Monday, June 27, 2016

CCHS question re variable INS_1

Question

Variable INS_1 is Insurance – prescription medications (literal question: Do you have insurance that covers all or part of the cost of your prescription medications?)

The questions of the researcher refer to the 2013-2014 iteration (available in <odesi> but not in STC Nesstar) and the 2007-2008 iteration:

1. For 2007-2008 do the yes and no responses refer to one of the individual years or is it the combined two years? If refering to one of the years, is the other year available and/or is there a combined number?

2. For 2013-2014, there is (in <odesi>), there are both 2013-2014 and 2014 entries. She would like confirmation that the 2013-2014 yes/no responses are the combined numbers and wonders if there is a separate number for 2013. The 2013-2014 numbers are approximately double that for the 2014 only. Perhaps one could just subtract to get the 2013 numbers, but I don’t want to offer that advice of my own volition! J

Answer

1. The two year file is a combination of two years, with the sample weights representing the Canadian population at a specific point in time. For example, the 2007-2008 file sample weights summate to roughly 28 million people, the Canadian population aged 12+ at a single point in time. A variable indicating the year of interview is not generally available on PUMF datasets but is available on master data sets to researchers each year. Annual files produced in off-years (e.g. 2009, 2011, 2013) tend to not become PUMFs while annual files produced in even years (e.g. 2010, 2012, 2014) do. Biannual PUMF files are created every two years.

2. The 2013 CCHS does exist as a master data set and is available to researchers that complete and have an approved RDC request. It has not however been made into a public use microdata file like 2014. We frequently get requests through the inbox about making a de facto PUMF by subtracting an annual file’s records from the biannual file (e.g. 2013-2014 minus 2014 = 2013). As the PUMF files do not have specific record identifiers at the respondent-level, and because PUMF files suppress other indirect identifiers to meet confidentiality guidelines, doing this would be a complex undertaking. If 2013 data is required, it might be best to request RDC access to the annual master microdata for that year.