Case: Statistics at work-Loveland computers ( by author)-Hint: Read on sampling methods.

After less than a week on the job as an administrative assistant to Loveland Computers CEO, Lee Azko was feeling almost overwhelmed with the range of projects that seemed to demand attention. But, there was no use denying, it sure felt good to put into practice some of the techniques that had been taught in school. And the next day on the job brought a new set of challenges.

I guess those folks in production must like you, Walter Azko greeted Lee by the coffee machine. I hope you are all done with purchasing because production has a quality control problem it needs help with. Go and see Nancy again.

Lee went to the assembly line but was greeted by an unfamiliar face. Tyronza Wilson introduced himself. Nancy said youd be down. Im in charge of checking the computers we use when we assemble high end computers for customers. For most of the components, the suppliers are so reliable that we just assume they are going to work. In the very rare case there is a failure, we catch it at the end of the line, where we run the computers overnight on a test program to burn them in. That means, we do dont want to be surprised by a part that fails when its been on the job for only a few hours.

Recently, we have been having problem with the 120 megabyte hard drives. You know, everyone used to be happy with 20 or 30 megabytes of storage, but new programs with fancy graphics eat up a great deal of disk space and many of the customers are specifying the large drive for their computers. To move large amounts of data, access time becomes very important that is a measure of the average time that it takes to retrieve a standard amount of data from a hard drive. Because access time is important for our customers, I cant just assume that every hard disk is going to work within specifications. If we wait to test access time at the end of the line and fine we have a drive that is too slow, we have to completely rebuild the computer with a new drive and a drive controller. That is a lot of expensive rework that we should avoid.

But it did not be even more expensive to test every one of them at the beginning of the process-the only way I can measure the access time of each drive is to hook it up to a computer and run a diagnostic program. All told, that takes the best part of quarter of an hour. I dont have the staff or the machines to test every one-and it is rather pointless because the vast majority of them will pass inspection.

Theres more demand than supply for the high capacity hard drives right now, so weve been buying them all over the place. As a result, there seem to be good shipments and bad shipments. If the average access time of shipment is too long, we return them to the supplier and reject their invoice. That saves us paying for something we cannot use, but if I reject too many shipments, it leaves us short of disk drives to complete our orders.

Obviously we need some kind of sampling scheme here-we need to measure the access time on a sample of each shipment and then make our decision about the lot. But I am not sure how many we should test.

Well, I think you have a good handle on the situation, said Lee, taking out a notepad. Let me begin by asking you a few questions.

Required: (a) What types of sampling schemes will Lee consider? (4 Marks)

What factors will influence the choice of scheme? (4 Marks)

What questions should Lee have for Tyronza?