Taylor Series Linearization to Calculate Variance in NHANES

MikeDereviankin

Hello, 

Looking at the NHANES data that incorporates demographic weightings and population stratum & PSU, is there a way to do a survey design in JMP? There is SAS code attached from the CDC below as well as R & Stata code. 

https://wwwn.cdc.gov/nchs/nhanes/tutorials/Module4.aspx

Code: 
SAS

PROC SURVEYMEANS data=one varmethod=taylor nomcar;
  STRATA sdmvstra;
  CLUSTER sdmvpsu;
  WEIGHT WTMEC4YR;
  DOMAIN Select;
   * more statements...;
run;

Ideally, I can get this done so that I can run the tabulate function, which I am very found of in JMP. 
 

ih

I am not sure whether you can do this in just JMP/JSL, but JMP makes it so convenient to run R code I wonder if it is worth reinventing the wheel here. You could let R do the analysis and then bring results back into JMP to use tabulate/graph builder.

 

I've not tested this, but after installing R your code would look something like this:

 

Names default to here(1);

//Open R
R Init();

//Open your data table here:
dt = Current data table();

//send data to R
R Send(dt);

//run your code
R Submit( "\[
	# install package if needed, really only needs to happen once
	if (!("survey" %in% installed.packages())) install.packages("survey")
	
	#load package
	library(survey)

	#code from the website, you will want to verify column names here
	NHANES_all <- svydesign(data=dt, id=~SDMVPSU, strata=~SDMVSTRA, weights=~WTMEC4YR, nest=TRUE)                    
	NHANES <- subset(NHANES_all, inAnalysis==1)
	m <- svymean(~Depression, NHANES)
	
]\" );

//return data from R
svymean = R Get( m )

//Close R
R Term();

More info on integrating R with JMP:

https://www.jmp.com/en_au/events/ondemand/mastering-jmp/jmp-and-r-integration.html

https://community.jmp.com/t5/JMPer-Cable/Getting-started-with-the-JMP-to-R-Interface/ba-p/43920

 

And here is some info on the R package they use:

https://cran.r-project.org/web/packages/survey/survey.pdf