Dan Waters
dwaters at usinternet.com
Wed Oct 26 10:17:15 CDT 2005
Hi Steve, Generally, with a survey, you want the people involved to be taking exactly the same survey, unless you are doing initial testing to see if the order of the questions will bias the results. Once the survey is ready, the randomized order of the questions should remain unchanged. Hope this helps, Dan Waters -----Original Message----- From: accessd-bounces at databaseadvisors.com [mailto:accessd-bounces at databaseadvisors.com] On Behalf Of Steve Erbach Sent: Wednesday, October 26, 2005 9:48 AM To: Access Developers discussion and problem solving Subject: [AccessD] A2003: Random order Dear Group, I'm writing a prototype Access 2003 ADP in preparation for writing a "practice" VB/ASP/ADO .NET project, and I wonder if you lot could help me out with the logic. It's a survey of 100 questions. There are 10 groups of 10 questions each. Each group of questions deals with a particular topic. Rather than have the survey takers see the questions grouped by topic so that they might "steer" the results, I thought it would be better to randomize the list of questions. My back-end is SQL Server 2000, but I think the principles apply to Access as well. So far I've got a table called OrgQuestions with this structure: QID, int, key TopicID, int QNumber, int QText, varchar(175) QOrder, int The QNumber goes from 1-10 for each of the 10 TopicID's. The QID goes from 1-100, conveniently enough! The QOrder column is what I added to indicate the (random) order in which the survey questions will be presented to the survey taker. In the code I first run a MakeTable query to create a temporary, single-field table containing just the QID field, renamed to TestQID. 100 rows numbered from 1-100. I open two RecordSets, rstPik and rstOrd. rstPik is that temporary list of Question numbers. rstOrd is the OrgQuestions table containing the QOrder field. I also create a 100 element array to store the random order of the questions. Then I run two FOR loops: For i = 0 To 99 sngSeed = CSng(Time) ' Multiply the random number generated by 100 - i to get a whole number big enough for ' the list of numbers I need to fill. That 100 - i gets smaller and smaller as the available ' list of numbers gets smaller. sngRnd = Rnd(sngSeed) * (100 - i) ' I do the + 1 here to account for the truncation that the INT() function performs. That is, ' if the Rnd() function gives me .00125, then, even if I multiply it by 100, the INT() function ' will return 0. The HIGHEST number that the RND() function * 100 returns is less than 1; ' so I need that + 1 there, too, to ensure that 100 comes back as one of the "random" ' numbers. intRnd = Int(sngRnd) + 1 rstPik.MoveFirst ' Move to the row indicated by the random integer - 1 because the MOVE() function ' moves X rows not counting the current one. I have to account for the "fencepost" ' error. rstPik.Move intRnd - 1 intStore = rstPik("TestQID") intArr(i) = intStore ' Here's where I eliminate the question number that has been used. rstPik.Delete rstPik.Update Next i ' Now the rstPik Recordset is completely empty. rstOrd.MoveFirst For i = 0 To 99 rstOrd("QuestionOrder") = intArr(i) rstOrd.Update rstOrd.MoveNext Next i In a nutshell, fill an array with the randomized sequence of question numbers and then dump those numbers into the column of the OrgQuestions table that holds that sequence. The issue here is, at least with SQL Server, the generation of the random sequence takes a bit of time. My SQL Server is hosted by my web host and this routine takes an appreciable amount of time. Does anyone have a suggestion as to how I might speed this up? With a zippy cable connection, this process takes over 20 seconds. -- Regards, Steve Erbach Scientific Marketing Neenah, WI www.swerbach.com <http://www.swerbach.com> Security Page: www.swerbach.com/security <http://www.swerbach.com/security> -- AccessD mailing list AccessD at databaseadvisors.com http://databaseadvisors.com/mailman/listinfo/accessd Website: http://www.databaseadvisors.com