Calculating
Sample Sizes
One of the more fundamental questions all researchers must answer before launching a survey is how many people do you want to survey. How big must your sample size be?
To determine how many people to survey, first you need to decide how much margin of error you can tolerate in the results. To put this another way: how accurate do you want your data to be?
The more people you survey, the more accurate your results will be. If you interview everyone in your target population, then you can say with complete accuracy what that population thinks.
But surveying everyone usually is prohibitively expensive, as well as unnecessary.
You can interview a sample, a portion of the total population, and project those results with some degree of accuracy onto the total population. But you have to expect some variation, or error, in the answers you get from the sample and the answers you might have gotten if you had actually surveyed everybody.
The margin of error tells you the range in which the "true" answer is likely to be found. The confidence interval tells you the probability that the "true" answer will fall within the range of the margin of error.
For instance, if the margin of error is plus or minus five percent and the confidence level is 95 percent, you can say this about your data: you can be certain that 95 percent of the time you will get answers that are within five percent (up or down) that you would have gotten had you surveyed the entire population.
After you know how accurate you want your data to be and you've determined the size of the population you want to get information on, then you must determine whether the results of your survey will be in
proportions or
percentages, or in means or
averages. With that information, you can easily calculate the appropriate sample size.
Note: keep in mind that margin of error is inversely proportional to cost - the higher the margin of error, the less accurate the results, but the smaller the sample and therefore the cheaper the survey costs. Since budget always plays a role in market research project decisions, you have to go with what is reasonable given your circumstances.
Maximize Your Performance Measures
Here are some excerpts from a recent article written by Polaris Vice President Lucy Klausner for
Competitive Advantage, a publication of the
American Society for
Quality, on ways to get the most out of your company's performance measurements:
Gain staff commitment. Start by ensuring that management and staff understand the purpose of measurement program. Try to and educate and engage them and help them see the big picture. Schedule meetings that will actively solicit their input and expertise about the customer experience and use that information in your measurements. Finally, try letting measurement be the team's idea and ask them to
help develop which measurements will have the most value to your organization.
Choose the right measurement tool. Each methodology (phone, mail, IVR, Internet, focus groups, in-depth interviews, etc.) has pros and cons associated with it, and as with all marketing research, you need to be aware of any potential biases. Key elements that will drive your methodology of choice should include, length of survey, audience, anonymity, confidentiality, timeliness quotas or forced representation, cost and accuracy.
Choosing what is right for you will depend upon your specific situation.
Carefully tie performance incentives to measurements. The most effective performance measurement programs almost always use measurement as carrot not a stick! They are sensitive to people's needs and clearly communicate what is going to be measures, how it will be measured and what can be expected based upon results. A successful program is one that is
communicated -- before and after the fact. Think about communicating results with your customers as well. To promote a team atmosphere, consider tying results directly to department performance rather then just to the individual.
|
