VCU Mass Communications researcher develops new metric for news release flow-through

By Malorie Burkett

RICHMOND, Va. (Sept. 8, 2004) – A Virginia Commonwealth University communications professor has developed an automated, statistically reliable evaluation tool for assessing the degree to which a news release influences a particular news article.

Ernest F. Martin, Ph.D., an associate professor in VCU’s School of Mass Communications, presented the results of his study adapting so-called “latent semantic analysis” to public relations during the 87th Annual Convention of the Association for Education in Journalism and Mass Communications in Toronto. The group’s public relations division cited the work as “Top Faculty Paper.”

Latent semantic analysis, or LSA, statistically infers relations for expected contextual usage of words in passages of text. According to Martin, LSA is widely used for information retrieval and text summarization, and it appears that it also can provide a method for measuring if key messages are flowing through to news stories. 

“We need to be developing useful and efficient metrics to add to our current public relation’s evaluation dashboards,” said Martin. “We work hard to build relationships with the media and to construct the perfect news release to tell our story.

“Obviously, public relations practitioners want the client’s key message picked up by the media as closely as possible to the one presented in the news release,” he said, adding that he recognizes that the final determination of what is used, when it is used and in what form is up to the gatekeeper.

“However, any public relations plan must provide a method by which the practitioner can measure results, and applying this type of analysis to news stories gives the practitioner a way of knowing if the key message is staying intact,” he said.

Martin said four tests were performed using sets of news releases and newspaper articles. The first exploration found that LSA can provide a reliable measurement of relatedness for the source news release to the target news article, even with two texts written very differently.

The second exploration explored lower-end LSA scoring by testing known, unrelated news releases, with news stories, establishing levels for key message non-relatedness.

The third and fourth explorations demonstrated ability of the metric to discriminate scoring levels for strong, medium and weak coherence between news release and news story, pointing to future research for refinements.

The School of Mass Communications is one of the largest programs in VCU’s College of Humanities and Sciences. At the undergraduate level, the school offers three sequences of specialized study in advertising, journalism and public relations. At the graduate level, the school offers a master’s degree in communications with programs in scholastic journalism and advertising.

EDITOR’S NOTE: A copy of Dr. Martin’s study is available in PDF format via e-mail or by fax. For information, please call University News Services at (804) 828-1231.