Re-coding the Health Insurance "Other Specify" Items in the 2011 SAQ Written by Carol Roan 5/14/2014 History: The SAQ version of health insurance questions was not well thought-out. Participants were answering items about "other" types of health insurance before we asked about those specific types in later questions. This meant that the same plan was sometimes entered twice; once in other specify fields and once in later fields. Jessica Schumacher and her staff checked for these redundant reports by looking at the other specify comments and the answers participants gave to later questions. Based on their observations they recoded the data to make for the most accurate measures. For many cases, this meant that the "screener" question was recoded from "yes, I have other insurance" to "no." In addition to the screener question the Schumacher group grouped the legitimate answers into meaningful codes for the variables. A document with specific details of their coding decisions is cor1026a.pdf. The Schumacher team changed the codes using the webkit. Once the changes were complete Carol Roan modified the step2 program file to create additional variables with the other specify codes. In addition the changes are also documented in an excel spreadsheet which is saved as cor1026b.xls Note that in the process of reviewing these codes we discovered that we were missing data due to a date entry issue at UWSC. More information about this is available in cor1025. *************************************************************** Here are my notes that map the raw variables to analysis variables *************************************************************** Question 1 - uncodeable. Question 2 - uncodeable, Company/Plan name Question 3 - no categories needed, all responses were recoded or a blank (R circled yes but did not write anything) --- hopefully the spouses responses will lead to the same result Question 4 - VA/Military/Tricare, Company/Plan Name, uncodeable, Medicare HMO Yes Uncodeable Yes Company or Plan name given Yes VA/Military/Tricare Yes Medicare HMO Goal jins09re; y_hins1_9 (other public) jins18re; y_hins1b_9 (spouse other govt) jins21re; y_hins2_3 (other private that DOES supplement medicare) jins24re; y_hins2b_3 (spouse other private that DOES supplement medicare) jins27re; y_hins3_3 (other private does not supplement medicare) jins30re; y_hins3b_3 (spouse other private does not supplement medicare) jins35re; y_hins4_5 (other type of private Rx drug insurance) jins40re; y_hins4b_5 (spouse other type of private Rx drug insurance) Leave all of the above the same as they are now with these response categories . -29 -4 -3 -1 1 yes 2 no Create 8 more variables with valid data for when the above values are = to 1 jins09os jins18os jins21os jins24os jins27os jins30os jins35os jins40os . -29 -4 -3 -1 1 left blank 2 Not Codeable 3 Compnay or Plan name given 4 VA/Military/Tricare 5 Medicare HMO In order to do this we need to recode the original variables. Since the original variables already have original codes I'm going to get tricky and make a code set that adds 10 to the final values 11 left blank 12 Not Codeable 13 Company or Plan name given 14 VA/Military/Tricare 15 Medicare HMO 16 State or Federal plan named McKaylah asked whether or not she would have to code the 11s or if that already exists. I believe that if an other specify is left blank then UWSC codes it as "b" in cases which gets translated to -3. For the raw variables above a -3 code simply means left blank. But how do I get to left blank on the please specify line?