It turns out to be very tedious form groups of subjects which are comparable on a set of variables, and not on other variables. It would be desirable and useful if there is any tool that can automatically create groups that do not differ statistically in these variables for quasi experimental studies.
Does anyone know some tool for this?
Thank you.
Your question sounds like it may be addressed using a propensity score matching or propensity weighting approach. There are some well-developed approaches in Stata and R (the packages that I'm familiar with) that not only match or create the weights, but also evaluate the balancing using statistical and graphical approaches.
Paul
Your question sounds like it may be addressed using a propensity score matching or propensity weighting approach. There are some well-developed approaches in Stata and R (the packages that I'm familiar with) that not only match or create the weights, but also evaluate the balancing using statistical and graphical approaches.
Paul
Dear Julio Alberto González Torre,
I concur with Paul Visintainer that this looks like a case for propensity score tools. Here is one handy source to find out more about propensity score matching:
Li M. 2013. Using the propensity score method to estimate causal effects: A review and practical guide. Organizational Research Methods 16(2): 188-226.
Absolutely, this sounds like a case for propensity-score matching at first glance. I am typically wary of recommending statistically nuanced techniques without the aid of a statistician, so please consult a local statistician if one is available. If a local statistician is not available, here is another good resource for the layperson that walks through the basics of PS matching and what it can & cannot do: Streiner DL and Normarn GR. The Pros and Cons of Propensity Scores. Chest 2012; 142(6): 1380-1382.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques