KNOWLEDGE NETWORKS REMAINS ONLY ONLINE RESEARCH FIRM TO DELIVER SIX ELEMENTS OF ACCURACY

Growing industry concern over data quality emphasizes importance of statistically valid methodology employed by KnowledgePanel®

Menlo Park, CA; October 4, 2006: At a time when attention is being focused strongly on the quality of online research – which is being used increasingly as the foundation for product and media strategies of large consumer companies – Knowledge Networks is already employing the only online methodology that provides six essential, proven elements of accuracy and validity. KN's nationwide online panel – recently renamed KnowledgePanel® – has been built and maintained according to those standards since 1998.

Two recent industry summits on respondent cooperation, a front-page Advertising Age article ("Consumers Sick of Marketers' Endless Surveys," October 3), a growing number of studies by blue-chip research buyers, and a Stanford University study comparison of major online research providers have all shown that there is cause for concern. Each has demonstrated the growing realization that opt-in online panels yield less accurate data and can lead to wrong business decisions.

In particular, attention has been focused on six touchstones by which online information and panels need to be evaluated:

• Coverage bias: If those without Internet access cannot participate, data will be skewed; all opt-in panels exclude the non-Internet population, but KnowledgePanel® does not.
• Self-selection bias: When respondents choose themselves (i.e., opt in, or volunteer) – rather than being chosen as part of a random sample (the KN approach) – then it becomes impossible to project research findings to other populations.
• Non-response bias: In many opt-in panels, a small fraction of those invited actually participate, and little effort is made to convert nonresponders. KnowledgePanel makes aggressive efforts to contact nonresponders.
• Panel maintenance: Nonparticipating panelists need to be identified and removed promptly; but opt-in panels often allow nonparticipating members to be considered members even after they have not answered a survey in a year. KN removes panelists after they have missed just six surveys.
• Panelist engagement: Research companies need to make strong efforts outside the survey experience to keep respondents attentive and involved, and help them troubleshoot in real time during survey taking. KnowledgePanel® delivers on both of these needs.
• Replicability of results: At least one prominent retest found that the same survey on the same opt-in panel produced different results when rerun one week later – a sign of disturbing weaknesses in panel methodologies and management. A similar retest on KnowledgePanel® this June produced essentially identical results from wave to wave.

KnowledgePanel® is the only online consumer panel that meets all of these six criteria. Created in 1998 by two Stanford University professors, KnowledgePanel® is based on the premise that online research can and should conform to established principles of sound survey research – most notably probability sampling – while also leveraging the advantages of the Internet.

To learn more about KnowledgePanel® and order the 2006 edition of The Decision Maker's Guide to Online Research™, go to www.knowledgenetworks.com/knpanel.

Rooted in a statistically valid, random sample of all U.S. telephone numbers, KnowledgePanel® provides Internet access to any chosen respondent who does not have it already. This means that KnowledgePanel® is the only online panel to represent the roughly 30% of the U.S. population – some 70 million consumers – who do not have Internet access at home.

This truly representative approach means that companies and institutions using KnowledgePanel® findings can obtain accurate new product and established business marketing projections to the populations of interest; such crucial tasks include

• accurately sizing markets
• gauging interesting products and services among the full U.S. population
• conducting government and academic surveys that demand the highest quality methods

In addition, KN's "we choose them" methodology drastically minimizes the biasing effects common with volunteer panels, such as

• more responses coming from those with a particular interest in the survey topic and/or survey taking generally, and
• inaccurate answers from "professional respondents," who attempt to maximize their profit from survey taking by joining multiple panels and answering questions in whatever way will gain them the greatest payback with the least work.

To learn more about the elements of reliability in online research, go to www.knowledgenetworks.com/reliability.

Knowledge Networks also sets a standard of excellence in its management of KnowledgePanel®, including prompt removal of nonparticipating panelists, rigorous limits on number of surveys received and length of surveys, and an extensive communications program that builds engagement with panelists and provides "live" person-to-person help on survey troubleshooting. This leads to in-panel response rates of 70% or higher.

KnowledgePanel® now has more than 40,000 members nationwide, and KN has collected thousands of data points on its panelists that can be accessed on their own or add valuable depth to surveys conducted on the panel.

Knowledge Networks delivers quality you can use—superior methodologies, design, and analysis that give you an edge when it comes to understanding consumers and making business decisions. KN's unmatched consumer research resources include the only probability-based, Web-based consumer panel. The company leverages its expertise in brands, media, advertising, and analytics to provide insights that speak directly to clients' most important marketing issues. In addition, Knowledge Networks has built a substantial practice in government and academic research, an area notable for its rigorous methodology standards.