Tuesday, January 28, 2020

CCHS variable CCS_185

Question:

I have a researcher working with multiple years of the CCHS (2007-2008 to 2015-2016) and a question has arisen - please see his message below:

In the CCHS documentation item CCS_185 (last time to have colonoscopy or sigmoidoscopy); the response options include both a time range of "5 TO < 10YEARS" and another time range of "5 YEARS OR MORE". These options should not be mutually exclusive (although they are presented in the dataset as such). Do you have any explanation for this? 

The researcher has indicated that it is represented this way in the data itself as well and is present across all the years. My initial thoughts are that there is a typo and that the second option should perhaps be "10 years or more", as otherwise these two response options overlap each other.

Can you please advise?

Answer:

There are a couple previously identified format/label issues with CCS for multiple CCHS cycles, including the format for CCS_185. The error for CCS_185 is a label error and CCS_185=6 should be ‘10 years or more’ (errata item 19). I would encourage the researcher to consult the errata found in the documentation folders (two separate CCHS annual component Errata documents; one for the 2000 to 2014 reference period, and another for 2015 and later) , this document contains information for all known errors as well as information on correcting known errors.

This may have previously been mentioned to the researcher, but the CCHS annual component was redesigned for the 2015 reference period and onward cycles. 
As a result of the 2015 CCHS redesign, combining or comparing cycles of CCHS data from before and after the redesign (e.g., combining/comparing 2014 (and earlier files) with 2015 annual files) is not recommended, and caution should be taken when comparing estimates across those years. Even estimates derived from content that has remained unchanged, are subject to the potential impacts of the other major changes to the survey (i.e. new survey frames and collection methods) and may not necessarily be comparable with past cycles. It would be very difficult to ascertain whether any changes or consistencies between estimates pre and post redesign are a reflection of the true population characteristic being examined, or the effect of the significant methodological and operational changes made to the survey. Please review the 2015 CCHS Redesign Summary found with the 2015 and also with 2016 survey documentation.