Given the cost of single ads (more than $300,000 for some prime time network 30-second exposures) the decision of what level of repetition to use in advertising a product is critically important. Over the years many studies have sought to determine the appropriate level of repetition. In general, these studies conclude that low levels of repetition within a purchase cycle are all that are necessary to obtain the required amount of attitude and behavior change necessary for purchase. Low repetition consists of three or four exposures during the product’s purchase cycle, which for most household packaged products is about four weeks.
Since advertising is forgotten (wears out) over time, there is also the question of whether a pulsed or burst schedule should be used in contrast to one using constant spending. Generally speaking, the situation variables which argue for the use of low repetition schedules favor constant spending.
With few exceptions, media research is concerned with measuring the size and composition of individual vehicle audiences. In the case of print media, the audience is typically defined as being comprised of individuals who say they have seen one or more major editorial features. With television and radio, the audience can be defined in various ways, such as sets tuned to a program or number of people listening or watching. Since there are substantial differences in measuring print versus electronic media audiences, the discussion will deal with each one separately.
Print Media: The Audit Bureau of Circulation (ABC) reports the paid circulation of a medium, the number of home subscriptions, and the number of newsstand sales. For newspapers, circulation is broken down by daily and Sunday editions. For magazines, circulation data are shown by census regions and city-size groups.
But circulation does not provide information about the number of readers. The problem of so doing for a given publication is a most difficult one. How does one define a reader? The typical method, commonly referred to as recognition defines readers as individuals who are exposed to any of the editorial contents of a given issue. Respondents are classified on the basis of whether they claim to have read any of the editorial contents of the particular issue even though we know that this method yields unreliable data.
There is also the problem of ‘when’ to interview respondents in media studies, since exposure is cumulative. Thus to interview early may result in some respondents who will read the issue later will be excluded. To interview late means that some who were exposed will forget. This is particularly true certain magazines having a long readership life. The usual procedure is simply to interview continuously over a given time period and, therefore, to provide an average readership count. Another problem is what issue to study. Because of the cost involved, only a very few issues are studied in a given year despite the fact that readership of individual issues varies widely.
To confound the problems cited above, the act of reading certain publications (particularly magazines) has a status connotation which affects reported readership substantially. Order effects are also present; that is, when respondents are questioned first about monthly magazines, readership figures are much higher on the average than if they are queried first on weekly magazines and daily newspapers. Because of all the above problems coupled with the use of relatively small samples, it is not surprising that there can be substantial differences in the measurement of audiences by different syndicated services. Newsweek reports that one syndicated service estimated that it had more than 20 million readers, while another placed the total at 16.89 million.
Media analysts need information on the duplication among magazines. Since consumer readership of three or more magazines is quite common, the problem of duplication is an important one; rarely are data available showing the duplication among three or more magazines. An extensive study to determine duplication among all possible combination of magazines is obviously an expensive, time consuming activity. Agostini cites a study in France in which duplications were obtained for 30 magazines taken 2 by 2, as well as for all the possible combinations of 15 of the 30 magazines. This resulted in 32,767 possible combinations. He proposed (and validates) the use of a short cut method to estimate the unduplicated audience of combinations of three, four, five and so on magazines using total audience and duplication data of the magazines involved taken 2 by 2. The major drawback to Agostini’s method is the need for outside data that require periodic studies to be kept current.