of Regional Research
Paul Donato, Audits & Surveys Worldwide
Barry Koch, Turner Entertainment Networks USA
Paper presented at the ARF/ESOMAR Worldwide Electronic and Broadcast Audience Research Symposium, San Francisco (USA), 21st-24th April, 1996
Summary: Los Medios y Mercados de Latinoamérica
is an industry-developed and -supported pan-regional multimedia survey of Latin America.
The methodological and technical difficulties associated with this project are enormous.
Within that context, the authors review and evaluate how this industry-sponsored action
came to be and try to draw implications for developing media in other regions.
Much of the material reported herein represents the opinions, sometimes divergent, of several sponsors of this project. These opinions are collected through a series of debriefing meetings and telephone interviews.
The authors wish to express their thanks to JoAnne Burke of FCB/True North and Sandy Wax of The Discovery Channel for helping to organize this paper and contributing to the editorial content. The authors also wish to thank the following sponsors for their editorial review and contributions: Rebecca Bearden of Fox Latin America, Filiberto Fernandez of TeleNoticias, Ned Greenberg of Canal de Noticias NBC, Fran Kennish of Young & Rubicam, I.J. LaCayo of Bates Worldwide, Cristina Martinez of J.Walter Thompson and Marcelo Salup of Foote Cone & Belding.
This paper reports on an initiative first undertaken by the Latin American cable programmers in 1993. The purpose of this initiative was to develop an infrastructure for buying and selling pan-regional advertising in Latin America. The 17 regional programmers formed an industry association called TAP LA (Television Associations of Programmers of Latin America). One of the first tangible deliverables of the association was the development and execution of a research platform which covered the 19 countries of the region.
It was no accident that a research platform should be the first real product of the organization. Without regional research, the concept of regional advertising was amorphous. Usually, a research platforms are "supply side in their origination". That is to say, the research suppliers identify a niche or role in the marketplace and proceed to fill it. In this case, TAP LA played a very active role in the development of this research platform.
This paper reviews the organizational aspects of the study. The paper is presented with the objective of sharing the developmental experience of Latin America, so that emerging media in other parts of the world might benefit from this.
In total, some thirty companies sponsored this research. We need to reflect the attitudes of these sponsors if we are to portray our conclusions as being representative of an industry. Towards this end, we conducted a series of interviews with a number of the sponsors about the organizational and methodological issues. Much of this paper contains the edited comments of the sponsors. The paper is organized into a flow that will take us through (1) identifying and reacting to the need for pan-regional media research systems among a group of potential buyers (advertisers and advertising agencies) and sellers (media companies, both electronic and print); (2) shaping the buyers and sellers into a formal consortium; (3) identifying the appropriate research product(s); (4) designing the research methodology; and (5) channeling reactions to the research data into positive methodological improvements.
More often, research suppliers identify a niche or role in the marketplace, develop a research program and bring it to the market. This would be called "supply side" research. In the case of Latin America, the research was more like "demand side" research. It should be emphasized that no pan-regional media research exists previously. At the outset, the consensus was that the visible hand of an industry association afforded some significant benefits such as detailed market knowledge as well as the perceived market needs.
One programmer said, "At first, we did not really have any formalized structure .... In a way, I think that it actually helped because we would focus on the specific need, which was to collect marketplace data. From that need grew the infrastructure. You need the association before you launch the regional media system."
To interpret this requirement for an association, one has to understand the uniqueness of the Latin American market and therefore recognize that these circumstances may dictate the benefits from a joint industry association as they cannot in other regions of the world.
As one agency researcher put it, for "other regional media studies, there is no industry association ... A supplier just comes and invents it, and brings it along .... where we were, which is right in the middle of the pre-existing association and the supplier just coming ... is exact perfect .... for this newer ground."
In some ways, Latin America is different from other regions. In other ways, it is very similar to those regions. To those who have worked in Latin America and Europe, the research infrastructure in Latin America is clearly less well-defined. Few countries have total population surveys for media behavior, which are believed to be changing at dramatic rates. Economically and politically, the largest economies of Latin America have undergone significant political shifts in the past ten years. Economically, hyperinflation still characterized half of all Latin American countries up to three years ago, but the situation is somewhat better now. All of these factors contribute a sense of newness in the Latin American marketing environment ---- a newness which does not characterize people's thoughts about Western Europe as a region. Clearly, there are also differences between Latin America and other regions such as Asia-Pacific, Africa, Middle East and Eastern/Central Europe.
However, it is important to note that two-thirds of Latin America speak Spanish, notwithstanding some idiomatic differences, and the other one-third is Portuguese-speaking Brazil. The income and consumer structures of Latin American countries are more alike than dissimilar. As one network sponsor said, "It would be extremely difficult to try and lay something like this in Europe at this point because so much has already been done." Therefore, industry associations may serve the greatest benefit to developing regional media environments, especially those that are homogeneous in language and consumption.
SHAPING THE CONSORTIUM: RESEARCH AS THE ORGANIZING PRINCIPLE
Why should an industry initiative begin with a research project? Why not a promotional effort? Why not a lobbying effort? The consensus was that the research needs were so serious that even fundamental decisions could not be made at that time. In the absence of research data, decisions about planning, programming and distribution were being made by piecing together faxes and facts from local offices in nineteen countries.
At the same time, regional marketing had no real shape or defining boundaries. Without a quantitative description, it could be too easily debunked on the basis of presumed national differences.
As one programmer suggested, "It is odd that we started with a research project, but there needed to be an excuse or a reason to try and bring the different programming entities together." Another programmer suggested, "It is the one thing which everybody can hopefully put their differences aside for."
Additionally, using research as an organizing principle presented this association of television programmers with the opportunity to expand the scope of the project to include agencies and other media. If Latin America was to be positioned as a viable marketing entity, it would have to be developed as such for all regional media. Thus, the "cable" study became a multimedia study. If that transition had not occurred, the study would have been viewed as something like a subscriber study, rather than a platform for framing the concept of regional marketing.
Speaking of the early stages of the proposal, one programmer said: "We did not know how agencies fit in or how other entities fit in. We just had the idea that we wanted to put together an industry-wide request for proposal, for some type of study which could be shared among cable networks."
An agency research director replied: "As long as a cable network or even a consortium of cable networks go out and does its own research, I am not going to chip in for it. I am not going to buy into it because we agencies are just too suspicious of what people are trying to do. So I think that it was really smart to get the agencies in. I think that it was even smarter to get the print media in, even though it was a little bit later, to not only pay for the study but to get greater credibility."
IDENTIFYING THE PRODUCT: PRE-PLATFORM PLANNING OR DOING WHAT YOU GOTTA DO
Commonsense holds that after you introduce a platform where none existed before, there would be endless disagreements about the data. But at least people were all looking in the same direction: "Could the gray market for cable television be so large?" "What channel assignments do we have in Mexico?" "How many different issues of a magazine are sold at a newsstand at any one time?"
An agency director recalled the way that one network used to sell in the early days: "They came in with subscriber counts that they got from each of the countries, and estimates of ratings based on the US in 1984 when cable was new there."
ALL THINGS TO ALL PEOPLE
Given the vacuum, it should serve as no surprise that a multimedia planning database would dramatically focus attention on a variety of issues when the results suggested something other than conventional wisdom. The biggest danger is that it is natural to use the platform to "answer all questions - even those that it was not intended to answer."
Lacking any other information on a specific issue, agencies may and can use the results of the study to evaluate the composition of a network within a country. Such an analysis may result in decisions based upon very small samples. Media companies are in a similar situation in that the study be the only source of information on the demographics of viewers in certain areas.
It is easy to attack the validity of such analyses, and often the entire platform is attacked in the process. Some better methods need to be established to identify the limitations of such platforms. Setting minimum sample sizes for analyses may help. But when the alternative is having no data at all, then the decision-makers will have to make use of any information, which could be even worse.
In a strange way, determination of the methods to be used is a function of balancing the international community's beliefs about methods and harmonization against the experience of the local research company doing the fieldwork in each country.
Nowhere was this more clear in our decision to adopt the instrument format which is used by multimedia platforms in much of the rest of the world. During the initial stage, one of the sharpest differences among the leading research companies was in their contrasting opinions on the viability of using a self-administered product-usage questionnaire in addition to an in-person interview.
The product data represented one of the valuable components for the agency sponsors. But one research firm said that it could be done, while the other said that it was impossible. In the end, it was local expertise that confirmed the sponsors' gut feelings. A South American field coordinator pointed out that the product-usage questionnaire was very similar to a TV diary which was currently placed and retrieved personally in many countries in Latin America, with much more stringent requirements.
As a result, the sponsors decided to pursue this instrument as a means of collecting consumer data. In the end, the instrument seemed to work better in Latin America than in the USA. About 72% of the Latin American respondents returned usable product-usage questionnaires. The reasons for the success are simple: The questionnaires is short (16 pages) and the incentive of US$5 was significant for poorer households.
Consider the following supplier observation on the early data: "This stuff is so esoteric that, unless you have done it all your life, it is not clear how you could ever figure what the numbers mean. Look at the magazine audience estimates. Historically, when you look at the magazine pilots that were done 10 or 15 years ago, they effectively produce the same seemingly inexplicable numerical levels. Unfortunately, they folded the tent on the pilot rather than try to figure out what the data were saying. I think that they had an easy retreat because it was only a pilot. Maybe we should not be so quick to back away from different numbers ---- usually good studies should tell you something even if the results are quite unexpected."
THE STRUCTURE OF KNOWLEDGE
Perhaps the most critical of oversights which occurred during the planning of the survey was the integration of all interested parties: New York City, Chicago, Miami, Los Angeles, London, Mexico City and Buenos Aires.
The issue is simple: New York City and Chicago have a greater concentration of people who are familiar with the theory and practice of international media measurement. Miami, Mexico City and Buenos Aires (and other Latin American agency hubs) have a greater concentration of people who know about Latin America.
The origins of the research platform came from the media looking to institutionalize the practice of regional planning. The early agency contacts at the agency level were with media researchers since the issues were methodological in nature. This was a mistake that could easily have been avoided if we sought to draw on the knowledge of all those who could contribute. Understandably and correctly, the Latin American specialists at agencies felt that it was silly to conduct this research without consulting them first.
However, a funny thing or rather two funny things happened after the first wave of the study was reported and before the second wave was fielded.
The first thing is the fact that because there was something to feel left out about, agency and media communications and commitment to Latin America necessarily started to improve. As one programmer put it: "Well, I think information like the data that are coming from the study forces the companies to look at the communication systems within their companies. There was a real need to get this information to other groups. Then you recognize the pitfalls when you don't have the communication systems in place..."
Miami agencies began to be briefed on media issues and eventually they became the primary contacts, with New York and Chicago now being briefed. Several sponsors hired Latin American analysts to study the data. Online services that never considered Latin America sent sales representatives to Mexico City to sell the systems that they already provide in New York, Chicago, Atlanta, Washington DC and Los Angeles.
Intended of not, having something tangible that is being fought over pushed Latin American media planning towards the way it is done in Europe and North America. That is, media planning becomes slightly less instinctual and intuitive and slightly more quantitative and programmatic. This led to the second great thing.
The Latin American specialists in Miami and Mexico began to take ownership of the research and media responsibilities, now that there is something to take ownership of. It was clear that many people now believe in the viability of regional media. At this time, there is a lot of trade press about whether or not Miami will become the regional media hub of Latin America. We cannot answer that question, because only time will tell. But we believe that our experience has shown that research is not such a bad place for an industry association to start. If the industry can frame an issue --- namely, the liability of regional media --- clearly, then media and advertising companies will be faced with the rational decision to restructure their companies such that their Latin American specialists will be designated the Latin American media experts.
NEXT PHASE: RATINGS
The idea was to conduct survey research first, which would then usher in a ratings system. In the beginning, there were some differences in opinion because some people felt that ratings should perhaps have come first. In retrospect, building a regional ratings system prior to conducting a field study such as Los Medios might have been a disaster. The costs of placing people meter systems are such that if you could not get it right the first time, you may not be able to do a second time.
Ironically, according to one agency's point of view, "from a buying point of view, ratings were less needed for the agencies because there had been no product usage data. There was no information what cable homes consume. I think that we can live with the current ratings. Even if they were not the best thing, we can live with what we have. But in terms of all the other stuff (such as regional consumption of products), it was a brand new need for which there was absolutely nothing out there in the marketplace."
Quite apart from the value and assurances of first establishing the credibility of the regional media, there was a myriad of practical reasons for testing out a sampling or estimation method with survey research before investing heavily in electronic data collection.
"If you don't know what the universe is, how are you going to conduct ratings? You want to flesh out most of the methodological issues with survey research work. This is because the survey research work is a glorified and most extensive enumeration survey that you could do. Otherwise, you are totally naïve going into the project."
What kinds of issues can survey research flesh out to support the development of regional ratings system? Consider the following examples:
THE UNIVERSE: Most of data sources and studies project to persons by age and sex as estimated by the United Nations organizations. But there is no standard definition of households throughout Latin America. We found that 3.7% of households shared kitchen facilities with others and 4.9% shared housing with other families. For example, in Panama, we found a casa de inquilino with twenty families sharing four kitchens. So do we have one housing compound? four households? or twenty households? Such practical field issues needed to be addressed beforehand.
CLUSTERING: Since telephone penetration in Latin America is less than 35%, installing people meters presents special problems. For households with telephones, the television tuning/viewing data can be transmitted by phone to the central computer. For households without telephones, the data can be transmitted by installing a cellular phone dedicated to data collection and otherwise inaccessible to the household. However, this is clearly a very expensive proposition. As an alternative, one might consider transmitting the data from a household without telephone via RF frequency to another metered household nearby with a telephone, and hence transferring both households' data over the one phone line. Such a protocol might be feasible if we understand something about telephone distribution patterns, which is provided by the survey research work.
The survey research can also contribute to an understanding of issues that many metering efforts may fail to discover themselves. Consider a household with Direct-To-Home satellite dish. Its television behavior is relatively independent of its neighbors, although they have about the same socio-economic characteristics. But consider the case of satellite cable, where a local neighborhood receives a low-cost form of television from US-based satellites. Virtually every household in this neighborhood receives the same channel lineup of US networks. Knowing something about one household's reception capability tells you everything about the neighborhood.
When the study was being planned, very little attention was given to the fact that Latin Americans can watch US-feed as well as Latin American-feed programs. "Twenty different companies failed to think that asking the language of broadcast for the networks would have been a good idea. But because it was an industry failure, there was no incrimination at the backend."
Any distribution person working for a cable/satellite programmer with worldwide operations would have suggested that this is a problem. But, at the outset, the representatives simply did not know enough to realize that this would be an issue.
WEIGHTING: There is a seemingly trivial example of sample allocation that until you have gone through the exercise, even a sampling statistician cannot appreciate. To simply the problem, let us suppose that you have 1600 sample respondents to allocate in two countries and that you want to report each country separately. How do you make the allocation? To have separate reports, you would want about 800 respondents in each country. If the two countries are Argentina and Colombia, which are about the same in population size, then a 800-800 allocation is approximately proportionate to population size and all is well. But if the two countries are Brazil and Puerto Rico, then a 800-800 allocation is very much disproportionate to population size. But a disproportionate allocation is not necessarily a bad thing in terms of sampling variance, because the measured variables (such as cable/satellite viewing, magazine subscription, product consumption) will be negatively correlated with the case weights and in fact reduces the variance. This counterintuitive finding can be demonstrated empirically.
CHANNELING: One of the most important benefits of implementing an annual project such as this is that it is easier to talk in a clinically detached way about what you should do differently and then just do it. By the time that you discover the problem, you are ready to go into the field for the next study. The timing of this study is such that the market has a chance to study and react to the data. Other studies which go to field too quickly requires two years to correct any problems discovered in the field or otherwise adapt to a rapidly changing environment.
In the second year of our study, the following changes were introduced:
These corrections have improved the study in significant ways.
As one programmer put it, "I think that the bottomline is that there is a very steep learning curve. We are certainly not at the top of it yet. But going through this process ... and it was a pretty arduous process in terms of the number of committee meetings ... has helped us to learn from it. Hopefully, when we talk about Asia and when we talk about the Middle East or wherever we want to go, we do not have to start from ground zero again."
As another programmer put it, "I guess that in terms of having this as a blueprint for other regions, you are really talking about a lot of the same players, both agencies and networks. In some cases, you are talking about the same individuals who have the responsibilities."
So the process of industry learning can build on top of what we already gone through together.