Persuasive Advertising - J. Scott Armstrong [106]
A more recent study showed some improvement as substantiation was obtained from about 20 percent of the companies surveyed, while about 30 percent did not reply (Coney and Patti 1979). Similar findings were obtained by Anderson (1986).
Evidence on the effects of providing customers with verifiable evidence
Verifiable evidence is persuasive. A meta-analysis of studies in which some arguments provided sources and others did not, found that the inclusion of sources led to higher persuasion in 17 of the 23 comparisons, and it increased perceived credibility in 7 of 11 comparisons (O’Keefe 1998).
A lab experiment on print advertising found that subjects were more skeptical of claims that they could not check prior to purchase, such as “this carpet will last eight years” or of those that they could never check, such as “our soups are made with the finest ingredients” (Ford, Smith, and Swasy 1990).
Our WAPB analysis of quasi-experimental data found eight pairs of ads in which one ad provided a verifiable source for the evidence, and the other did not. Recall for verifiable ads was 1.23 times better than the other ads.
6.4. Data presentation
Consider an exchange on National Public Radio on June 13, 2000: “In 1999,” said Judith Applebaum of the Women’s Law Center, “3,800 cases were filed before the Equal Employment Opportunity Commission regarding wage discrimination.” In response, Diana Furchtgott-Roth from the American Enterprise Institute said, “But there are 60 million women in the workforce—that’s 0.006 percent, or to look at it another way, one in 15,000 women.”
Another example of data presentation: Assume that a sudden outbreak of a rare disease has threatened the lives of 20 of the 40 residents of a village in Vermont. Would you donate to a fund for a treatment that could save all these people? Alternatively, would you donate to a fund that would save the 20 people across the United States suffering from this same disease? Donations are expected to be higher in the former because the proportion of lives saved is higher (Fetherstonhaugh et al. 1997).
The principles in this section are most relevant for strong arguments that are presented in long print ads or on the Internet.
6.4.1. To aid understanding, use absolute numbers for small values and frequency rates for large values
Percentages are difficult for some people to understand. A survey of 1,000 Germans asked what “40 percent” means. Some respondents said “one-quarter,” some “4 out of 10,” and others “every 40th person.” About one-third of the respondents provided an incorrect answer (Gigerenzer 2002).
To communicate clearly, avoid relative rates. I was planning to run my first marathon when my wife read about a study claiming that this would double my chances of dying, as contrasted to sitting at home that day. However, all I needed to know was that my chances of dying while running a marathon was only 1 in 50,000, and I could tolerate those odds. But my wife is persuasive.
Relative rates are often used in advertising. Consider this claim: Hip fractures because of osteoporosis are expected to rise from 0.03 percent to 0.07 percent of the U.S. population over the next 50 years. Does that seem serious to you? To gain attention for the problem in 1999, the World Health Organization claimed that “The number of hip fractures worldwide due to osteoporosis is expected to rise threefold by the middle of the next century.”
Another example: An ad was headlined, “In patients with type 2 diabetes, Lipitor reduces risk of stroke by 48 percent.*” To their credit, the asterisk in this ad explained that the rate of strokes was 2.8 percent for a placebo versus 1.5 percent for Lipitor (Wall Street Journal, April 2, 2007).
Evidence on the effects of using percentages and frequencies
One paper reported on four experiments on the ability of people to understand statistical