Media publicity can offer firms a seemingly cheap and impactful way to communicate about their goods and services. It has a broad reach, is regarded by consumers with less skepticism than advertisements and, in a sense, is "free" (although U.S. industry spent $19.4 billion on public relations staff in 2008).

But to what extent does such publicity affect consumer behavior? Results have been scant on the ground but Khim-Yong Goh, Kai-Lung Hui and I.P.L. Png offer new evidence to show how and when publicity might generate a sales response, using the unique case of the U.S. federal Do Not Call (DNC) Registry.

The DNC registry was launched in June 2003 as a free service for people to opt out of telemarketing. More importantly, it was promoted largely through media reports, not advertising, offering a natural setting for studying the impacts of news reports on consumer choice.

The authors collected 2,004 reports from about one month prior to the launch of the registry to one year after, in 136 newspapers circulating in more than 2,300 American counties. The reports were matched to DNC registrations and the interaction with other factors were also investigated, including newspaper political affiliation, editorial staff size, newspaper consumption and page counts.

The results offered clear support for the idea that publicity can indeed influence consumer behavior.

For every one per cent increase in newspaper reports, DNC registration increased by 0.018 percent. The impact was greatest in national newspapers (0.043 per cent) rather than local ones (0.011 per cent), after controlling for circulation. This has important implications for managers.

"Media generally base advertising rates on circulation so managers expect to pay more for more eyeballs," the authors say. "Our results suggest that, for the same rate per eyeball, managers should prefer national to local papers."

The content of the message was also found to affect registrations. Reports that contained both the toll-free number and URL of the DNC Registry were associated with more registrations, as were those that mentioned the number of people who had already signed up to the registry. But this impact was reduced when the reports were too long.

"Informative content such as the means to access a product/service (for example, the toll-free number and URL) could be helpful in promoting the use of the product or service, but a long report, or even a long headline, could hurt," the authors say.

"There is also robust evidence that persuasive reporting - in this case, citing the number of other people who have signed up - mattered. These findings have immediate bearing on how managers prepare press statements and work with the media to generate publicity."

A final and most interesting result was that the politics of the media mattered, too, even though the DNC Registry is a relatively neutral service. Newspapers that were politically neutral as measured by whether they supported a particular candidate in the 2004 U.S. presidential election, had a bigger impact on registration (0.050 per cent) than those supporting the Democratic candidate (0.012 per cent) and the Republican candidate (no significant impact). Moreover, reports in Democratic papers had a negative impact on Republican voters and vice versa, indicating they adjusted for political slant.

The authors advise that "instead of merely considering the reach and demographics of readership, managers must also attend to the politics of media. To the extent that our findings generalize to other goods and services, managers would achieve greater impact by focusing their publicity efforts at neutral rather than politically affiliated media."

This finding is new and the authors recommend further research into the impact of non-political reports in politically-affiliated media.