Matching biometrics with NHANES Good morning everyone,
I'm trying to figure out how to do some matching with NHANES datasets. Basically, I have a separate population of participants in a weight loss program, for which we do not have biometrics (we have age, gender, height, and weight for these participants).
I've constructed two datasets, one for the participants from the weight loss program with age, gender, height, and weight, and a data set from NHANES for data cycles 2005-2016, which also contains age, sex, height, and weight, but also has biometric lab data for Blood Pressure, HDL, Serum Glucose, Glycohemoglobin, Fasting Glucose, Triglycerides, LDL, and Apolipoprotein (B). For the NHANES dataset, I've followed the instructions for merging multiple years, and have selected the proper weights for the individuals. 
What I'd like to do is use the NHANES dataset to figure out biometric data for the weight loss program participants. 
Can someone help me with what the steps for that would be? I've been doing a ton of reading around propensity matching, and inverse probability weighting, but I'm not 100% sure which one I should be using? It looks like propensity matching might not be the way to go, because that's more trying to estimate the effects of treatment, whereas inverse probability matching is more for filling in missing data (which I think this problem technically counts as?). 
But again, I'm not entirely sure, and I haven't been able to find explicit instructions for this particular problem, just general ideas. If someone has done something similar, or has input it would be awesome! 
Or perhaps a logistic regression could be used estimate the missing data? Those sound like they have some promise with this problem. If someone just has a clear cut "use this method" for this task, I can do the research and work myself, I was just hoping someone might be able to put me on the right path forwards :) . Again, any help or input would be greatly appreciated. Thanks in advance, have a great day!
 A: I decided to go with multiple linear regression, and some matrix math to calculate beta coefficients - use Age, Weight, and Height, as my independent variables, and say Waist Circumference as my dependent varaible. Set it up as matrices, with another column of all 1's for the Age, Weight Height, and use this equation (which basically calculates vectors of y from the hyperplane of X in however many dimensions you have as independent variables + 1):
Inverse(Transpose(X) %*% X) %*% transpose(X) %*% y

Which gives you 4 beta hat coeffecients:
B0 - a constant
B1 Age coefficient
B2 Weight coefficient
B3 Height coefficient which are then put into this equation to calculate new y (i.e. waist circumference) values:
y=B0+B1*(Age)+B2*(Weight)+B3*(Height)

Then throw some standard error regression and use an iterative process to calculate the remaining biometrics coefficients, which can then be used to calculate new values for participants in the program, who only have Age, Weight, and Height (I separated Males and Females into different datasets to start with, then did the above, and Males and Females were separated as well when calculating new y values for the weight loss program participants. Think that's probably the best way to go about it, allows you to calculate the biometrics for participants who don't have exact matches in the NHANES population.
