Description of Selected Past Projects

Election 2008 & Beyond
Client: University of Chicago
Funded by:Ford Foundation
Period of Performance: 2008 to 2010
Reference: Professor Cathy J. Cohen, Ph.D., David and Mary Winton Green Professor of Political Science and former Director of the Center for the Study of Race, Politics and Culture at the University of Chicago

Project Description: The Election 2008 & Beyond project measured longitudinally the attitudes, knowledge, and behaviors of U.S. adults in the month leading up to the November 4, 2008 general election and subsequently at two points in time after that election (June-July 2009 and December 2009-January 2010). The study was designed to measure the effective political mobilization of racial and ethnic minorities as a result of the Obama candidacy and then to measure the extent to which that mobilization would persist after the election. The study was designed to have adequate sample sizes for measuring the effects of race, ethnicity, gender, and social class. A key component of the project was the oversampling of African Americans and Spanish-speaking Hispanics by using KnowledgePanel Latino.

To meet the goals of the project, 3,181 randomly selected U.S. adults aged 18 and older completed the baseline survey, with just over 13% of respondents being Asian and the rest of the respondents split approximately equally between African Americans, Hispanics, and Whites. Within each race/ethnicity group, approximately 33% of respondents were aged 18-34 with the remaining respondents aged 35+.

To complete the project, KN implemented several steps including the following:

The study samples were a mix of cross-sectional and longitudinal interviews. Fresh cross-sectional cases interviewed for the baseline survey were retained and sampled with certainty for the two follow-up surveys. Sample loss was replaced with the insertion of fresh cross-sectional samples for the follow-up surveys, with the goal of obtaining a minimum of 750 African-American respondents, 750 Hispanic respondents, 750 white respondents, and 450 Asian respondents at each wave, while also maintaining the 18-34/35+ age split within each race/ethnicity group.
KnowledgePanel LatinoSM was used for sampling the Hispanic portion of the population, providing greater coverage of the non-English-speaking Latino population. The sample was designed to reflect the Census distribution of Hispanics by language groupings: Dominant English speakers; Bi-linguals; Dominant Spanish speakers.
The survey was translated into Spanish by experienced bi-lingual survey research staff to facilitate accurate and appropriate language use.
A variety of techniques were employed to minimize attrition between survey waves. These included the retention of baseline survey respondents even if they left the KN panel, enhanced use of respondent incentives, pre-notification e-mails for each survey wave, and a series of reminder e-mail messages and telephone calls to nonrespondents.
Post-wave in-depth interviews were conducted by telephone after the baseline and final followup survey (48 interviews per round), probing the answers respondents provided to the survey.

More information about the research project is available at http://www.2008andbeyond.com

2009

American National Election Surveys, 2007-2009
Client: Stanford University
Funded by: National Science Foundation
Period of Performance: 2007 to 2009
Reference: Professor Jon Krosnick, Ph.D., Stanford University

Project Description: The American National Election Studies (ANES) has involved the collection of national survey data on voting, public opinion, political participation, and other related research topics since 1948. This project is a "public good," meaning that data are collected to serve social scientists throughout the world. As part of the 2006-2009 ANES, a national Internet panel survey of 21 waves is being conducted. A new ANES web panel was created by Knowledge Networks and maintained for this purpose, funded by the National Science Foundation with professional support by Stanford University.

A key objective of the ANES Web panel is to track election-related judgments and behaviors during the course of the 2007-2008 general election campaign and in the post-election period. To permit this, approximately 4,000 randomly selected U.S. adult citizens were recruited to complete monthly surveys. The end result will be a unique resource for social science researchers worldwide for longitudinal analysis.

KN created a new Web panel for the ANES. This panel does not overlap with the existing KN Web-enabled panel (KnowledgePanel®); the ANES Web panel represents a separate sample and was created specifically and exclusively for the ANES. KN built the ANES Web panel in October through December 2007, with additional recruitment conducted during the fall of 2008. Seven of the monthly surveys cover political content specifically for ANES use; the remaining monthly surveys are available to other scholars, who can purchase the opportunity to conduct research on non-political topics. The current effective sample size for each upcoming wave is approximately 2,500 interviews.

In developing this custom panel, KN implemented several steps, including the following:

Selected the random-digit-dialing (RDD) sample
Prepared the sample file for use during telephone-based screening and recruitment
Screened households for eligibility and randomly selected an adult age 18 years or older as of November 4, 2008
Conducted a telephone recruitment interview with the selected respondent
Created a sample file for the ANES Web panel interviews containing the recruited respondent's demographic data, home addresses, e-mail address, and other data collected during the RDD recruitment interview
Conducted monthly interviewing
Minimized attrition via retention measures
Evaluated the effectiveness of the retention measures at maintaining ANES Web panel survey participation over time
Delivered to Stanford University all the data files, reports, and other deliverables

For this project, KN was responsible for all tasks related to panel initiation, panel management, and conducting surveys. In conducting this project, KN leveraged its computer systems, infrastructure, standard operating procedures, and panel management experience acquired as a result of recruiting and maintaining the proprietary KnowledgePanel® (50,000 adult members) since the inception of the company.

National Annenberg Election Survey, 2007-2009
Client: Annenberg Public Policy Center, University of Pennsylvania
Funded by: Annenberg Foundation Trust at Sunnylands
Period of Performance: 2007 to 2009.
Reference: Professor Diana Mutz, Ph.D.: Telephone, University of Pennsylvania, (215) 573-1974

This tracker survey was designed to support longitudinal data analysis for the entire 2008 general election campaign and post-election period, tests for possible panel conditioning effects, and support weekly point estimates for population-based shifts in policy and political attitudes. The study population consists of U.S. non-institutionalized adults age 18 and over. The study population therefore contains both likely and unlikely voters; a likely-voter screen was not employed for sample selection. Tracking participants are sampled from KnowledgePanel, and all panel data collection was undertaken by Knowledge Networks.

The overview of the sample design in terms of time and data collection waves is as follows:

Wave 1: A large pre-primary baseline survey – Between October 1, 2007 and the winter holidays
Wave 2: A political primary period wave – January to March, 2008
Wave 3: A low-intensity wave between the end of the primaries and the start of the general-election campaign – April 1 to August 28, 2008
Wave 4: A general-campaign wave – August 29 to November 4, 2008
Wave 5: A post-election wave -- November 5, 2008 to January 31, 2009

Wave 1 consisted of fresh cross-sectional sample interviews. Waves 2 though 5 consisted of a mix of follow-up interviews of baseline participants and fresh cross-sectional samples. The latter are included for purposes of statistical tests for panel conditioning. The table below summarizes the wave sample sizes:

Table 1: Number of Completed Interviews by Wave of Data Collection and Sample Type

Number of Completed Interviews by Wave of Data Collection and Sample Type

Survey of Young Americans
Client: Harvard University, Institute of Politics (IOP)
Funded by: Harvard University, Institute of Politics
Period of Performance: 2009, on-going
Reference: John Della Volpe, Director of Polling, IOP

Project Description: This is a series of fall and spring surveys of young Americans (18 to 29 year olds), including both those who are attending a college or university and those who are not. The fall survey is slightly smaller (N=2,000 English-language survey takers) than the spring survey (N=3,000 English- and Spanish-language survey takers). The topics cover the major issues of the time period. For example, the most recent survey (Spring 2010) asked about concerns with economic, health care, and national security issues; the job performance of President Obama, Congressional Democrats, and Congressional Republicans; volunteering in community service and participating in political activities; trust in particular individuals and institutions; political, economic, and social views; financial concerns and employment priorities; and the use of social networking tools. We obtained 3,117 completed interviews for the spring 2010 survey. The cooperation rate was 67% for the KN sample (N=2,561, including 273 interviews in Spanish) and 10% for an opt-in sample (N=556). The margin of error for the polls is approximately +/- 2.3 percentage points at the 95 confidence level.

National Poll - Children 08-09
Client: C.S. Mott Children's Hospital (University of Michigan)
Funded by: C.S. Mott Children's Hospital (University of Michigan)
Period of Performance: July 2008 - June 2009

Project Description: The C.S. Mott Children's Hospital National Poll on Children's Health, located within the University of Michigan Health System's Division of General Pediatrics' Child Health Evaluation and Research Unit, measures public opinion, perceptions, and priorities regarding major health care issues and trends for children in the United States. Knowledge Networks conducted a nationally representative household survey for C.S. Mott Children's Hospital via a method used in many published studies. The survey was administered from August 1-31 to a randomly selected, stratified group of adults aged 18 and older (N=2,245) with and without children from the KnowledgePanel. For this analysis, a subset of parents with children age 3-17 years was used (N=1,608). The sample was subsequently weighted to reflect population figures from the Census Bureau. The survey completion rate was 62% among panel members contacted to participate. The margin of sampling error is plus or minus 2 to 7 percentage points, depending on the question.

Public Opinion & Deliberation
Client: University of Pennsylvania
Funded by: National Institutes of Health, National Human Genome Research Institute
Period of Performance: 2008 to 2009

Project Description: This project is designed to assess public opinion and deliberation about ethical issues in genetic and genomic research. The study began with a baseline survey completed by more than 3,700 panelists assigned to one of three groups: a) those who were invited to participate in a series of three online discussion groups in which information on genetics research was presented (before and after which there were brief surveys to assess current opinions on the subject), b) those who were not invited to the discussion groups but did receive the pre-and post-group surveys as well, and c) a control group who was not invited to the discussion groups and did not receive the pre- and post-group surveys. At the end of the project, a final survey will be conducted with all baseline respondents. This project is in progress, with the baseline survey and two of the three online discussion groups completed to date.

U.S. Department of Defense: JAMRS Advertising Tracking Study
Client: Fors Marsh Group LLC
Funded by: US Department of Defense, Joint Advertising, Market Research & Studies (JAMRS)
Period of Performance: April 2006 to Present (under contract through September 2009)
Reference: Fors March Group contact information: Kara A. Marsh, Ph.D., Research Manager, Fors Marsh Group LLC, email: kmarsh@forsmarshgroup.com , Phone number: 703.243.7399

Project Description: In the current era, perhaps more than at any time since the end of the Cold War, our national and domestic security depends greatly on the quality and quantity of an all-volunteer armed force. In this milieu, public attitudes and perceptions about military service and the sacrifices and commitment required are increasingly important, and measuring them accurately is even more vital. This is true not only of the personnel that serve, but also of the potential recruits and the familial networks that support and influence them.

In coordination with Fors-Marsh, LLC., Knowledge Networks has supported this effort for the past three years by providing high-quality survey data, gathered daily, regarding Joint and Service-specific advertising performance on critical topics: general brand awareness, advertising awareness (by medium), imagery, perceptions of the Military, propensity to join the Military, likelihood to recommend military service to youth, and any special interest topics to be determined.

The populations of interest for this study include youth, adults who influence youth, individuals eligible to join the ranks of military reserves, and previous members of the US military. Influencers and Reserves are drawn exclusively from the KN Panel. Youths, who represent a difficult-to-survey population, and prior-service military personnel, who are a rare subgroup within the population, are drawn from the KN Panel supplemented by off-panel sample. These cases are combined using a dual-sampling method of weighting. This dual-sampling approach is based on having a probability-based sample that is augmented with non-probability cases. The two are combined according to results of a pre-identified question that is expected to be most highly correlated with other survey estimates. For this study, the estimate used is propensity to join or recommend military service. The advantage to starting with the KN panel is that it provides a representative base of the U.S. population. The study data from the panel can be analyzed separately or jointly with augment samples from other sources. Knowledge Networks has successfully employed a dual-frame approach, where appropriate, throughout this effort.

The JAMRS Ad Tracking data provides a research infrastructure upon which many important decisions rest. Because of the gravity with which any changes to the JAMRS methodology are justifiably considered, our approach to conducting the JAMRS Advertising Tracking Study is based on the best-practices that we have developed for Internet surveys used in situations requiring high-quality data such as for government-funded studies and those published in first-line academic journals. The principles of these best-practices and assumptions have guided us in our conceptualization of what is important for the successful execution of the JAMRS Advertising Tracking Study. Our essential proposition is that the study is best served by the investment of substantial resources to obtain a high response rate combined with a dual-sampling methodology, where needed, that allows for the augmentation of low-incidence data and an appropriate weighting of those data with the results of a high-quality probability sample.

Knowledge Networks has successfully conducted Years 1 thru 3 of the Ad Tracking Study and is currently under contract for the first six months of Year 4, pending the re-competition for an additional three-year study period.

2008

2007-2008 AP-Yahoo Election Survey
Client: Associated Press
Funded by: Associated Press / Yahoo
Period of Performance: November 2007 to December 2008
Reference: Trevor Tompson, Director of Polling, The Associated Press

Project Description: From November 2007 to December 2008, Knowledge Networks conducted an eleven wave longitudinal election study on behalf of the Associated Press (AP) and Yahoo. The year-long longitudinal design allowed AP and Yahoo to report on the political pulse of the nation and measure opinion change leading up to the primary elections through the Presidential election in November 2008. The interview completion goal for the study was to collect approximately 2,000 completes at the final wave of the study. Another key component of the project was the ability for AP to conduct follow-up telephone interviews with panel members who agreed to speak to a reporter. This allowed AP to include respondent stories and quotes in news articles and reporting throughout the election year.

During the 13 month study KN implemented several steps, including the following:

Selected nationally representative sample from KnowledgePanel® for the longitudinal study
Selection of three fresh cross-sectional samples conducted at waves 3, 6, and 9
Delivered a variety of custom post-stratification weights to allow for analysis of key demographic groups
Fielded 11 surveys across the 13 month study
Program and field surveys on a quick-turnaround basis - specifically during the last few months of the Presidential campaign
Developed incentive plans to maximize response in key demographics and speed of response during the late stages of the study
Handled custom email and phone reminders
Production of top-line reports for AP website and news stories
Arranged for access to KN panel members who agreed to speak to AP reporter
Conducted the Election Day survey with a field period of one day for voter feedback; delivered weighted data the night of the election
Developed retention measures for maintaining the original sample selected for the longitudinal study
Delivered to AP all the data files, reports, and other deliverables
Hosted publicly available website for survey data and reports.

Studies of Spatial Reasoning in Politics
Client: Stanford University
Fund by: National Science Foundation
Period of Performance: February 2006 - January 2008

Project Description: Knowledge Networks conducted a series of brief surveys (approximately five questions) for Professors Paul Sniderman, Michael Tomz, and Robert Van Houweling, with 1,000 surveys per wave. The researchers conducted dozens of experiments with complex randomization to test hypotheses related to factors that influence the party identification in the U.S. electorate. The research from this project has been published recently in the American Political Science Review and also will be published in a forthcoming issue (Tomz et al, 2008; Tomz et al, forthcoming).

2007

Retirement Perspectives Survey: Who Failed to Enroll in Medicare Part D and Why?
Client: University of California - Berkeley
Funded by: National Institute on Aging, UC-Berkeley
Period of Performance: October 2005 to April 2007

Project Description: Enrollment in the Medicare Part D prescription drug benefit began January 1, 2006, and concluded May 15 of the same year. To assess participation in the ground-breaking federal program, economists Florian Heiss, Daniel McFadden, and Joachim Winter used the Knowledge Networks panel for baseline and post-enrollment interviewing. Their project, called the Retirement Perspectives Survey (RPS), began with baseline interviews in November 2005 and sampled panelists age 65 and older. The post-enrollment survey involved a sub-sample of 2,360 cases that had completed the baseline RPS interview in November. More than 90 percent of all seniors sub-sampled for the post-enrollment interview completed the survey in the two weeks following the close of the enrollment period.

In March and April 2007, all of the 2005 interviewed baseline cases were re-interviewed for this study in an extensive survey averaging 35 minutes.

Project Outcomes: Estimates of prescription drug coverage from the KN study were consistent with those released from the Centers for Medicare and Medicaid Services in June 2006 after the Part D enrollment period had closed. While the program did accomplish its primary objective of boosting coverage to 90%, the research team found that majorities of seniors are less satisfied with Medicare and with the government as a result of this program. Their analyses identified vulnerable subpopulations and described which groups failed to enroll despite remaining uncovered for prescription drugs. An early report of their findings appeared in Health Affairs on August 1, 2006, less than three months after the drug-benefit enrollment period ended.

2006

Behavioral Game Experiment Regarding Hurricane Katrina
Client: National Bureau of Economic Research
Funder: National Science Foundation
Period of Performance: March 8 - June 16, 2006

Project Description: Knowledge Networks conducted a study examining people's attitudes and behaviors towards charitable donations to victims of Hurricane Katrina on behalf of the National Bureau of Economic Research. Specifically, the study presented images and audio manipulating different stimuli (e.g., victims' skin color, employment status, household income, and religious beliefs) to see how these factors influenced how much people would be willing to donate to a charity in a state affected by Katrina. Respondents were asked to allocate any portion of $100 to the charity or to keep it for themselves, and they could divide the amount in any way they wished. To increase the salience of the experiment, respondents were informed that Knowledge Networks would pay the amounts indicated for one out of every 10 participants. Each respondent was randomly assigned a number from 0 to 9. The "winners" were those whose number matched the first digit chosen in a pre-specified drawing from the Louisiana State Lottery Pick 3 lottery game. The project was conducted in four phases, three pilot tests and the main study, which was divided into three versions: a full survey for Black and non-Black participants, a version in which no lottery was presented, and a version in which participants were told that race was being studied. Participants in each phase were selected on the basis of having not completed any previous phase of this project or any other Knowledge Networks study about Hurricane Katrina. We prepared a paper discussing project outcomes: "What Determines Giving to Hurricane Katrina Victims? Experimental Evidence on Racial Group Loyalty" (joint with Christina Fong), American Economic Journal: Applied Economics, forthcoming, April 2009.

Boston University: Survey on the Age of Drinking Onset & Related Problems
Client: Boston University Medical Campus, Youth Alcohol Prevention Center at the School of Public Health
Funded by: National Institute of Alcohol and Alcohol Abuse, National Institutes of Health
Period of Performance: December 2005 to April 2006

Project Description: Knowledge Networks was selected in December 2005 for this NIAAA-funded study that included novel experiments for tests of KN panel quality. The study principal investigators are located at Boston University's Medical Campus. KN surveyed 3,000 persons age 18 to 39 with a history of having consumed alcoholic drinks. In addition, there was a telephone survey of 500 KN panelists using the same screening criteria and the same instrumentation, adopted as needed for telephone-based administration. Lastly, there was a third sample for the study called the Non-Response Followup Study or NRFUS. The NRFUS consisted of 500 self-administered Web interviews and some telephone interviews with persons that refused to join the KN panel when selected for panel recruitment. The three study samples provided powerful comparison data sets for isolating the effects on response of mode of data collection versus sample origins and sample composition.

The NRFUS survey, when combined with the KN panel survey data, obtained a 37% weighted response rate based on AAPOR Response Rate No. 3 methodology. The AAPOR Standard Definitions for coding case dispositions was employed. The AAPOR No. 3 Response Rates for the three NRFUS samples were 28%, 63%, and 58%, respectively, for these nonresponse groups that were randomly sampled for the survey: NRFUS: Panel Recruitment, NRFUS: Non-Connected Households, and NRFUS: Post-Profile. KN employed a third-party professional locating service to update our contact information for all the sampled respondents. An advance letter, pre-incentives in cash, and multiple telephone call-attempts (including refusal conversion) were employed.

California Department of Health Services: Evaluation of Media Campaign
Client: Rogers & Associates
Funded by: California Department of Health Services, Tobacco Control Section (CDHS-TCS)
Period of Performance: November 2005 to December 2006

Project Description: In 2005, California took a new direction in its evaluation of its own anti-smoking television campaigns by engaging the Knowledge Networks Panel. Switching from an RDD-based telephone design to the KN methodology added a longitudinal component to measure behavior change and advertisement exposure. In addition to retaining probability sampling, KN's online panel provides lower attrition rates, higher participation rates, and lower overall costs compared to a longitudinal telephone-based survey.

The original study included three survey waves: December 2005, May 2006, and December 2006. Each wave followed a flight of CDHS-TCS commercials. Key data points include smoking and quitting behaviors and extent of advertisement recall, and attitudes towards tobacco-related issues were collected. The online mode of administration enabled respondents to see images of the anti-smoking commercials on-screen to aid recall questions.

Following the 2005-2006 study, the Tobacco Control Section retained Knowledge Networks for a second-year study to be conducted in 2007. The 2007 project included a similar longitudinal effort to track media awareness and quitting behaviors in California. In addition, a national study was conducted with KnowledgePanel to determine non-California benchmarks for smoking attitudes and behaviors. A separate methodology experiment was conducted during 2007 to compare ad recall estimates obtained with streamed video of ads versus an approach using text and still images.

Chicago Council on Foreign Relations: American Public Opinion & U.S. Foreign Policy, 2004 and 2006
Client: Chicago Council on Foreign Relations
Funded by: CCFR
Period of Performance: June 2004 to August 2004, and May to July 2006

Project Description: The Chicago Council (CCFR) and the Program on International Policy Attitudes (PIPA) of the University of Maryland undertook a joint research effort to determine U.S. public opinion about America's role in the world. KN conducted the survey of 1,195 adults using its Web-enabled panel; participants required an average of about 40 minutes to complete the survey. The survey results were published in October 2004 in a report entitled "The Hall of Mirrors: Perceptions and Misperceptions in the Congressional Foreign Policy Survey."

This project was repeated by KN in the summer of 2006.

ENH Research Recruitment of Non-Schizophrenic Controls
Client: ENH Research
Funded by: National Institute of Mental Health, NIH
Period of Performance: August 2003 to August 2006

Project Description: In this NIH-funded study, KN was responsible for administering an extensive battery of items for the assessment of mental health and subsequently for collection of blood specimens for the same research subjects for future analysis. KN worked with the researchers to identify and implement protocols for the safeguarding of the privacy and informed consent rights of human subjects. Respondents had to consent after reading the consent form online and sign the paper version of the consent form in person to be eligible for the study. The study, conducted over three years, required recruiting 4,500 research subjects, with a large oversample of African Americans (approximately 25%).

Project Outcomes: The survey data and the DNA cell line data were used jointly to compare general population and the patient population. It is hoped that through the analysis and comparison, researchers will be able to identify the genes that cause schizophrenia and eventually find a genetic cure for this health condition.

IRS TAB Conjoint Study I & II
Client: Pacific Consulting Group
Funded by: IRS
Period of Performance: January 2006 to July 2006

Project Description: Knowledge Networks conducted two conjoint studies on behalf of Pacific Consulting Group about the services offered by the IRS. The primary objective of both studies was to investigate people's preferences among various means of contacting the IRS to resolve their tax-related issues. The first conjoint study was launched between January and February 2006, and the second study was launched between June and July 2006. Although both studies had similar research objectives, the second study consisted of a larger sample size (N=2,196) than the first study (N=618). Additionally, the second study tested a significantly larger number of IRS services than did the first study. Respondents were those who were responsible for filing their taxes in the past year and not employed by the IRS. Both studies over-sampled those who visited the IRS Tax Assistance Centers in the last 24 months. The second study also over-sampled those whose household income was less than $36,000 per year. The within-survey completion rates were 81% for the first conjoint study and 75% for the second study.

NOAA Longitudinal Seafood Consumption Survey
Client and funding agency: NOAA, National Marine Fisheries Service
Period of Performance: December 2004 to March 2006

Project Description: This project consisted of a two-stage survey involving screening of a general population of U.S. adults and then a main interview of qualified adults. The screening took place with 21,000 households' primary grocery shoppers to identify fish consumers. The screening survey identified households' levels of seafood consumption. This screened-in sample was augmented with 6,000 freshly screened households not previously selected for this study. At least 1,200 primary grocery shoppers representing 1,200 unique households were identified as eligible for the main interview, supplementing the 21,000 that were previously screened. The main instrument provided monthly and quarterly estimates of Seafood Fish Consumption over a 12-month period. The sample design involved three cohort samples that were re-interviewed every fourth month and three additional fresh cross-sectional cohorts that were introduced in months 7, 8, and 9 to maintain monthly samples sizes in the range of 700 to 800 interviews per month, for a total of 9,175 interviews over the 12-month data collection period. Each cohort represents a random sample of cases that were eligible for the survey based on stage 1 screening. In total, 9,175 main interviews were completed over the 12 months of the project.

Project Outcomes: These data will support analysis of within-subject change in fish consumption over the course of the calendar year, as well as supporting estimations of monthly and quarterly fish consumption rates. This information provides NOAA's National Marine Fisheries Service with longitudinal consumption data for both policy and research purposes.

Stratus Consulting, NOAA Coral Reef Protection Survey
Client: Stratus Consulting
Funded by: National Oceanic and Atmospheric Administration (NOAA)
Period of Performance: December 2004 to December 2006

Project Description: This project was a stated preference survey for measuring valuations of the coral reefs near Hawaii. The preference survey measured valuations in light of attributes related to no-fishing zones, cost, harm to shipping, etc. The survey started with a screening of the general population households from the KN national panel. It had two segments: one for 1,000 general population adults and one for 200 users of various outdoor activities related to coral reefs, such as fishing, snorkeling, diving and other activities. KN selected these latter cases randomly from the panel database and then screened them in the field for user status with a four-question screener. The average interview length was approximately 25 minutes.

Project Outcomes: These data were incorporated in an NOAA report to assist in the formation of policies to protect coral reefs and for estimating the non-use value of the coral reefs.

University of Pennsylvania Continuing National Health Survey (ANHCS)
Client and funding agency: University of Pennsylvania
Period of Performance: December 2004 to January 2006

Project Description: In 2005 Knowledge Networks (KN) conducted the Annenberg National Health Communication Survey (ANHCS) for the Annenberg Schools of Communications at the Universities of Pennsylvania (PENN) and Southern California (USC). The survey design consisted of monthly interviews of independent, general population, adult samples drawn from the Knowledge Network's Web-enabled panel. The interview completion goal for the entire year-long project was 3,600 completed interviews. Due to the length of the survey, an incentive was offered.

The survey measured a variety of health-related topics including exposure, behavior, knowledge, and policy. A 15-minute core of the instrument stayed virtually unchanged throughout the year. A five-minute section of the interview was in place for an extended period of time, approximately five to six months. A final five-minute section changed more frequently, approximately every few months. KN conducted two pretest surveys and worked with the PENN and USC to finalize the 15-minute core of the survey. The pretest surveys included half-sample splits in order to test different question wordings and design. For the main surveys, each month PENN and USC notified KN of any changes to the monthly survey. Changes were made accordingly and the monthly survey was fielded usually during the middle of the month with a field period of approximately one to two weeks.

Data delivery was scheduled once a month after data collection for that survey had been completed. KN generated post-stratification weights for each monthly data delivery. The overall survey completion rate for the project, excluding the pretests, was 74%. Based on the execution of the 2005 study, Pennsylvania and USC have contracted with KN to continue the ANHCS study in 2006 and again in 2007.

In 2006, PENN and USC worked with KN to conduct a follow-up survey with participants of the 2005 ANHCS. The follow-up survey design was to conduct interviews one year later with the respondents of the ANHCS from March through July 2006. The monthly follow-up surveys were conducted March through June 2006 with KN panel members who were still active on the panel. Approximately 78% of the original sample was active on the panel one year later at the time of the follow-up survey and KN collected responses from 90% of those cases for an overall follow-up completion rate of 70%.

2005

Harvard University, EPA Water Quality Survey: Main Study
Client: Harvard University
Funded by: EPA
Period of Performance: August 2005 to December 2005

Project Description: The EPA funded Profs. Joel Huber from Duke University's Fuqua School of Business and W. Kip Viscusi of the Harvard Law School to examine the ways people choose between moving to an area with better water quality vs. an area with a better cost of living. They commissioned Knowledge Networks to conduct a general population Web-enabled panel survey to determine the value individuals place on clean lakes and rivers in the U.S. The study employed contingent valuation methodology to examine the amount of money individuals are willing to pay for clean lakes and rivers in their region. Compared to other methodologies, contingent valuation methodology is quite challenging for respondents because it requires them to comprehend and remember significant amounts of information at the same time they are making difficult choices among alternatives that have several simultaneously varying features (e.g., cost and pollution differences across alternatives). Knowledge Networks used color graphics to reduce respondents' burden by making the choice process less cognitively taxing. For each choice screen, respondents chose among three alternatives, with each alternative and its associated features arranged vertically within a matrix. The survey effectively used color and layout to organize the labels and features to make it easy for respondents to choose their most preferred alternative. The choice process was assisted by a design innovation—for each successive choice in a given series, features from previous screens were displayed. Knowledge Networks' graphics capabilities were employed because it allowed new choice features to appear on the screen while information that had appeared on previous screens was displayed in muted text. The within-survey completion rate was 81%, and the survey data were delivered for analysis to the principal investigators 23 days after the survey was fielded.

Project Outcomes: These data are part of the input to a willingness to pay (WTP), contingent valuation model developed by Prof. W. Kip Viscusi of the Harvard Law School. This model estimates the per unit valuation of increased variables in water resource quality, such as fewer chemicals, better oxygenation, fewer pollutants, etc.

Synovate/FDA Survey of Persons with Food-specific Allergies
Client: Synovate
Funded by: U.S. Food and Drug Administration
Period of Performance: June 2005 to August 2005

Project Description: The Survey of Persons with Food-Specific Allergies was funded by the Food and Drug Administration as part of a congressionally mandated study. Specifically, the study was conducted to improve the way package labeling presents information about allergen cross-contact.

The survey design included three separate groups of participants: those with food allergies, those who cook for people with food allergies, and the control group of those who neither have food allergies nor cook for someone with food allergies. Knowledge Networks screened 11,905 participants for food allergies or caring for someone with food allergies. Out of 9,009 respondents, 530 were screened in and completed the survey for food allergy sufferers, 209 were identified as and completed the survey for caretakers of someone with food allergies, and 504 were selected for and completed the control group interview.
The survey instrument included images of food products with varied types of labels regarding food allergies. Knowledge Networks conducted a pretest of the questionnaire with a small number of participants.
Project Outcomes: The FDA reported the survey results to a House sub-committee in a report delivered August 2006. The Knowledge Networks study showed that all three groups prefer labels with signal words such as "Allergy Information: May Contain Peanuts." This and other results from the survey will inform the FDA's future consideration of product label guidelines.

2004

2004 Political Attitudes Survey
Client: Harvard University & University of Arkansas
Period of Performance: September 2004 to November 2004

Project Description: This project used a sub-sample of the Knowledge Networks' panel to conduct a regionally-specific assessment of political attitudes and policy preferences. On behalf of political scientists Sunshine Hillygus at Harvard and Todd Shields at the University of Arkansas, Knowledge Networks gathered 1,150 interviews from the 11 original Southern States (Alabama, Arkansas, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Texas, and Virginia). An equivalent sample was drawn from non-southern states in addition to a third stratum of 500 general population cases.

Project Outcomes: The study's investigation of moral issues in specific regions of the country provides a wealth of opportunity to understand the dynamics of an election where moral issues ruled the day. The researchers first published their work in Hillygus, D. Sunshine, and Todd Shields. 2005. Moral issues and decision making in the 2004 presidential election. PS: Political Science & Politics 38: 201-209. The investigators continue to present other analyses of the data in preparation for additional publications.

Harvard University, USDA, Food Borne Risk Survey
Client: Harvard University
Funded by: USDA
Period of Performance: March 2004 to June 2004

Project Description: This project developed and tested methods for valuing reductions in food-borne risk associated with bacterial pathogens. Willingness to pay (WTP) was elicited to reduce risks of morbidity that vary in duration and severity, and in the conditional mortality risk. Each respondent was asked about two different risks. In each case, the variables that differed were food, morbidity, symptoms, duration, and mortality. These attributes were randomly assigned so that each of the combinations was asked of some respondents, but no respondent answered more than one question about any single level of any of these five attributes. If the respondent had a child, the second question asked about a risk to the child; otherwise, it asked about a risk to the respondent. The last part of the survey had questions about food-handling practices and the standard package of demographics from the panel profile database was attached for each record. This project included a 400-case pilot survey and a 2,000-case main survey.

Project Outcomes: These data are part of the WTP research at the Harvard Center for Risk Analysis (HCRA) under the direction of Prof. James Hammitt. This center studies a variety of health risk behaviors and outcomes funded by a variety of government agencies. This project combines WTP analyses with actual behaviors about food handling. Ultimately these data will find their way into the policy formation literature.

University of California, Irvine. NSF 9/11 Stress and Trauma Longitudinal Survey
Client: University of California, Irvine, Department of Psychology and Social Behavior
Funded by: National Science Foundation
Period of Performance: September 2001 – October 2004

Project Description: The University of California at Irvine Stress and Trauma Study was a longitudinal study funded by the National Science Foundation aimed to determine the process of adjustment to a traumatic event on a national scale. The main objective of the study was to examine the extent to which demographic factors, psychological and mental health history, lifetime exposure to stressful events including the 9/11-related incidences, and coping strategies predict psychological outcomes over time.

This study spanned more than three years with a design of seven waves of data collection. The first wave was fielded to a national probability sample of 3,496 adults drawn from Knowledge Networks' Web-enabled panel immediately after the 9/11 attacks in 2001. A total of 2,729 individuals completed the Wave I survey (78%). In between waves of data collection, KN maintained contact with research subjects to remind them of their role in the longitudinal study. The last wave of data collection (Wave 7) was conducted between September and October 2004—three years after 9/11—with over half of the original respondents from Wave 1 having attrited from the KN panel. Despite this difficulty, the Wave 7 data collection still achieved a 79% survey completion rate as a result of extensive locating efforts and refusal conversion of respondents that had been dropped from the KN panel.

Articles based on the KN-collected data have been authored by Professor Roxane Silver and her colleagues in Applied Developmental Science, Basic and Applied Social Psychology, Canadian Journal of Psychiatry, Families, Systems & Health, Journal of Aggression, Journal of the American Medical Association , Maltreatment & Trauma, and Motivation and Emotion.

2003

The Health Science and Policy "E-Dialogue"
Client: University of Pennsylvania, Annenberg School of Communication
Funded by: The National Science Foundation
Period of Performance: December 2003

Project Description: This project followed the successful use of the KN panel for the Election 2000 E-Dialogue study and used a sequence of Internet-enabled surveys and online group deliberations to explore factors that shape public views on pressing healthcare issues. In this project, the researchers studied ways to maximize discussion and decision quality in online deliberations about the health care system, focusing on issues such as the cost of prescription drugs, consumer choice, and the fiscal health of Medicare. This project expanded the concept of the online discussion technology to include respondent panels with different memberships:

Health care policy "elites" with a significant stake in the issues and some measure of influence over policy outcomes;
Members of health-care "issue publics" for whom health matters are personally important, and who pay greater-than-average attention to healthcare issues; and
Ordinary citizens who take a more-or-less passing interest in healthcare.

Each group of respondents took part in a carefully designed series of surveys and group deliberations, including:

An initial benchmark survey,
Regular periodic surveys on particular issues,
Online group discussions (among 9 or 10 panelists), and
A final wrap-up survey.

The project's goal was to develop real-time information about how various critical segments of the public process key healthcare issues, and possible differences in the way they frame healthcare problems was successfully attained.

RTI Eliciting Fatal Tradeoffs for Valuing Fatal Cancer Risks
Client: RTI
Funded by: EPA
Period of Performance: May 2003 to June 2003

Project Description: In this EPA-funded study, principal investigator George Van Houtven was interested in finding out about individual preferences for reducing cancer related risks relative to other types of risks. He examined how these preferences were affected by the length of latency and morbidity periods and how they differed across different types of cancer. The results show that individuals generally have a strong preference for avoiding fatal cancer risks; however, as expected, this cancer premium is inversely related to the length of the cancer latency period. Significant differences also exist across cancer types. The main implication of these findings is that, particularly for shorter latency periods, standard Value of Statistical Life (VSL) estimates from the literature may understate the benefits of policies that reduce fatal cancer risks.

In total, 1,010 individuals, each from a different household, completed the survey. The within-survey completion rate was 74%. Some respondents were excluded because they failed a practice choice question or did not express preferences for any of the options offered to them. In the end, the size of the analysis sample was 788 respondents. However, the Heckman sample selection model showed no evidence that the differences between the final analysis sample and the initial case release sample led to biased estimates of the coefficients in the second stage model. Respondents were randomly assigned to the following conditions: three different types of fatal cancers (stomach, liver, or brain cancer, each compared to fatal auto death risks); three different assumed latency periods for the cancer (5, 15, or 25 years); and two different assumed morbidity periods for the cancer (2 or 5 years).

Project Outcomes: The principal researcher and other co-authors have written a paper entitled "Cancer Premiums and Latency Effects: A Risk Tradeoff Approach for Valuing Reductions in Fatal Cancer Risks." The manuscript has been submitted to Journal of Risk and Uncertainty.

2002

Stanford University Survey of Health and the Internet
Client: Stanford University and the U.S. Department of Veterans Affairs
Funded by: U.S. Department of Veterans Affairs, National Institute on Aging
Period of Performance: December 2001 to March 2002

Project Description: KN conducted this survey under the direction of a distinguished group of public health researchers at Stanford University who sought to examine the prevalence of the use of the Internet for health care to estimate its impact on health care utilization. Specifically, the researchers used the survey to measure the extent of Internet use for health care among a representative sample of the U.S. population, to examine the prevalence of email use for health care, and to examine the effects that Internet and email use has on users' knowledge about health care matters and their use of the health care system.

KN conducted the survey in December 2001 and January 2002 among a sample drawn from the KN panel. Responses were analyzed from 4,764 individuals aged 21 years or older who were self-reported Internet users. The sample included a large oversample of U.S. military veterans.

The authors reported in their 2003 article in the Journal of the American Medical Associations that approximately 40% of respondents with Internet access reported using the Internet to look for advice or information about health or health care in 2001. Six percent reported using email to contact a physician or other health care professional. About one third of those using the Internet for health reported that using the Internet affected a decision about health or their health care, but very few reported impacts on measurable health care utilization; 94% said that Internet use had no effect on the number of physician visits they had, and 93% said it had no effect on the number of telephone contacts. Five percent or less reported use of the Internet to obtain prescriptions or purchase pharmaceutical products.

The authors concluded that, although many people use the Internet for health information, use is not as common as is sometimes reported. Effects on actual health care utilization are also less substantial than some have claimed. Discussions of the role of the Internet in health care and the development of policies that might influence this role should not presume that use of the Internet for health information is universal or that the Internet strongly influences health care utilization.

To help peer reviewers evaluate the KN panel methodology, the Stanford researchers prepared an extensive document entitled ‘Validity of the Survey of Health and the Internet and Knowledge Networks' Panel and Sampling.' The document contains a review of KN survey results compared to those from national studies (National Health Interview Survey) and from the Survey of Health and the Internet.

University of California Los Angeles, EPA Valuation Project titled "Not All Deaths Are Created Equal," (EPA Grant #446456-JS-22859)
Client: UCLA
Funded by: EPA
Period of Performance: September 2002 to November 2002

Project Description: This project consisted of asking respondents in the US and Canada about their "valuation" of various ways of dying and the associated costs and risks for each outcome. This was a complex conjoint design and analysis that required respondents to view the choice alternatives and attributes of each. There were six distinct operational phases.

Phase 1: The questionnaire was programmed and tested, with full quality assurance. The design was a complex conjoint design with random sorting of attributes. Revisions were made after each laboratory test and the Canadian pretest.

Phase 2: Knowledge Networks conducted cognitive interviews with 40 research subjects in a laboratory provided by KN. In total there were 4 days of cognitive interviews, with 10 interviews per laboratory day.

Phase 3: After the cognitive interviews were completed, Knowledge Networks pretested the questionnaire with a Canadian email sample for a total of 1,000 completed interviews. These data included the Standard Geographical Classification (SGC) codes linked to statistical information from the Canadian census.

Phase 4: For the main study, information packets were prepared, produced, and mailed to the Knowledge Networks panel members sampled for the study. This packet consisted of two pages of visuals aides. UCLA provided the content for the packet. This packet was provided to respondents because of the complex nature of the conjoint choice sets (11 attributes and three alternatives each).

Phase 5: The main study consisted of a 20-minute interview with 4,400 Knowledge Network panelists and included respondent incentives. There was a "field check" after the first 200 cases to confirm that the conjoint programming was correct and that there were no systematic patterns in the randomization.

Phase 6: This was a 10-minute interview of approximately 1,000 parents and legal guardians of minors. These parental respondents had completed the questionnaire in the main study (Phase 5).

KN provided basic demographic profile data on non-responding households with the final data file as well as 25 demographic and social context variables and a set of health profile data.

Project Outcomes: This project continues the line of research on stated preference analysis that DeShazo and Cameron have pursued. (See DeShazo, J.R. and G. Fermo. (2002.) "Designing Choice Sets for Stated Preference Methods: The Effects of Complexity on Choice Consistency," Journal of Environmental Economics and Management, 43(3):360-385. The paper was identified by Ian Bateman, Editor of Environmental and Resource Economics, as one of the three most influential articles of the year at the 2002 World Congress.

2000

America's Experience with Census 2000: Experiments in Mobilizing Public Participation
Funded by: The Russell Sage Foundation, The Ford Foundation, The William and Flora Hewlett Foundation, The Annie E. Casey Foundation, The John D. and Catherine T. MacArthur Foundation, The Andrew W. Mellon Foundation and The Carnegie Corporation of New York.
Period of Performance: February 2000 – April 2000

Project Description: Knowledge Networks was called InterSurvey in 2000, and it worked closely with the U.S. Census Bureau with the financial support of seven foundations to provide real-time survey data on the census environment in 2000. InterSurvey tracked the American public's exposure—through advertising, community mobilization, and news stories—to Census 2000 and some of the consequences of that exposure. Between March 3 and April 13, 2000, 4,673 Americans were surveyed on a weekly basis. A baseline questionnaire and five monitoring surveys constituted the core of the study.

The surveys were among the first conducted with the InterSurvey panel, a representative sample of adults in U.S. households that included both households with and without computers and was recruited using a standard probability sampling method, random digit dialing. In order to gauge the public's reaction to the advertisements for Census 2000 produced by Young and Rubicam, InterSurvey designed the study so that people could view actual television commercials. This multi-media capability was possible with the InterSurvey technology, which conducted surveys with people over the Web via WebTV, a set-top box that connected the household to the Internet through their television set and to which multi-media were downloaded.

Project Outcomes:

The Census 2000 promotion and mobilization campaign was highly successful in increasing the public's awareness, knowledge, and participation.
Exposure to the Census 2000 campaign was particularly effective in the African American and Hispanic communities, as well as for households that received the long form.
Concerns about the census as a violation of privacy increased dramatically in the week before "Census Day" on April 1, 2000.
Growing privacy concerns appear to have had a negative impact on cooperation. Nevertheless, even among those who believe the census is a violation of privacy, those with higher levels of exposure to Census 2000 are more likely to participate than those who are concerned about privacy but who have a lower level of exposure.
Public awareness of the controversy over the Census 2000 long form is widespread and is likely having a negative impact on participation.
Questions on the census long form concerning income and physical and mental disabilities are most highly ranked by Americans as being too personal for the census to ask, though one-third of the public says that none of the questions on the long form are too personal to ask.

A full description of this ground-breaking study can be found in The Hard Count: The Political and Social Challenges of Census Mobilization by D. Sunshine Hillygus, Norman H. Nie, Kenneth Prewitt, and Heili Pals, published by the Russell Sage Foundation, April 2006, ISBN-13978-0-87154-363-9.

The "E-Dialogue" Project for the 2000 Election
Client: The Annenberg School of Communication at the University of Pennsylvania
Funded by: The Annenberg Public Policy Center
Period of Performance: December 1999 – 2000

Project Description: This innovative project followed on the "deliberative polling" projects that had been conducted using a face-to-face methodology and substituted the very expensive face-to-face discussions with online discussions using the KN (then InterSurvey) panel. The completed panel for this project consisted of 900 members from the national household sample who were eligible to vote. There were two initial 10-minute profile surveys of the entire sample in early January. Of this sample, 800 cases were used to construct 60 "virtual focus groups" of 13-14 members each, and 100 cases were used to form a control group.

There were eight events selected during the campaign, and these were the focus for the activities of all but the control group. After completing the initial profile surveys, the panel members assigned to groups completed eight three-segment experiences: a brief 5-minute survey; an issue-specific, moderated, on-line discussion (average of 45 minutes); and a brief 5-minute post-discussion survey. Materials such as text, sound, and graphic images were downloaded to participants in advance of these three-segment experiences. The control group completed all the surveys, but did not participate in on-line chats. A final two-segment "end-of-project" survey was completed with all 1,200 originally profiled panelists.

The complex scheduling of the events and the design of the stimuli for the surveys was done jointly by InterSurvey Inc. staff and Annenberg staff, and the Annenberg staff moderated the on-line discussions which were run using a technology developed for WebTV applications by Iacta Inc.

Reports, papers and presentation from these data are being published by Prof. Vincent Price of the Annenberg School of Communication and include Vincent Price and Joseph N. Cappella. 2002. Civic engagement, social trust, and online deliberation, presented at the May 2002 Annual Meeting of the American Association for Public Opinion Research and Online Deliberation and Its Influence: The Electronic Dialogue Project in Campaign 2000. Paper presented at the May 2001 Annual Meeting of the American Association for Public Opinion Research