What are the elements which you find to be crucial for telephone tracking studies to be successful?

First, those involved in tracking studies have to be very concerned about wave-to-wave replication. This means that everything related to the tracking study should remain the same. The same sources should be used for sample selection. Second, it is important that the respondents' qualifications stay the same. This means that those involved in preparing for each wave must give careful thought as to the methodology used in the previous wave. Ideally, all variables should be held constant except those that are being tested.

There are some other factors which should be taken into consideration. We find we must be very careful regarding local time zone calling. We want to be consistent and use the same calling times into areas because there are differences in who is home at different times. We also recommend limited daytime calling on consumer tracking studies because there is a disproportionate number of retired people and mothers with young children at home which could result in a skewed sample.

The client may also want to consider including some "control" markets, regions or groups in the design of the tracking study. This is particularly useful in determining if an observed change is because of the variables being tested or because there has been an overall market shift.

What about recommended sample size?

This is probably the most frequently asked question and an important one. The client should consider first the number and nature of market segments (e.g., regions, working women, males 18-34 years old, households with young children) that will be scrutinized upon completion of each wave of the study. Then it should be decided what the minimum number of interviews for that segment is acceptable. In general, we recommend a minimum of 100 interviews per segment, optimally more if the budget allows.

Second, the size of the total sample per wave is dependent on the desired level of precision. The greater the sample size, the lower the sampling error. Most market research firms will assist the client in estimating the appropriate number of interviews for each wave.

What about scheduling tracking studies?

If it is an annual or semiannual tracking study, data collection should ideally occur during the same calendar time periods as previous waves. Timing on ad tracking programs, however, are generally scheduled to reflect the flight and/or weight of the campaign.

Even though everyone involved in planning a study did their best to anticipate every contingency, there are times when questions must be added to a later wave. How do you handle this problem?

We know that inserting new quesitions into the middle of the survey can affect the data. Therefore, whenever possible, we try to put additional questions at the end of the questionnaire. This can usually be done with the effect of maintaining the integrity of the benchmark data and still obtain valid information for the questions which have been added. It is also very important that the "wording" of each question be kept exactly the same from wave-to-wave.

Some industries have a history of heavy advertising expenditures. How do you recommend to your clients to reduce potential problems in case an unusual amount of advertising expenditure is made by a competitor during a wave period of the tracking study?

In some industries, it is not uncommon for companies to "spike" expenditures over a short period of time. When this is a possibility, we recommend that our clients spread the data collection of each wave out between 6-8 weeks. This tends to reduce the impact of these "spikes" in advertising expenditures and provide more accurate results on volatile measures such as unaided advertising awareness and product awareness. On smaller tracking studies we recommend each wave be started and finished within five to seven days - thus measuring a very narrow point in time.

There are also situations where "continuous tracking" is the recommended methodology. In this case, a small number of interviews are conducted each day with tabulation occurring on the quarter or every six months.

Are studies which require a large sample handled any differently than those which require more modest numbers?

Normally, on large studies, we suggest CRT interviewing because it may be more cost-efficient. The software for CRT interviewing is highly sophisticated and can offer some cost savings.

The most useful feature of CRT interviewing is the ability to perform a preliminary "tab" of the results part way into the study. On large tracking studies that take some time to complete, this feature is especially useful in providing the client "top-line" results based on the first four or five days of production.