From: Alexis O'C. <ale...@ou...> - 2006-03-03 16:46:10
|
Jon Maber wrote: > A difficulty is that if you use the functionality frequently enough > either one or zero names will be dropped from the list compared to > previous time. All changes to the results can then be attributed to that > person. The 'less than 50%' restriction protects the anonymity of the > first 50% of users but not the last 50%. > > The answer is (I think) to allow access to the tool at any time but when > it's run the first thing to do is choose a date. The tool will list > users who hadn't yet completed the questionnaire at midnight on that > date. The date is chosen so that if the command was run at some unknown > point in the past either the same list will be returned as was returned > then or a list with at least N more names. The user will be told the > date when the list is output and will be told the number (but not names) > of people on the list who have now completed the questionnaire. > > Jon > Hmmm... prehaps my original post wasn't clear enough ;-). I was also saying a cut-off date was required. Up until this date is reached, the results can not be retrieved in any way, not via the summary tool, nor CSV file, etc. As respondents submit a response, their name drops off the non-respondent list. Once all the respondents have replied or the deadline is reached (what ever comes first) the resource owner can then get the results. As they have received the results all at once, they do not have a clear way or attributing an individual result with an individual user. Anonymity has not been leaked over time as it would be if the results *and* non-respondents list were both available prior to everyone having responded or the deadline being reached. (The deadline is required because there can be perfectly legitimate reasons for not everyone having responded). Alexis |