View Question
Q: Simple Statistics Problem ( No Answer,   5 Comments )
 Question
 Subject: Simple Statistics Problem Category: Science Asked by: wethepeople-ga List Price: \$10.00 Posted: 20 May 2006 22:12 PDT Expires: 19 Jun 2006 22:12 PDT Question ID: 730870
 ```Send me an Excel worksheet that shows how to determine the statistical significance of the following scenario: In Week 1, 100 customers enter Carmen's Candy Store and 5 of them purchase a Luscious Pops that week. In Week 2, Carmen moves the luscious pops display from the back wall onto the counter. This week 105 customers enter the store and 15 customers buy a Luscious Pop. Did moving the candy display have a statistically significant effect on the sales? To answer the question I'm looking for an Excel worksheet I can use to run these kinds of questions again using different input values.```
 There is no answer at this time.

 Subject: Re: Simple Statistics Problem From: orngu10-ga on 21 May 2006 12:00 PDT
 ```I'm not a researcher, but here goes: week # cust # purch # non-purchasers 1 100 5 95 2 105 15 90 pval: 6.74877E-06 Note that the only formulae are in the last column (column D): D2=B2-C2, D3=B3-C2, D5=CHITEST(C3:D3,C2:D2). This will provide a general solution to your problem (just enter the number of customers and purchasers for each week in columns B and C), but beware: if any of your cell entries fall below 5, CHITEST() may no longer be an appropriate test statistic.```
 Subject: Re: Simple Statistics Problem From: epidavros-ga on 28 May 2006 15:16 PDT
 ```You cannot determine statistical significance from two measurements. By definition two measurements are lineked by a straight line.```
 Subject: Re: Simple Statistics Problem From: youreh-ga on 09 Jun 2006 09:20 PDT
 ```If you consider the customers to be independent of eachother and it is believed that the only thing influencing the number of pops sold was the placement of the pops, then each week will be like a binomial distribution (and hence the variances can be estimated independently). More importantly, useful data then can be extracted. (If there was just one person in each week, then it would be a problem, but even then there are ways to work with the data... just not very powerful ones.) wethepeople, are you always looking to have just two types of states? (For example, purchasers and non-purchasers.)```
 Subject: Re: Simple Statistics Problem From: activealexaoki-ga on 10 Jun 2006 22:53 PDT
 ```I will be brutally honest since the question sounds rather request for service and does not "Answer". Hire someone, unless you are so unfortunate. Moreover, Google Answers do not support attachments and writing formulaes here cannot guarrantee that you can find the function to begin with (otherwise you would have done yourself!) So I will just write down, with similar table as previous commentator. And you request only the formluas, I won't explain. -Table- week cust purch 1 100 5% 2 105 14% mean AVERAGE(cust:cust) AVERAGE(purch:purch) sd STDEV(cust:cust) STDEV(purch:purch) p ([2purch]-[meanpurch])/[sdpurch]```
 Subject: Re: Simple Statistics Problem From: activealexaoki-ga on 10 Jun 2006 22:58 PDT
 ```I will be brutally honest since the question sounds rather request for service and does not "Answer". Hire someone, unless you are so unfortunate. Moreover, Google Answers do not support attachments and writing formulaes here cannot guarrantee that you can find the function to begin with (otherwise you would have done yourself!) So I will just write down, with similar table as previous commentator. And you request only the formluas, I won't explain. -Table- week cust purch ------------------------------------------------------------------------ 1 100 5% 2 105 14% mean =AVERAGE(purch:purch) sd =STDEV(purch:purch) pval (this is correction) =1-NORMSDIST(([2purch]-[meanpurch])/[sdpurch])```