Hi Folks,
I am somewhat new to mixed, and have a question about fairly large confidence intervals. Anything I can do about it? Here is a description of my model and some code. Thanks.
Data: unbalanced panel dataset of depth to water for irrigation wells (from the years 1910-2013), which are situated in counties.
Explanatory variables: At the county level: the population density of the county (which changes by year); At the well level: the altitude of the well, the distance of the well from the river (both of which don't change); and then I have dummy variables for the drought or wet conditions of that particular year (drought1, drought2, wet1, wet2, with wet3 excluded).
Performing EM optimization:
Performing gradient-based optimization:
Iteration 0: log pseudolikelihood = 21291.913
Iteration 1: log pseudolikelihood = 21350.855
Iteration 2: log pseudolikelihood = 21356.944
Iteration 3: log pseudolikelihood = 21356.946
Computing standard errors:
Mixed-effects regression Number of obs = 32,169
No. of Observations per Group
Group Variable Groups Minimum Average Maximum
countyid 31 1 1,037.7 5,400
id2 2,300 1 14.0 70
Wald chi2(7) = 47.18
Log pseudolikelihood = 21356.946 Prob > chi2 = 0.0000
(Std. Err. adjusted for 31 clusters in countyid)
Robust
lnper_cdtw100 Coef. Std. Err. z P>z [95% Conf. Interval]
lnpopden_county .0014967 .0008484 1.76 0.078 -.0001661 .0031595
lnalt_avg -.0164247 .0072642 -2.26 0.024 -.0306622 -.0021872
lnriv_km2 .0014593 .0008141 1.79 0.073 -.0001362 .0030549
drought1 .0074959 .0020855 3.59 0.000 .0034083 .0115834
drought2 .0044572 .0070613 0.63 0.528 -.0093827 .0182971
wet0 -.2147777 .0598072 -3.59 0.000 -.3319976 -.0975578
wet1 .018222 .0059625 3.06 0.002 .0065358 .0299082
_cons 4.757492 .0614828 77.38 0.000 4.636988 4.877996
Robust
Random-effects Parameters Estimate Std. Err. [95% Conf. Interval]
countyid: Independent
var(lnpopd~y) 1.54e-06 .0000194 3.16e-17 75331.21
var(_cons) .0000168 .0000483 6.06e-08 .0046738
id2: Identity
var(_cons) 7.06e-16 9.36e-14 1.2e-128 4.21e+97
var(Residual) .0155087 .0027515 .0109536 .0219581
I am somewhat new to mixed, and have a question about fairly large confidence intervals. Anything I can do about it? Here is a description of my model and some code. Thanks.
Data: unbalanced panel dataset of depth to water for irrigation wells (from the years 1910-2013), which are situated in counties.
Explanatory variables: At the county level: the population density of the county (which changes by year); At the well level: the altitude of the well, the distance of the well from the river (both of which don't change); and then I have dummy variables for the drought or wet conditions of that particular year (drought1, drought2, wet1, wet2, with wet3 excluded).
Code:
mixed lnper_cdtw100 lnpopden_county lnalt_avg lnriv_km2 drought1 drought2 wet1 wet2 ||countyid: lnpopden_county ||id2: lnalt_avg lnriv_km2 , robust
Performing EM optimization:
Performing gradient-based optimization:
Iteration 0: log pseudolikelihood = 21291.913
Iteration 1: log pseudolikelihood = 21350.855
Iteration 2: log pseudolikelihood = 21356.944
Iteration 3: log pseudolikelihood = 21356.946
Computing standard errors:
Mixed-effects regression Number of obs = 32,169
No. of Observations per Group
Group Variable Groups Minimum Average Maximum
countyid 31 1 1,037.7 5,400
id2 2,300 1 14.0 70
Wald chi2(7) = 47.18
Log pseudolikelihood = 21356.946 Prob > chi2 = 0.0000
(Std. Err. adjusted for 31 clusters in countyid)
Robust
lnper_cdtw100 Coef. Std. Err. z P>z [95% Conf. Interval]
lnpopden_county .0014967 .0008484 1.76 0.078 -.0001661 .0031595
lnalt_avg -.0164247 .0072642 -2.26 0.024 -.0306622 -.0021872
lnriv_km2 .0014593 .0008141 1.79 0.073 -.0001362 .0030549
drought1 .0074959 .0020855 3.59 0.000 .0034083 .0115834
drought2 .0044572 .0070613 0.63 0.528 -.0093827 .0182971
wet0 -.2147777 .0598072 -3.59 0.000 -.3319976 -.0975578
wet1 .018222 .0059625 3.06 0.002 .0065358 .0299082
_cons 4.757492 .0614828 77.38 0.000 4.636988 4.877996
Robust
Random-effects Parameters Estimate Std. Err. [95% Conf. Interval]
countyid: Independent
var(lnpopd~y) 1.54e-06 .0000194 3.16e-17 75331.21
var(_cons) .0000168 .0000483 6.06e-08 .0046738
id2: Identity
var(_cons) 7.06e-16 9.36e-14 1.2e-128 4.21e+97
var(Residual) .0155087 .0027515 .0109536 .0219581