Origins: Market Research
At the turn of the twentieth century, a burgeoning field of market research made a science out of selling.38 Its early practitioners employed primitive survey methods to collect data on consumer preferences and track the effects of advertising on purchasing decisions. To the advocates of this new approach, a sophisticated understanding of the human mind was to be combined with the practical concerns of the business manager. As the psychologist Walter Dill Scott, an early proponent of market research, explained, “the principles and the methods” of modern psychology were to be “practically applied by the advertiser.” Its aim, according to Scott, was to aid “that increasing number of American business men who successfully apply science where their predecessors were confined to custom.”39
Work by Scott and others was part of a broader effort to study the distribution, sale, and use of consumer goods, a concern that eventually found an academic home in the nation’s growing number of professional business schools. At the Harvard Graduate School of Business Administration, established in 1908, marketing was a required course for all students. In 1911, the school’s first dean, Edwin Gay, established the Bureau of Business Research in order to develop the “scientific study of marketing.”40 Harvard soon became an influential center for market research, attracting leading figures in the field such as Paul Cherington, a former trade journalist who joined the original business school faculty in 1908 and became Harvard’s first professor of marketing in 1917. Cherington promoted the idea that careful study could reveal how best to reach consumers in particular markets. The methods he devised would greatly influence the development of survey research.41
At the same time business schools promoted the academic study of marketing, the nation’s leading advertising firms established research departments of their own devoted to the study of consumer psychology.
The J. Walter Thompson agency, then the largest advertising firm in the country, was at the forefront of so-called scientific advertising, or what company president Stanley Resor described as the search for laws of human behavior that could “guide the work of influencing the public mind.”42 In 1915, the company established a research department of its own, and in 1920 Resor hired John B. Watson, regarded as the founder of behavioral psychology, to be a member of its research staff. Watson’s role at J. Walter Thompson was more symbolic than substantive; he spent much of his time lecturing on the virtues of scientific advertising to business executives, trade associations, and industry leaders in the United States and abroad. The real research work at Thompson fell to Paul Cherington, who became the company’s director of research in 1922.43
Continuing the work he began at Harvard, Cherington perfected ways to study consumer behavior. He described the goal of his research as an attempt “to determine not only the number of actual or potential customers, but to get as good an idea as is possible of who they are, what their economic status is, what their buying habits and practices are, and what controls their purchases of the goods to be sold.”44 Such information was essential to the effective use of advertising, and it required collecting data on consumers themselves as well as their buying habits. To do so, Cherington perfected a way of drawing a representative sample of consumers for study. Specifically, Cherington segmented households into four classes, or strata, based on income: “Class A” households included what Cherington described as “homes of substantial wealth above average in culture,” “Class B” consisted of middle-class homes “directed by intelligent women,” “Class C” included “industrial homes” of skilled workers, and “Class D” homes consisted of unskilled laborers and immigrant stock, “where it is difficult for American ways to penetrate.”45 Under Cherington’s direction, teams conducted household surveys with a representative number, or quota, of respondents in each class. The technique came to be known as the quota sampling method.46
The techniques developed by Cherington and others quickly became the standard for market surveys. By producing a representative sample of manageable size, quota sampling made it possible to survey the public at a reasonable cost with results that were far more reliable than simple
“man-on-the-street” interviews. More broadly, quota sampling offered a way to aggregate, as well as disaggregate, the public. By selecting a set of categories or strata that the researcher deemed representative of the wider population, polls segmented the public into various categories of income or place. Looking back at Cherington’s class distinctions, for example, we can see both the methodological pitfalls and the inherent biases of such an approach.47 At the time, however, quota sampling enabled advertising and marketing experts to assemble survey results into a portrait of the mass market, or to disassemble and cross-tabulate them to reveal a particular market niche.48
One of the earliest and most profitable applications of the new tools of survey research was in radio. Although advertisers and broadcasters recognized radio’s ability to reach a wide and diverse audience, tapping into these commercial opportunities required some understanding of its consumers. However, radio lacked the handy metric of circulation that newspapers and magazines used to estimate readership, and broadcasters struggled to fix a price for selling products over the air. Without basic data on who was listening and when, it was difficult to assign a value to radio airtime.49
Survey methods offered a solution to this problem, and beginning in the late 1920s, both advertisers and broadcasters turned to those skilled in market research in order to measure the radio audience. In 1927, NBC commissioned Harvard Business School psychologist Daniel Starch to conduct a national survey of listening habits. Based on a canvass of 17,000 people in twenty-four states, Starch estimated that 47 million people listened to radio, and, using the rental value of homes as a proxy for income, Starch found that radio households were generally of a higher economic status.50 As a writer in Commerce and Finance observed, insights from the Starch survey extended well beyond the selling of soap: “One of the things political candidates and strategists doubtless would like to know just now is how the great invisible audience, made of many millions ... is reacting to campaign oratory.”51 Surveys revealed this “invisible audience” in ways that would make radio and other forms of mass communication a valuable political tool.
In addition to the general characteristics of radio listeners, the industry also wanted to know how many people tuned in to specific programs. In 1929, a consortium of advertisers commissioned
Archibald Crossley, a pioneer in the use of telephone surveys, to develop such a measure. Crossley established one of the first market research firms in the country, Crossley, Inc., in 1926.52 Located in Princeton, New Jersey, Crossley’s firm conducted regular telephone surveys of radio listening habits that asked respondents which programs they listened to during the previous day.53 The resulting “Crossley Rating,” as it became known, reported the percentage of radio homes that listened to a particular program, as well as a basic demographic portrait of the radio audience—something of great interest to advertisers.54
The great innovation of the Crossley Rating was that it offered the first ongoing audience measures for radio programs, something the popular press soon paid as much attention to as the industry for which it was designed. Time magazine called Crossley a “pioneer” in the measurement of “the unseen audience,” noting that top radio personalities such as Jack Benny “worried more about their ‘Crossleys’ than their hairlines.”55 However, the method Crossley used was flawed. By relying on telephone surveys, Crossley biased his sample toward upper-income households, something that became increasingly problematic as the radio penetrated American society. These shortcomings invited competition. In the 1930s, George Gallup, whose name would become synonymous with polling, developed audience measures that supplemented telephone surveys with personal interviews. Boasting greater accuracy (and accompanied by aggressive marketing), Gallup’s method replaced the Crossley Rating as the industry standard.56