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

On 12 Jan 1996, Pablo Sanchez wrote: > > In article <DL33Dv.EoI@world.std.com>, infosrfp@world.std.com writes: > > > > Ten years ago, while working with medical and scientific data > > for which partial results were computerized pending final > > reports from a laboratory, I was accustomed to flagging such > > "missing, but expected" values with a project-specific code; this distinguished > > "missing" cases from values that were already firmly known to be > > zero, null, void, "not applicable", lost, unobtainable, or whatever. > > > > Are you potentially looking at NULL? That's considered a non-value > so to speak... > He wants a multi-state NULL. One state indicates that we don't know the value but it will be available soon (whatever soon means). The other state shows that we never ever expect to know the value. Reminds me of the medical abstracts application written back in the late 70s. The Director of Medical Records wanted four potential values for 'SEX' (Male, Female, unknown, other). I questioned the need for both unknown and other. After the explantion she got all four values! --Steve +----------------------------------------------------+ | Steve Butler Voice: 206-464-2998 | | The Seattle Times Fax: 206-382-8898 | | PO Box 70 Internet: sbut-is@seatimes.com | | Seattle, WA 98111 Packet: KG7JE@N6EQZ.WA | +----------------------------------------------------+ All standard and non-standard disclaimers apply. All other sources are annonymous.