True circulation

Editor's note: Chris Van Derveer is senior project director and Ron Bass is senior research director at ORC Macro, a New York research firm.

One of the major challenges facing magazine publishers is to determine exactly how many people actually read their magazines. Simply looking at the number of subscribers would seem to be the obvious method of determining total readership. But this is not the best method, and it also provides data which is wildly inaccurate. To rectify this situation, a market research technique for developing highly accurate total readership figures has been developed. This method is called pass-along research. When magazines clearly establish that they have high pass-along numbers they can bill greater fees for advertising space.

The origins of the process go back to an event encountered by The Audit Bureau of Circulations (ABC). Almost 15 years ago, one of the business publications audited by ABC decided to profile subscribers by conducting a market research survey. Pass-along questions were later added to this survey, which is now referred to as a "Pass-Along Study."

Since ABC had promised to audit and publish the results, the Bureau was asked to help design a study which would be able to stand up to the closest scrutiny in terms of statistical validity. ABC solicited advice from leading individuals and foundations in the marketing research field and, from their suggestions, the current "officially sanctioned" sampling procedures and techniques for collecting and projecting the data were designed.

Following is a step-by-step description of the optimum procedures which should be employed to produce a valid pass-along result.

1. The magazine first decides what its pass-along statement should include, i.e., what industries, job titles, and other categories it wants to report. The research supplier, in consultation with ABC, determines the minimum number of interviews necessary to deliver stable and representative data. Typically, the research employs one of two sizes of random samples, 1,000 or 1,667 subscribers.

2. The magazine delivers a copy of its last ABC- or BPA-audited circulation file to the research supplier. The research supplier then re-sorts the file into industry and job title categories previously chosen and sorts the subscribers in each category into zip code order. This ensures that any random selection of respondents will also be geographically representative. International addresses are included as well.

3. Using random starting points, the vendor selects every nth name for inclusion in the survey. If 1,000 names are selected, this is the primary sample. From this total random group of 1,000 names, the research supplier must complete interviews with 60 percent of them, or 600 respondents, to produce a statistically valid result.

4. A mail survey is employed as the first method of contact. The entire primary sample of 1,000 or 1,667 names is first sent a mail survey questionnaire along with a cover letter and a small token of appreciation ($1). A period of two to three weeks is required to close this wave of the mail survey portion of the study, as most returns will have been received in that time. Following this first wave of mailings, a second wave is sent out to those who have not responded to the first request. For those who have not completed the mail survey after the second mailing, the research supplier will then attempt to complete the interview via a telephone survey. The supplier will make as many attempts as are required to reach at least 60 percent of the primary sample; sometimes, eight, 10, or even more attempts are required. Unless interviews are completed with at least 60 percent of the primary sample the resulting data will not be judged representative and will not be accepted by ABC or BPA.

For the contact method to be accurate, the mail survey and the follow-up telephone surveys must also be conducted with all overseas readers.

5. All data are edited for accuracy, coded, and keypunched. The mail survey and telephone survey data are merged and then tabulated. Weighting procedures are applied to put the various subscriber categories into their known proper proportions and to "project" the sample to the full subscription base.

6. ABC or BPA verifies the result and issues an official statement showing the total verified pass-along audience plus the subscriber base.

The involvement of ABC or BPA is integral to the success of the project. The auditor reviews, monitors, and/or verifies every aspect of the study. The auditor verifies the research sample selection; that is, if the sampling design indicates that the 405th name on the subscriber file is to be part of the selected sample, the auditor verifies that the 405th name on the file has, in fact, been selected as a primary name contacted.

Interviews are monitored on a random basis throughout the interviewing period to ensure that they are conducted with the designated respondent and that the respondent's answers are accurately recorded.

In addition, ABC or BPA performs the following:

  • validates that the information on the questionnaires has been accurately coded and keypunched;
  • verifies the accuracy of the data in the computer tabulations by checking against counts generated by hand-tabulating actual questionnaires;
  • verifies the accuracy of a "projection" of the data to represent the entire subscriber base.

Questions about validity

Many questions have been raised about the validity of the pass-along vs. other types of readership studies. One question asked often is "Isn't a census better than a sample projection"? This is not necessarily true. A census attempts to speak to everyone. (In this case, every subscriber to a particular magazine.) In practice, many people are unreachable or refuse to answer at least some part of whatever survey form they are asked to complete, and then the so-called census is actually completed by far less than everyone. The data reported in a census are often distorted due to the fact that there are no controls over who is and who is not answering all questions. Sample surveys, however, can always be controlled to represent every type of subscriber segment. Therefore, the results are accurately projectable to the total subscriber list.

Another alternative to our stated pass-along method would be to develop pass-along counts by tabulating the results of a pass-along question printed on their qualification/re-qualification cards. There are four major problems with this method of measuring a pass-along audience.

1. The pass-along counts are not profiled by title, making the data less useful to media buyers.

2. It is reasonable to suspect that many of the pass-alongs are going to non-qualified recipients, since a qualified reader could order a personal copy at no cost. (Why would an interested reader wait for his turn on a routing list if he could have his own copy for free?)

3. The qualification method can exaggerate the pass-along total, since it allows the same reader to be counted twice, with no mechanism for eliminating duplication. Take the case of Mr. Jones, who indicated on his qualification card that he passes his copy along to a new person, Mr. Smith. After receiving his secondhand copy for a few months, Mr. Smith decides he'd like his own copy, so he sends in his own qualification card. Since data from qualification cards are kept for three years, the records now count Mr. Smith twice, once as a pass-along recipient and once as a subscriber.

4.Qualification card information (for the primary reader as well as for pass-along readers) is usually reported from cards collected up to three years ago. In that period, many individuals have changed titles or jobs, or their companies might have gone out of business. Therefore, statements based on this type of data are undoubtedly out of date.

Powerful tool

The issuance of an ABC- or BPA-audited number for total readership is a very powerful tool in the hands of the magazine publisher. If the pass-along data can verify that the total readership for the magazine significantly exceeds the subscriber base, then the magazine can use this data to request higher advertising fees from its advertisers. In slower economic times, strong pass-along data can be employed to retain advertisers who might be considering a reduction in their advertising budgets.