .943994 .9662557
nempsize==5 | .8903509
.0140018 .9238475 .9558893
nempsize==6 | .8852459
.0216589 .8497627 .9167613
nempsize==7 | .9722222
.0173717 .8021689 .8907183
nempsize==8 | .9756098
.022264 .8424765 .9128225
------------------------------------------------------------------------------
Finite
population correction (FPC) assumes simple random sampling without
replacement
of PSUs within each stratum with no subsampling within PSUs.
Weights
must represent population totals for deff to be correct when
using
an FPC. Note: deft is invariant to the
scale of weights.
.
.
.
end
of do-file
.