Hello, I am doing multilevel modeling on a complex survey data with following variables bmi age sex education setting psu strata pweight
bmi is a continuous outcome variable; age is a continuous explanatory variable at level 1; sex is a categorical explanatory variable at level 1; education is a ordered explanatory variable at level 1; setting (urban or rural) is a categorical explanatory variable at level 2 (PSU level); psu is clustering in survey design ( I am also using psu for grouping that means psu is level 2); strata is stratification; pweight is weights
TITLE: Multilevel modeling with sampling characteristics of complex survey data. DATA: FILE = met.dat; VARIABLE: NAMES = bmi age sex education setting psu strata pweight; CLUSTER = psu; STRATIFICATION= strata; WEIGHT=pweights; WITHIN= age sex education; BETWEEN= setting; ANALYSIS: TYPE = COMPLEX TWOLEVEL; MODEL: %WITHIN% bmi ON age sex education; %BETWEEN% bmi ON setting;
Are the input commends correct? will this model correctly take account of sampling characteristics and multilevel modeling with psu as level 2. Finally outcome I am looking for is estimates, SE and p value for fixed and random effect. I am also looking for variance by psu and residual variance. how can i get these outcomes.
Thanks again Linda for prompt reply on my last post,here is one more question
I am doing MLM with following variables bmi age sex education setting psu.
individual is Level 1 PSU is level 2 or grouping variable
bmi is a continuous outcome variable; age is a continuous explanatory variable at level 1; sex is a categorical explanatory variable at level 1; education is a ordered explanatory variable at level 1; setting (urban or rural) is a categorical explanatory variable at level 2 (PSU level);
TITLE: Multilevel modeling DATA: FILE = bmi.dat; VARIABLE: NAMES = bmi age sex education setting psu; CLUSTER = psu; WITHIN= age sex education; BETWEEN= setting; ANALYSIS: TYPE = TWOLEVEL; MODEL: %WITHIN% bmi ON age sex education; %BETWEEN% bmi ON setting;
Are the input commends correct to see the effect of level 1 and level 2 explanatory variable on outcome variable at level 1?
I don't believe that R can do latent variable decomposition of an individual-level variable. The only comparison to R would be the first part of the example where there is an observed variable on within and an observed variable on between.
Hi Linda I am trying to generate some simulation data fitted with multilevel complex survey model (COMPLEX TWOLEVEL). That means I am looking for data as it coming from complex survey (sampling weights and clustering). I couldn't find in Mplus how to incorporate weights in it. Please let me know how can I generate this data for my Monte Carlo study.
Therefore, I can only generate data for multilevel Monte Carlo study without weights? Any Suggestion which software can generate weighted Monte Carlo data. Can I bring that data to Mplus for COMPLEX TWOLEVEL analysis. I appreciate your help
I do not know of any software that generates weighted data.
Mohd Masood posted on Wednesday, November 07, 2012 - 6:58 pm
I am planning for a simulation analysis for multilevel complex data. As we have discussed that Mplus can not generate data with weights. Now I am planning to generate a population and doing a complex sampling from that.
Please guide me how can I generate a population fitting a multilevel Model.
Here are the parameters and variables for my population
Total population= 100,000 Number of PSU= 200 Mean PSU size= 500 Variance in outcome variable due to PSU= 20%
See Example 12.4 and the Monte Carlo counterpart examples to the examples in Chapter 9. The CSIZES and NCSIZES options are what are used to generate the clustered data.
Mohd Masood posted on Monday, December 03, 2012 - 8:52 pm
I am generating the data using example 12.4 but I am not sure how can I save this data. Please let me know how can I save this data in a excel or spss or stata file as i want to analyze this data in R.
Mohd Masood posted on Wednesday, December 12, 2012 - 12:42 am
I am working on the example 12.4. I do get the explanation for the following codes
MONTECARLO: NAMES ARE y1-y4 x w; NOBSERVATIONS = 1000; NREPS = 500; SEED = 58459; CUTPOINTS = x (1) w (0); MISSING = y1-y4; NCSIZES = 3; CSIZES = 40 (5) 50 (10) 20 (15); WITHIN = x; BETWEEN = w;
But don't get the explanations for the following codes. Please let me know where i can get the explanations for them.
MODEL POPULATION: %WITHIN% x@1; iw sw | y1@0y2@1y3@2y4@3; y1-y4*.5; iw ON x*1; sw ON x*.25; iw*1; sw*.2; %BETWEEN% w@1; ib sb | y1@0y2@1y3@2y4@3; y1-y4@0; ib ON w*.5; sb ON w*.25; [ib*1 sb*.5]; ib*.2; sb*.1; MODEL MISSING: [y1-y4@-1]; y1 ON x*.4; y2 ON x*.8; y3 ON x*1.6; y4 ON x*3.2;
I don't know what you mean by your "intercept" model. You have a growth model with an intercept growth factor and a slope growth factor. The ICC for the outcome variable changes over the time points. As always
ICC = (between variance)/(between var+ within var).
Mohd Masood posted on Sunday, January 06, 2013 - 9:03 pm