Re: Coding "missing values" - standards, customs, tricks of the trade?

TOWLER <TOWLER@SLB.COM> wrote: >You may have got a reply on this but here goes. The General Social Survey >(conducted by Univerrsity of Chicago) uses the following codes for missing > data. > > Refused answer Didn't know No answer > answer recorded > >1 column 7 8 9 >2 columns 97 98 99 >3 columns 997 998 999 > > >Hope this is useful. I use these codes all the time, along with other >fellow psychologists! > Additionally, when programming BSA or BES studies in Britain, I use negative codes to indicate route skipping - eg if asking questions of employees use -1 for "skp,not employee" use -2 for "skp,not fulltime" use -3 for "skp,not want ptime" cascading downwards as needed, as they can fit into ranges in the missing values definitions. I also have begun using these negative values to denote that the question was not asked in that version of the questionnaire, or that the self-completion part of the questionnaire was not returned. Leading to cascades such as use -1 for "No self-completn" use -2 for "skp, A version" use -3 for "skp,not in paid work" use -4 for "skp,no disabld coworker" This enables me to have more than the three values in missing values by using ranges. Ann by learning and courtesy