HI, I have a general question about the difference in 2 commands. I have multilevel survey data and an ordered 3 level dependent variable (outcome). I'm wondering what the difference is in the commands between a)survey setting the data and running meologit as a svy command vs. b)running meologit and adding in sampling weights. Does anyone have thoughts on that? (Note: I didn't use dataex because I'm not worried about replicating this or how it is running, just whether there are differences between these 2 commands).
Many thanks to any who might reply!
--Ann
Examples below, outcome=ckd_3cat and independent var=stunting
example a:
svydescribe
Survey: Describing stage 1 sampling units
pweight: w_ind_norm
VCE: linearized
Single unit: missing
Strata 1: site
SU 1: hogar
FPC 1: <zero>
. svy: meologit ckdu_3cat stunting
(running meologit on estimation sample)
Survey: Ordered logistic regression
Number of strata = 2 Number of obs = 773
Number of PSUs = 314 Population size = 814.402204
Design df = 312
F( 1, 312) = 2.55
Prob > F = 0.1115
------------------------------------------------------------------------------
| Linearized
ckdu_3cat | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
stunting | .4876653 .3055801 1.60 0.112 -.113593 1.088924
-------------+----------------------------------------------------------------
/cut1 | 2.569918 .207242 2.162149 2.977686
/cut2 | 6.545487 .6893908 5.189044 7.90193
------------------------------------------------------------------------------
example B:
meologit ckdu_3cat stunting [pweight = w_ind_norm] || hogar:
Fitting fixed-effects model:
Iteration 0: log likelihood = -241.89419
Iteration 1: log likelihood = -237.86994
Iteration 2: log likelihood = -237.30416
Iteration 3: log likelihood = -237.3033
Iteration 4: log likelihood = -237.3033
Refining starting values:
Grid node 0: log likelihood = -232.26489
Fitting full model:
Iteration 0: log pseudolikelihood = -232.26489
Iteration 1: log pseudolikelihood = -227.83392
Iteration 2: log pseudolikelihood = -227.57108
Iteration 3: log pseudolikelihood = -227.56916
Iteration 4: log pseudolikelihood = -227.56916
Mixed-effects ologit regression Number of obs = 773
Group variable: hogar Number of groups = 313
Obs per group:
min = 1
avg = 2.5
max = 10
Integration method: mvaghermite Integration pts. = 7
Wald chi2(1) = 2.57
Log pseudolikelihood = -227.56916 Prob > chi2 = 0.1091
(Std. Err. adjusted for 313 clusters in hogar)
------------------------------------------------------------------------------
| Robust
ckdu_3cat | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
stunting | .5356627 .3342881 1.60 0.109 -.1195299 1.190855
-------------+----------------------------------------------------------------
/cut1 | 3.25844 .3152659 2.64053 3.87635
/cut2 | 7.401028 .7907553 5.851176 8.95088
-------------+----------------------------------------------------------------
hogar |
var(_cons)| 1.625323 .5789674 .808588 3.267021
------------------------------------------------------------------------------
Related Posts with difference between meologit with sampling weights and svy:meologit?
matrix operation not foundDear All, I need some advice on how to deal with an error: matrix operation not found I get th…
Help with merging variablesGood afternoon, I am new here and would greatly appreciate some help. I have two variables about we…
Poisson and Negative Binomial Distribution and Stata CodesDear Stata Forum, My question relates more with the theory of Poisson and Negative Binomial Poisson…
Count the number of unique observation conditional on variableHi Statlist users, I am currently having an issue counting the unique number of observation conditi…
PSM method againHello every one i m doing my analysis for my research and i need to do endogeneity test by using PSM…
Subscribe to:
Post Comments (Atom)
0 Response to difference between meologit with sampling weights and svy:meologit?
Post a Comment