Civiqs automates the complex process of conducting opinion surveys, including sampling, interviewing, and modeling results. The Civiqs online survey platform is custom-built from the ground up to maximize responsiveness, representativeness, and accuracy. Civiqs manages and maintains its own online survey panel, to ensure the highest quality responses.

Combining the speed of the Internet with new methodologies for scientific sampling and modeling, Civiqs achieves the precision of telephone surveys at a fraction of the cost.

Why Poll Online?

Scientifically rigorous, sampling-based survey research is the best and most reliable way to learn about public opinion. Civiqs keeps up with Americans’ political attitudes with its large, national, online survey panel, and its custom-built polling service at civiqs.com.

The Internet is the predominant way that Americans communicate today. In 2000, only half of American adults were Internet users, and high-speed residential Internet access was nearly nonexistent. Americans who used the Internet were younger, wealthier, and better educated than those who did not. Nowadays, those differences have nearly vanished. Over 88% of all American adults are online, and 73% have home broadband connections. Another 12% use smartphones with e-mail and web access, even if they do not have high-speed connections at home. The number of cellular phone subscriptions per 100 Americans is greater than 120, and increasing.

“The Internet is the predominant way that Americans communicate today.”

At the same time, the use of landline telephones in U.S. households is collapsing. The rate at which Americans subscribe to landline service is nearly half of what it was in 2000. More than half of all U.S. households do not have any landline telephone service—and adults without landline telephones are disproportionately younger and people of color: populations that would be missed by landline-only surveys. For the first time, at the end of 2016, a majority of American adults were wireless-only.

These trends make it increasingly difficult to conduct polls by telephone alone. Relying on landline telephones to contact a representative sample of Americans is almost guaranteed to produce inaccurate survey results. Yet calling mobile phones adds significant costs and complexity to traditional telephone-based survey methods. Even when Americans can be reached on the telephone, fewer and fewer are choosing to participating in polls: response rates on telephone surveys have fallen from 36% in 1997 to less than 10% today. Response rates on mobile phones are even lower than on landlines. Polls using interactive voice response (IVR, sometimes referred to as “robo-polls”) have still lower response and completion rates, are restricted from calling mobile phones, and, in many states, are prohibited completely.

“The spread of Internet access and the proliferation of handheld mobile devices makes the Internet the future of polling.”

The spread of Internet access and the proliferation of handheld mobile devices makes the Internet the future of polling. Online polling methodologies have been studied extensively, and been found to generate accurate results when researchers follow a series of best practices in sampling, user experience, interview design, and statistical modeling. By polling online, surveys can be fielded at a speed and scale that would never be possible over the telephone. Methodologically sophisticated online surveys have many additional advantages:

  • Very high survey completion rates.
  • Response rates above what can be achieved by telephone.
  • Round-the-clock interviewing, so respondents can answer when it is most convenient.
  • No interviewer effects, as every survey is conducted in exactly the same way.
  • The ability to interview the same people repeatedly over time.
  • Easy integration of voter registration data for list-based sampling.
  • Voter file matching on survey responses.

As with any polling methodology, the quality of the research design and implementation determines how confident you can be in your data and results.

The Civiqs Research Panel

Civiqs owns and operates a large, nationally representative opt-in online survey panel, or list of Americans who have agreed to take polls at civiqs.com. Panelists reside in all 50 states and Washington, DC, and match the contours of the United States along key geographic, demographic, and ideological lines. The Civiqs panel is expanding constantly, with new people joining every day. This keeps the panel fresh with respondents from all states, demographic groups, and ideological perspectives. The size, scope, and quality of the Civiqs panel, combined with Civiqs’ scientific sampling methodology, enables Civiqs to generate representative survey samples from the Civiqs panel.

By managing its own panel, Civiqs ensures that every respondent has verified contact information, every response is legitimate, and that the demographic, geographic, and political characteristics of every respondent can be trusted for accuracy. Civiqs is not dependent upon any third-party organizations for panel management or administration. Civiqs carefully monitors its panel usage to verify that surveys are answered on a timely schedule by a representative sample of respondents. Civiqs has optimized the frequency with which panelists are contacted for research, to keep survey response rates high and limit panelist attrition.

Civiqs uses a variety of techniques to invite new people to join the panel. The most common approach is through Internet advertisements that offer the chance to take a short, introductory survey at civiqs.com. At the end of the survey, Civiqs asks respondents if they would like to take future surveys; and if so, to enter their email address. By submitting their email address, individuals join the panel. Participation in Civiqs polls is free and open to all residents of the United States of America. Panel members must be at least 13 years of age.

Civiqs periodically asks its panel members to provide demographic information about themselves, including birth year, race, and level of education, to confirm that surveys are representative of all Americans. Civiqs occasionally asks panelists for their home location, for statistical sampling purposes, and names, to verify voter registration status. Civiqs maintains extremely strict security measures to protect the safety of all panelists’ data.

Civiqs gives its panelists a chance for their opinions to be heard. In return, Civiqs treats every panelist with care and respect. Panelists may contact Civiqs at any time by writing to help@civiqs.com with questions, issues, or concerns. Individuals in the Civiqs panel who no longer wish to participate may unsubscribe, and thereby leave the panel, at any time. Directions for unsubscribing from the panel are included in every email Civiqs sends.

All Civiqs panelists are covered by the Civiqs privacy policy. Participation in surveys at civiqs.com is governed by the Civiqs Terms of Use. The survey responses that are given by Civiqs panelists are only ever used for research purposes. Civiqs will never contact panel members with advertisements or sales solicitations. Civiqs does not sell or release panelists’ contact information to third parties for advertising or sales purposes.

Scientific Sampling

Civiqs uses a scientific, list-based sampling methodology to determine which panelists are selected to receive polls. Every day, Civiqs conducts a large number of interviews with its own panel on a range of topics of political interest. Samples for these surveys are generated automatically on a daily basis. Civiqs samples can also be used to conduct polls that are designed to be fielded quickly: either web-only, or to complement telephone data collection in a mixed-mode survey.

The Civiqs sampling methodology is engineered to correct for any demographic or ideological non-representativeness in the panel. If groups are underrepresented or overrepresented in the panel relative to their prevalence in the population, sampling directly from the panel will lead to biased results. Civiqs addresses this using a national list of registered voters to control the rates at which individuals from different groups are polled. Panel members from groups who are underrepresented in the Civiqs panel are sampled more frequently, and those from groups who are overrepresented in the panel are sampled less frequently.

As with any sample-based public opinion poll, the characteristics of Civiqs survey respondents—and the responses they give—are subject to random sampling variation. Civiqs monitors a variety of demographic and political measures to assess differences between the sample and population in any particular sample. Based on those factors, Civiqs weights and models the survey results to improve the accuracy of its polling.

To conduct a poll, Civiqs randomly selects a number of target individuals equal to the desired sample size from a list of registered voters. For each target voter, Civiqs identifies members of its panel who are most geographically and demographically similar to the selected voter, using a statistical matching methodology. Civiqs then sends emails to matched panelists, in order of similarity, to invite them to take a poll at civiqs.com. Panelists who do not respond within one day receive a follow-up, reminder email. Emails are sent on a regular schedule until a survey is completed for each target individual. When more than one matched panelist responds to a poll for a selected target, Civiqs only counts data from the most similar respondent. Panelists can expect to receive a new poll approximately once every few weeks.

Civiqs achieves very high rates of congruence between the demographic profile of the target individual and the characteristics of the survey respondent. After sampling, any remaining differences between the distribution of the sample and the distribution of the population are reconciled in the post-interviewing weighting and modeling analysis.

Interviewing

Surveys that are quick, easy-to-use, and mobile-friendly keep survey respondents happy and engaged. Civiqs recognizes that a well-designed survey interface, on both desktop and mobile devices, encourages respondents to complete surveys truthfully and completely. When interviews are long, boring, or confusing, respondents may answer inaccurately, or not respond at all. Civiqs provides an intuitive and convenient survey-taking experience that enables panelists to easily share their opinions and have their voices heard.

When a panelist is chosen to participate in a Civiqs poll, he or she is sent an email with an encrypted link to a survey at civiqs.com. Clicking the link brings up the survey in the panelists’ web browser or smartphone. Panelists click or tap their responses to each question to move through the questionnaire. Results are submitted automatically. Civiqs monitors the rates at which invitation emails are opened and clicked to ensure that delivery rates are consistent across different demographic groups, dates, panel recruitment cohorts, and email service providers. Civiqs polls regularly achieve response rates at double the industry average for online survey research.

Unlike in a traditional telephone poll, Civiqs does not need to ask a battery of demographic questions at the end of every survey. In most cases, Civiqs has already recorded this information from previous polls. This removes length and redundancy from Civiqs questionnaires, and streamlines the poll-taking experience. Civiqs will occasionally ask panelists to update their demographic information; particularly if the panelist has moved to a new residence.

Civiqs uses question formats, wordings, and layouts that are optimized for the web. By keeping surveys accessible and short, panelists complete Civiqs polls quickly, with almost no drop-off. Most surveys are finished in under a minute. Civiqs questionnaires are designed to encourage respondents to answer all of the questions, but that is never required. Respondents may answer as many or as few of the questions as they are interested in and feel comfortable with. There is never any charge to Civiqs panelists for participating in Civiqs research polls.

Optimizing question layouts for mobile users is critical, as more than half of Civiqs surveys are completed on panelists’ mobile devices. Civiqs tests question formats on both desktop and mobile interfaces to ensure ease of use, and uses only those proven to be most effective for collecting accurate responses. Civiqs avoids the use of grid layouts, sliders, and other question configurations that are difficult to read, difficult to interact with, or tedious to answer.

Civiqs’ results have been validated and tested for accuracy by leading voter file providers and survey firms. Because Civiqs surveys are all online, every Civiqs poll is conducted identically, which eliminates interviewer bias and minimizes social desirability effects. Civiqs data shows no patterns of data falsification that plague many polls; especially IVR surveys and long telephone questionnaires. Civiqs surveys also avoid the problem of straight-lining, in which respondents give invalid responses in order to complete the survey as quickly as possible. As an independent survey research organization, Civiqs does not engage in push polling or any other advocacy activities that masquerade as scientific research.

Civiqs’ interviewing service can send questionnaires to a representative cross-section of the Civiqs survey panel the next day. Results are delivered as an anonymized dataset, just as from a traditional call center. Every dataset is reviewed for quality, with close attention to detail. Please contact Civiqs at inquiries@civiqs.com to request an estimate.

Daily Tracking Models

The Civiqs public opinion tracker conducts daily interviews on a range of important political issues. The tracking questions include topics of interest to many investigators:

  1. Job performance and favorability ratings of politicians, political parties, and other groups.
  2. Attitudes on political issues including healthcare, foreign policy, abortion, the environment, minimum wage, marijuana legalization, and gun control.
  3. Support for the legislative priorities of the Trump administration, including a border wall, repeal of the Affordable Care Act, job creation, infrastructure, and immigration.
  4. Economic perceptions: current, past, and future.
  5. Vote intention at the state and national level, including a generic House ballot.

To produce daily measures of public opinion, Civiqs uses a proprietary statistical tracking model to correct for random differences between the demographics of the sample and the population, and smooth out day-to-day sampling variation. Modeling is performed with a dynamic Bayesian multilevel regression model with post-stratification weights, or MRP. The model enables Civiqs to detect systematic changes in public opinion with a high degree of statistical sensitivity.

The Civiqs tracking model captures shifts in the attitudes of various groups over time. These changes may be either gradual or rapid. Civiqs estimates the relationship between panelists’ attitudes on each question and their gender, age, race, education level, and party identification, on every day of polling. The model is specified using all combinations of two-way interactions between categories of the predictor variables, to improve model fit, give the model greater flexibility, and produce more reasonable measures of posterior uncertainty.

To capture trends in public opinion, the model applies a Bayesian dynamic linear model (or Gaussian random walk) to the effects of the demographic factors. The model’s estimate of attitudes within every subgroup on any given day starts from the previous day’s estimate, and is then adjusted upward or downward, based on the new polling data observed that day. In this manner, the model generates smoothly trending estimates of public opinion for every combination of the demographic predictors on every day—even for small groups with limited data. Estimates of the proportion of each subpopulation who select each response to a question are constrained to sum to 100% over the possible choices.

As attitudes shift from day-to-day within each demographic subgroup, so do attitudes at higher levels of aggregation, up to and including the topline population level. Civiqs uses population data from a list of registered voters to combine small subgroup estimates into estimates of attitudes in larger groups: for example, the attitudes of all women, or all self-identified Independents. In the post-stratification step, the estimated conditional response probabilities in each subgroup are multiplied by the share of the population in the respective group, and then summed across groups. This weighted-sum post-stratification approach is also used to produce 95% posterior credibility estimates around the trendline estimates.

The Civiqs results dashboard enables users to explore the results of its daily tracking polls. The MRP model generates a rich set of trendlines that give researchers the ability to drill down into the attitudes of even small demographic and geographic subgroups. Civiqs’ results dashboard provides an intuitive interface to see the effects of hundreds of combinations of demographic factors, to instantly track where an issue or candidate stands with the public, and how those attitudes have trended over time.

Custom Panel Research

Small, targeted populations are difficult to survey with traditional representative sampling methods. They typically require an oversample that adds cost, time, and complexity to data collection. A custom online survey panel with exactly the right types of respondents is an efficient and cost-effective solution.

Civiqs can recruit a custom panel of survey respondents with nearly any demographic, geographic, or political characteristics. For example, persuadable voters within a congressional district, voters who are highly likely to turnout in an upcoming election, or high-school educated men under the age of 35. Civiqs emails panelists when a new survey is available, and collects response data using its interviewing platform at civiqs.com. A custom panel is for the exclusive use of the specified research project. Civiqs only contacts custom panel members if a new research poll is available for that project.

Civiqs recruits survey respondents for custom panels by mail or online, in consultation with researchers to find an approach that best suits their needs. In the online approach, Civiqs displays targeted Internet advertisements to web users who meet the characteristics of the custom panel. The advertisements invite recipients to click a link to a short poll at civiqs.com, where they can opt-in to receive future surveys by providing their email address. Civiqs displays ads until the panel reaches its intended size.

In the mail option, Civiqs sends invitation letters to a preselected group of households. The letters contain information about Civiqs and instructions for joining the research study. Individuals who receive invitations to join a panel are provided a personalized identifying code that they can use to opt-in at civiqs.com. Panelists who visit the website, verify their identity, and submit their email address are entered into the custom panel to receive future polls. In both the mail and online options, Civiqs can selectively target individuals from particular groups so that the panel has sufficient numbers of respondents with certain demographic, political, or geographical features.

Civiqs can offer paid rewards to custom panel members for answering polls. Immediately after a respondent finishes a poll, Civiqs sends an email containing an electronic gift card, or donation to a charity of the panelist’s choice. The rewards may be as small as $1, or as much as $20 or more. Civiqs does not charge any additional fees for delivering reward payments.

Members of Civiqs custom panels, as with the general Civiqs panel, are covered by the Civiqs privacy policy and Terms of Use. Civiqs will never send advertisements or sales solicitations to custom panel members. Civiqs does not sell or release panel members’ contact information to third parties for purposes of advertising or sales.

Field Experiments

Civiqs offers an all-in-one experiment administration service to help campaigns and advocacy organizations test the impact of real-life persuasion efforts—such as door-to-door canvassing, phone-banking, digital ads, and direct mail—quickly and inexpensively, using online polls. Civiqs’ time-saving expert project management ensures the successful completion of every project from start to finish, including mail, online surveys, data management, and incentive fulfillment.

Civiqs employs a repeated survey design for persuasion experiments that is less expensive, faster, and more precise than traditional approaches. Field experiments are the best and most rigorous way to measure the effectiveness of a political campaign, message, or advertisement in the real world. The repeated survey design, which uses web surveys and an pre-recruited online panel, outperforms many current approaches which, in addition to being expensive and labor-intensive, do not have the statistical power to detect what are often small persuasion effects. Civiqs researchers guide every facet of the study, making implementation easy.

To conduct a field experiment, Civiqs recruits a panel of survey respondents unique to each study. By mail or online, Civiqs invites selected individuals or households to join a research study at civiqs.com. Voters who sign up with a verified email address and answer a baseline survey can be sent an immediate electronic cash reward. Civiqs employs a gold-standard double opt-in procedure to validate every email address at sign-up, so follow-up surveys are delivered accurately and efficiently, with high response rates.

Following the baseline survey, Civiqs randomizes treatment and control (placebo) conditions across surveyed panelists. When the persuasion campaign is complete, Civiqs emails panelists with one or more follow-up surveys to measure any changes in attitudes across treatment conditions or between repeated surveys over time. Unlike telephone surveys—which require lengthy phone operator time to contact voters—Civiqs web surveys are emailed automatically following the treatment, with no additional labor for data collection, and very high re-contact rates. The panel design ensures that the same individuals are interviewed both before and after the persuasion campaign, which greatly increases the statistical power of the research design.

With the repeated survey design, there is less wasted effort, leading to lower costs. No time or money needs to be spent persuading—or attempting to follow up with—voters who will never respond to a measurement survey. Every voter who receives the treatment will have already agreed to answer future polls. Since the repeated survey design requires both fewer contacts and fewer respondents to achieve the same precision of results, it is feasible to conduct placebo-controlled studies to compare attitudes among contactable voters in treatment and control groups.

The experimental methodology of the repeated survey design is based on the research of Professor David Broockman of Stanford University and Joshua Kalla of the University of California at Berkeley.