News Release

Internet porn-blocking software needn’t block health information, study finds

Peer-Reviewed Publication

Michigan Medicine - University of Michigan

ANN ARBOR, MI – A comprehensive new study of Internet filtering software finds that libraries, schools and parents can bar access to pornographic Internet sites without necessarily blocking important access to health information.

But, the study also finds, setting Internet filters to their most restrictive levels will indeed keep computer users from seeing many health sites – and will only give marginally better protection against porn than the least restrictive setting tested.

The findings, published in the Dec. 11 issue of the Journal of the American Medical Association, come from a scientifically designed study of six filtering packages that was performed for the Kaiser Family Foundation by a University of Michigan team.

The researchers tested which health and pornography sites the different filters blocked – or allowed – depending on how software controls were set. It compared the filters' performance in Internet searches based on 24 health search terms, and in allowing access to 586 pre-selected "recommended" health sites for teens.

Believed to be the most rigorous comparative study of its kind, the project holds important implications for libraries, schools and families in the ongoing debate over how to strike a balance between access to the Internet's wealth of information, and the desire to protect children and teens from exposure to electronic pornography.

The topic was made even hotter by the U.S. Supreme Court's Nov. 12 decision to hear a case challenging a federal law requiring libraries to install computer filters.

"In general, we found that filters were remarkably good at distinguishing between health information and pornography, when set at the least restrictive setting. But at highly restrictive settings, almost a quarter of health sites were blocked, though there was little improvement in porn blocking over the least restrictive settings," says lead author Caroline R. Richardson, M.D., a lecturer in the Department of Family Medicine at the U-M Health System and a research scientist at the Veterans Administration Health Services Research and Development Service in Ann Arbor.

Adds co-author Paul J. Resnick, Ph.D., an associate professor at the U-M School of Information, "Much of the policy debate until now has been about whether filters should be installed at all. Our study suggests that careful thought and debate should also precede decisions about what settings to choose."

Richardson, Resnick and others from the U-M Health Media Research Laboratory and the Kaiser Family Foundation conducted the study with an eye toward teens, who are likely to turn to the Internet for health and sexuality information.

"Filters can strike a good balance between protecting kids from pornography while still giving them access to online health information, but only if they're configured carefully," said KFF vice president and senior author Vicky Rideout, M.A.. "Otherwise, filters can be a serious obstacle, especially on issues such as pregnancy, sexually transmitted diseases, and birth control."

Previous Kaiser Foundation studies have shown that many teens have used the Internet to look for information on health and sexuality, including topics they may not feel comfortable talking about with parents, teachers or health care providers – from depression to sexual orientation.

Three-quarters of schools and nearly half of libraries have installed filtering software on Internet-connected computers, to prevent access to certain kinds of information. Only some schools and libraries disclose the restriction-level that they choose for their software.

In the mid-1990s, there were some reports that such software had blocked access to information on health topics like breast cancer. Since then, filter manufacturers have worked to improve their products' ability to distinguish between pornography and legitimate health information. But until now, no publicly available study had so thoroughly assessed the performance of the most common filters with respect to over-blocking of health information.

In 2000, Congress passed the Child Internet Protection Act, which required filters on computers at libraries and schools that receive federal funds. The portion of the law applying to libraries was struck down in court but an appeal will be heard in the current Supreme Court session.

The U-M and Kaiser researchers performed their study with the help of a specially designed JAVA computer program that conducted Internet searches and stored the results in a database.

The team started with unfiltered searches for 24 health and sexuality terms, and six pornographic terms, using six search engines popular with teens --Yahoo!, Google, America Online (AOL), Microsoft Network (MSN), Ask Jeeves, and Alta Vista.

Some of the health terms were unrelated to sex (for example, diabetes), others involved sexual body parts (such as breast cancer), some were related to sex (for example, birth control), and some controversial health topics (such as abortion).

They then tested whether access to the sites was permitted by seven software packages: N2H2, CyberPatrol, Symantec Web Security, SmartFilter, 8e6, and Websense. All are widely used by schools, libraries, or both. More than 3,000 health and 500 pornography sites were ultimately tested against the filters.

At the least restrictive setting, designed to filter out only pornographic pages, the filter software blocked an average of 1.4 percent of health information sites and about 87 percent of porn sites. At moderate settings, designed to filter pornography, and a few other categories such as nudity and information on drugs and weapons, the filters blocked an average of 5 percent of health sites and 90 percent of porn sites. At the most restrictive settings, which barred a broad range of categories, the blocking of health sites reached an average of 24 percent, but porn blocking only increased to about 91 percent.

Interestingly, the software packages were much better at blocking porn sites that resulted from the pornographic-term searches (89 percent) than they were at blocking the few pornographic sites that turned up when health-related search terms were entered (62 percent).

Certain health search terms resulted in much more blocking -- even at the least-restrictive settings, about 10 percent of health sites turned up by searches using the terms "safe sex", "condom", and "gay" were blocked.

The percent of sites from those terms blocked at moderately restrictive settings was even higher, and at the most restrictive blocking setting, several of the more controversial search terms had blocking rates above 40 percent.

When the researchers turned their attention to a list of 586 health sites that had been recommended for teens by the Yahoo! and Google directories, they found that an average of 0.5 percent were blocked at the least-restrictive filter settings, 2.5 percent were blocked at the moderately restrictive setting, and 23 percent were blocked at the most restrictive setting.

In all, the authors report, setting filtering software to block only pornography will not necessarily have a serious impact on access to general health information, but more-restrictive settings may indeed have an impact. And even pornography-only restriction settings may bar computer users from seeing information on sexual health.

The researchers recommend that software manufacturers devote further research to improving their products' ability to discriminate between pornography and sexual health sites.

They also advocate the adoption of an appeals process by which an Internet site owner can ask that a site be allowed through a filter that blocks it.

Says Richardson, "The Internet makes a big difference in access to health information, and in the future it will make a difference in communication between patients and health care providers, and in modifying health behaviors. We need to be careful not to inadvertently allow pornography-blocking software to cripple the Internet as a tool for improving health."

In addition to Richardson, Resnick and Rideout, the study's authors include Holly Derry, MPH, of the Health Media Research Laboratory, and Derek Hansen, Ph.D., a student at the U-M School of Information. Richardson is a former Robert Wood Johnson Clinical Scholar at the U-M Medical School, under the direction of Joel Howell, M.D., and Rodney Hayward, M.D.

Note: An executive summary of the study and a link to the complete searching and blocking results will be available at http://www.kff.org. The summary will also be available by calling the Kaiser Family Foundation's publication request line at 1-800-656-4533.

Note to editors: The results of the study will be discussed by the authors at a briefing at 1 p.m. EST on December 10, 2002 at the National Press Club in Washington, D.C. Comments from various organizations involved in the Internet filtering issue will also be presented at that time. A webcast of the briefing will be available at http://www.kaisernetwork.org at 4 p.m. that day.

###


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.