The parallel test study consisted of a seven-question survey completed online by 510 U.S. Lightspeed Consumer Panel members and by mobile phone by 510 U.S. Lightspeed Mobile Panel members. The survey was also completed online by 500 U.K. Lightspeed Consumer Panel members and by mobile phone by 500 U.K. Lightspeed Mobile Panel members. Survey invites – by text to the mobile survey participants and by email to the online survey participants – were sent at the same time for each test condition. The survey was open for three days and no reminder invitations were sent.

The results of the study clearly demonstrated the significant strengths of a mobile survey, particularly in regard to response rates, response times, and demographic reach.

Response Rates and Response Times

In the U.S., we found that while the overall response rates of the mobile and online surveys were similar (18% for mobile vs. 16% for online), the response rates of younger male and female respondents (ages 18 to 44) were much higher for the mobile survey (Table 1).

U.S. Response Rates

Table 1. Parallel test response rates for U.S. mobile and online respondents – Total by gender/age

The U.S. response times (the time that elapses between a sent survey invitation and the start time of the survey) demonstrated the mobile survey’s ability to reach a target audience quickly. Over a third of our U.S. mobile response times were within 6 minutes, compared to 1 hour and 45 minutes for our online survey (Table 2). The mean response time for the mobile survey was 3 hours and 18 minutes, compared to 6 hours and 19 minutes for the online survey.

In the U.K., the response times for the mobile study were even more impressive. We received 75% of our U.K. mobile responses within 49 minutes of sending out the survey invite (Table 3). The mean response time for the U.K. mobile survey was under one hour.

Table 2. Parallel test response times (the time that elapses between a sent survey invitation and the start of the survey) percentiles for U.S. mobile and online respondents

Table 3. Response times percentiles for the U.K. mobile and online respondents

Demographic Composition of Samples

The mobile survey respondents tended to skew younger than the general population in both the U.S. and U.K. studies. The U.S. mobile panel was more successful in delivering traditionally hard-to-reach 18 to 24-year-olds, and especially significantly more successful in delivering 25 to 44-year-olds (Table 4).

Table 4. Age composition of U.S. mobile and online samples compared to Census

Also, the U.S. mobile panel was much more successful in delivering minority respondents. The two test conditions performed very similarly to each other and to Census on income.

In the U.K., we observed a similar dynamic, as the mobile survey was able to reach more 16 to 44-year-olds than the online survey (Table 5).

Table 5. Age composition of U.K. mobile and online samples

Survey Responses

The responses to the actual survey questions revealed intriguing trends as well. In the U.S., our mobile panelists were much more likely to report responding to texts more quickly than our online panel respondents (Table 1).

In general, do you respond quicker to emails or text messages?

Table 1. U.S. mobile and online respondents

However, when we dissected the U.S. online survey results by age group, we saw that the differences were not driven by the survey mode. Rather, they were driven by age (Table 2).

In general, do you respond quicker to emails or text messages?

Table 2. U.S. online survey questionnaire results by age

Looking at our U.S. mobile survey results, the trending of older respondents who respond to emails more than text message is not nearly as strong. This suggests that the mobile platform is capable of receiving much faster responses not only from younger mobile survey participants, but also from older participants (Table 3).

In general, do you respond quicker to emails or text messages?

Table 3. U.S. mobile survey questionnaire by age

In the U.K., the proportion of those who more quickly respond to texts than email is close to two to one for both test conditions (Table 4).

In general, do you respond quicker to emails or text messages?

Table 4. U.K. mobile and online respondents

Mobile panel respondents in the U.S. were more likely to say they carry their mobile phone all the time, but respondents from both test conditions (98.3% mobile and 87.6% online) report carrying their phones all or most of the time (Table 5).

How often do you carry your mobile phone?

Table 5. U.S. mobile and online respondents

In the U.K., we observed similar values for both test conditions, with mobile respondents slightly more likely to carry their cell phone all the time (Table 6).

How often do you carry your mobile phone?

Table 6. U.K. mobile and online respondents

Both test conditions checked their email with the same frequency. While mobile respondents were more likely than online respondents to say they check their email every few minutes, 85.6% mobile and 83.7% of online respondents report checking their email either every few minutes or several times day. Mobile respondents were also slightly more likely to report spending more time online.

Conclusions

The study results reaffirm our confidence that Lightspeed Mobile Surveys can deliver specific data within minutes to a younger, traditionally hard-to-reach audience. It can also uniquely reach an audience on the go and gives you the power to receive in-the-moment feedback.

We have applied the same leading panel management methodologies used on our online panels to the Lightspeed Mobile Panel. The panelists are regularly engaged through a monthly Mobile Survey. We closely monitor the life of each panelist from recruitment, to activity level, to ongoing profiling — ensuring effectiveness and usability.

Based on our experience, we can confidently predict response rates through our Lightspeed Mobile Panel.

In the next phase of Lightspeed Research’s comprehensive mobile panel test, we will examine prior mobile panel studies longitudinally by analyzing trends in response rates, completion rates, and speed of response patterns for individual studies. We will compare them to the same metrics used for our online panel. We will also compare the demographic composition of our U.S. mobile panel to the online panel and document changes over time.

For more information about how Lightspeed Mobile Surveys can deliver the data you need to make informed business decisions, contact one of our regional teams:

The Americas
1 (908) 630 0542

Europe
44 (0)20 7896 1900

References:

Lozar Manfreda, K., Bosniak, M., Berzelak, J., Hass, I., Vehovar, V. (2008). “Web surveys versus other survey modes: a meta-analysis comparing response rates.” International Journal of Market Research.

We found that respondents may tolerate longer surveys if the surveys are perceived to be engaging. When examining response and drop-out rates, we also found the level of engagement in a previous survey is strongly associated with subsequent respondent behavior. In this issue of Lightspeed Ahead, we examine how completion times and demographics affect respondent engagement. We will also further examine the validity of our findings by applying the Engagement Index to four similar surveys.

Age and Gender

Based on the data collected from over 280,000 respondents, women tended to be more engaged than men. The ‘low” and “medium” respondent engagement classifications were comprised of 54% women and 46% men, whereas the “very high’ engagement classification consisted of 59% women and 41% men (Table 1). Interestingly, our youngest (16 to 24 year-olds) and oldest (65+) respondents tended to exhibit the lowest engagement levels (Table 2).

Respondent Gender within Engagement Categories

Table 1. Respondent gender within engagement categories (N = 283,245)

Respondent Engagement within Age Groups

Table 2. Respondent engagement within age groups (N = 283,245)

We also examined four studies from our United Kingdom online panel to discover if the comprehensive research findings from our 280,000 respondent Engagement Index study were evident at an individual-study level. We carefully selected studies that shared lower than average engagement scores, the same survey length, and had relatively similar formats. We deliberately selected survey topics that varied widely – mobile phones, bread, business banking, and television shows. By controlling for survey design and length, we were able to isolate the effects of respondent demographics on engagement as much as possible.

Gender did not appear to be an important engagement factor in any of the four closely-examined U.K. studies. However, the four studies did have very similar age trends to the longitudinal analysis gathered from the 240+ studies. Our youngest (16 to 24 year-olds) and oldest (65+) respondents tended to exhibit the lowest engagement levels (Table 3).

Online survey research faces a great deal of competition for respondent attention and expression, ranging from online social networks, alternative opinion sites, and even escalating work hours. Recognizing that a time-pressed respondent may require special effort to engage in a survey experience — and maintain that engagement throughout — is a first step in managing the new realities of survey research.

We are learning that respondent engagement has many dimensions, indicators and related behaviors. Lightspeed Research is part of the Kantar Group — the information, insight and consultancy division of WPP, the world leader in communications services. Kantar recently instituted a series of seven questions asked at the end of every online survey, designed to address respondent satisfaction and assess survey experience.

By now, these questions have been asked of hundreds of thousands of respondents participating in many different types of surveys and industry categories for many clients. With the ultimate goal of establishing meaningful baselines and/or norms, these survey experience measures ask respondents to assess how enjoyable the survey was; whether it was interesting, easy to complete, repetitive, relevant or long; and how likely they are to participate in a future survey.

This wealth of data will allow us to answer many questions about respondent experiences with our surveys and, over time, will provide clues to help Lightspeed Research and its clients optimize the survey experience. Eventually, we should be able to predict which surveys will be more (or less) engaging by associating survey characteristics with survey satisfaction levels.

We were interested in first exploring the relationship between a survey experience and panel retention over time. Is a highly engaged respondent more likely to accept future invitations, less likely to drop out of surveys, once undertaken, and overall more likely to cooperate in the future?

To explore this question, we gathered data on a large number of respondents who had taken a survey between September and December 2007, and then observed their survey behavior over the next six months, through June 2008. We collected data concerning the number of invitations sent to each respondent, invitations accepted, drop-out rates in attempted surveys, and other related metrics.

Looking at more than 182,000 cases, including over 240 surveys in which respondents completed the seven survey satisfaction questions, we developed an “Engagement Index.” The index was constructed by using factor analysis against the seven variables, then weighting each according to the variance explained by each variable. Although all seven variables are correlated, and in the expected direction, surprisingly, the variable that contributed most to the overall variance was “enjoyment” of the survey experience (47.2%), followed by how interesting the survey was (15.7%), and how easy the survey was thought to be (10.1%.) Interestingly, perceived length of the survey contributed only 6.3% of overall variance. Indeed, in a recent Kantar Operations analysis of 3 million online surveys, length was the cause of drop-outs in only one-fifth of the cases. This suggests that respondents may tolerate longer surveys if they find them engaging — lending even more urgency to the quest to discover what constitutes an “enjoyable” and interesting survey from a respondent point-of-view.

Once the Engagement Index was established, and using a subset of 25,808 respondents (10,374 EU and 15,434 in the US), we were able to conduct a variety of additional analyses, by level of engagement and with respect to the behavioral metrics over the six-month period following the survey experience.

It turns out that level of engagement in a previous survey appears to be associated rather strongly with subsequent behavior, as measured by response rate, completion rate, and drop-out rates over a six-month period.

As these results show, the survey experience affects future cooperation and panelist retention. We see significant differences in panelist behavior over a six-month period after the initial survey was taken when controlling for level of engagement. Additionally, the survey experience strongly affects respondent satisfaction.

Several questions remain to be answered, and we are mining our data to learn more about the relationship of engagement to a variety of issues such as response patterns within an individual survey. Additionally, we will investigate what the notion of “enjoying a survey” means to a respondent, especially since this factor appears to account for so much of the overall experience. We will also be exploring the extent to which survey design characteristics can be associated with engagement level, and whether engagement might be an innate respondent quality or variable depending on exposure to a survey experience, or perhaps a combination of the two.

Practical outcomes of this research are expected to include:

• How and when intervention(s) to prevent panel and survey attrition due to lower engagement levels can best be practiced
• Which survey characteristics are most strongly associated with consistently high respondent engagement levels and how to apply this knowledge to future survey designs

Lightspeed Research has already committed additional resources to gain a better understanding of respondent engagement. For example, in November 2008, our London office conducted eye-tracking and usability testing among a sample of panelists in order to evaluate how panelists actually interact with our online surveys. We will soon be able to apply these results and consult more effectively with our clients on the multiple design factors that may influence respondent perception and engagement.

We have already worked in partnership with several clients to improve the survey experience overall, and have run multiple pilot studies to help improve interface design and questionnaire dynamics. Tests demonstrate that data quality significantly improves when these steps are taken.

References

Yan, T., Conrad, F.G., Tourangeau, R., Couper, M.P. (2007). Should I stay or Should I Go? The Effects of Progress Indicators, Promised Duration, and Questionnaire length on Completing Web Surveys. Paper presented at the American Association for Public Opinion Research (AAPOR) 62nd Annual Conference.

Whether you’re a Lightspeed Research client, prospective client or other research industry professional, our new site welcomes you with easy-to-follow navigation, a refined graphic design, and complete information about our products, services, resources and more.

“We built the site using a customer-focused design process,” said Seth Ruoss, Executive Vice President of Marketing and Strategy. “We spent significant time trying to understand our site visitors and their needs. The number one priority for us was to deliver a website that enables visitors to easily find the information they want. In addition to our world class global online panels, we have a suite of diverse research products and services, so it is crucial that our content is well organized through an easy-to-navigate site.”

Along with details about Lightspeed Research’s products and services, the new site also features a new Resource Center. There you’ll find white papers, press releases, and access to recent issues of this newsletter — including an archive of the insightful research articles, invaluable tips and “Ask the Expert” advice. You’ll also find the new Lightspeed Research Panel Book available for download, listing complete information on our panel demographics and specialty attributes available by country.

Also just added is a Careers section featuring employee insights about “What It’s Like” to work at Lightspeed Research, as well as our Job Opportunities.

The site’s user-friendly, intuitive design makes it easy to find answers to your questions about how Lightspeed Research can deliver the information you need to make informed decisions. If you have a unique question not answered on our site, please send us an email using the “Contact Us” page. Or for immediate answers, you can always phone your regional office — listed under About Us/Office Locations — or call us at 908-630-0542.

If you have any questions, comments or suggestions to improve our site, please Contact Us.

ANSWER:
“Grids,” or multiple survey questions arranged in a block that utilize a common scale, do present some challenges. However, proactive attention to design, survey psychology, and some judicious editing upfront can dramatically improve the effectiveness of question grids as well as data quality.

First, scrutinize grids for questions that appear repetitive, such as those that repeat a phrase or alter the question or statement by only a word or two. These may be tough for even very attentive respondents to differentiate, sometimes resulting in data that looks like “satisficing” has occurred.

Similarly, avoid questions that use language or brand descriptions that fail to express concepts from the user or target audience perspective, particularly in grids. It’s easy to make the mistake of writing a questionnaire as though the respondent were a brand manager, deeply engaged in thinking about every nuance of the product and familiar with the technical terms used in brand marketing. But as every researcher has learned at one time or another, this particular error can produce very poor data. Allowing both the time and money for a brief pretesting period or soft launch will quickly identify these problems.

Perhaps the most important goal is to recognize that respondents face a distinct cognitive burden when participating in a survey, and to consider all methods that may reduce this burden while maximizing data quality outcomes. There is a unique psychology to survey response and interactions with online surveying, and it is critical to be aware of how these affect outcomes.

Online survey design can offer some options that may also improve grid performance. For example, some studies have shown that providing visual feedback to the respondent, such as darkening the entire box around the selected radio button in a grid row, helps orient respondents within the grid and prevent attrition (Kaczmirek, 2008). Apparently, in one test more than one-third of respondents clicked around the radio button they intended to select but failed to properly select it. An immediate or dynamic visual feedback design alerts the respondent to the missed selection before they progress to the next selection. This is arguably a more user-friendly approach than a validation error message to a missed question, and the technique improved response by more than 3.5%.

Another design option that is frequently — and hotly — debated is whether “progress indicators” (visual or verbal feedback on the page indicating respondent progress in the survey) improve response and data quality, particularly where grids make up a large proportion of a survey or extend over several screens/pages. Some studies have found that progress indicators have no effect in preventing attrition (Heerwegh, 2004), while others show a negative effect (Crawford, Couper, and Lamias, 2001) or slightly positive effect (Couper, Traugott, Lamias, 2001).

More recently, it appears that the type of progress indicator used may be important in producing the desired effect of motivating respondents and improving cooperation and quality. Some survey deployments that use progress indicators suffer from “jumps” due to filter questions, so the respondent seems to suddenly move ahead unexpectedly. This tends to confuse the respondent, or convince the respondent that the survey software has produced an error. High dropout rates are often observed at this point.

Other progress indicators are “dynamic,” smoothly moving the respondent through the survey while hiding the effect of filter questions or piping. Progress Indicators that showed “jumps” resulted in much higher attrition than no progress indicators at all, but dynamic indicators fared about the same as none at all. This suggests that if tempted to use a progress indicator to move respondents through a lengthy grid section, it would be advisable to avoid “jump-type” indicators and look for a dynamic implementation, if possible.

As far as deploying fancy Flash designs, animations, etc., as a way to improve respondent attention, a recent study actually showed that these increased nonresponse by at least 6%, which is problematic, so this issue will require additional study. The biggest improvement in response — and data quality — appears to be associated with greater attention to survey usability, even more than increasing incentive payments. This also corresponded with higher respondent engagement/satisfaction at the survey conclusion. (Kaczmirek, 2008)

References

Tourangeau, R., Rips, L., and Rasinski, K. (2007). The Psychology of Survey Response, Cambridge University Press.

As the use of online market research continues to rise, so does the industry’s concern about data quality. In fact, a recent article in Research World (May 2008) reveals results of a survey of 700 US market research professionals conducted by Market Research Careers in January 2008. Quality of data was rated as the most important attribute in selecting a research vendor – trumping quality of insights, cost, and speed of delivery.

Concerns about data quality are multifaceted. Some issues are the responsibility of the industry at large, such as keeping surveys engaging and relatively short. Yet all too often, the industry has migrated long, tedious surveys from offline to the Internet with the goal of preserving historical trends and norms.

Other issues are clearly the responsibility of the online sample provider:

• Who are the people taking our surveys? How do we know they are who they say they are?
• When we need multiple sample sources, how can we ensure respondents are unique and aren’t taking the survey more than once?
• How can we ensure people are providing honest, thoughtful answers?

Lightspeed Research takes these issues seriously, continually reviewing new technologies to build our fraud-fighting arsenal. As a result, we’re developing Lightspeed RealRespondents and Lightspeed RealResults. These major components of our fraud detection programs are designed to ensure honest, thoughtful opinions from real survey respondents. Unlike off-the-shelf products and services, our solutions will be fully integrated into every phase of our business – from our recruitment process, to our panel management system, to our sampling tools.

Fighting Fraudulent Panelists

An industry leading verification process, Lightspeed RealRespondents is a series of real-time check points new panelists must pass when registering for our panels. Each step casts a wider net, providing 99.5% accuracy in eliminating fraudsters. All steps as outlined below will be in place by the end of 2008.

Key checkpoints include:

• IP Address Verification – Identifies registrants who have a blacklisted IP address (the numeric address of a computer on the Internet) as a result of past fraudulent activity
• Proxy Detection – Detects a proxy server used to mask the registrant’s true IP address and past fraudulent activity
• IP GeoFencing – Locates the registrant’s country of origin through their IP address and redirects the registrant to the appropriate country’s panel registration survey
• Postal Address Verification – Verifies the registrant’s postal address and zip/postal codes against a global address directory
• CAPTCHA – Prevents automated programs from joining our site through a challenge-response test
• Unique MachineID – Tags each computer with a unique ID generated by proprietary algorithm to ensure only one registrant per computer can join our panels
• Email Address Verification – Queries our database to ensure the email address is unique (all panelists must verify their email address through a double opt-in registration process)

Catching Questionable Data

Leveraging both our technology and survey quality management expertise, Lightspeed RealResults identifies and removes questionable survey data through a series of quality checks. Respondents who don’t participate in the survey to the best of their abilities are identified, and all survey answers they provided are removed from the final data. Their actions are reviewed for possible removal from the panel.

Lightspeed RealResults employs the following automatic quality checks:

• Respondent Engagement – At the beginning of surveys, respondents must agree to provide honest, thoughtful answers to each question
• Survey Speedsters – Respondents who rushed through the survey are identified by comparing survey completion times to the norm
• Grid Speedsters – Respondents who rushed through grid questions are identified by comparing grid completion times to the norm
• Trap Questions – Survey questions with obvious answers can determine whether a panelist is fully engaged with the survey
• Respondent Satisfaction – Feedback from panelists is gathered and assessed to help determine the quality of the survey

Eliminating Duplicate Respondents

When more than one source is used for a survey, Lightspeed Research’s quality solution includes a tool to identify and eliminate duplication within and across sample sources. Utilizing digital fingerprinting, each computer is tagged with a unique ID generated by a proprietary algorithm. We also continue to test innovative tools offered by technology companies to supplement our capabilities as needed. Currently in the testing phase, this solution is scheduled for global rollout by the end of 2008.

These are just a few components of the solutions Lightspeed executes to ensure data quality. Our practices regarding recruiting, privacy, panel management, sample management, and survey execution further exemplify our commitment. By continuously enhancing our technology and practices, we ensure the highest level of online data quality available in the industry.

As always, additional information about our practices is available from your account representative.

If you’re thinking about bringing online a study formerly done via telephone or in a mall – large or small, whole or in part – here are the key things to keep in mind, along with information on how Lightspeed Research can help you prepare for the move.

First, consider whether your migration includes a “modal shift”, such as changing from an interviewer-administered surveying mode (telephone or mall interviewing) to online surveys. Most online migration projects do involve modal shifts, and the change to self-administered surveying will have implications for every aspect of your study – from sampling to design to analysis.

Survey Design Considerations

Perhaps the most profound effect will be in how questions can be asked. Questions can produce different response patterns when asked by an interviewer versus when read by a respondent on a screen.

For example, a telephone interviewer might ask the question, “What is your marital status?” and simply code the spoken response into one of the available categories: single, married, separated, divorced, or widowed – without reading each of the categories to the respondents. Respondents might simply report their status as either “single” or “married,” as might be expected in response to a casual inquiry from a stranger. As a result, the data would tend to end up predominantly in one of those two categories, and rarely in one of the others, unless the interviewer was instructed to probe further.

However, if identical wording was used in an online survey, the respondents would be exposed to each of the five categories and would likely provide a more specific response, distributing data across the categories. (Dillman, 2007)

Simply put, telephone survey questions are usually written to be read aloud by a trained interviewer, while self-administered surveys, such as those for the web, are meant to be read directly by the respondent. Self-administered survey questions must be straightforward and self-contained, without lengthy instructions. It’s important to think about these distinctions when repurposing survey questions for online surveys.

Another effect to consider is survey length. A persuasive telephone interviewer can keep a respondent engaged longer than is otherwise possible. In contrast, a self-administered questionnaire must be engaging by itself, and shorter overall, to optimize both completion rates and data quality.

If the survey is a long-standing one, there may be scales and questions associated with baseline norms or other trend data, so changing the questions may be difficult or impossible. In this case, consider “unimode construction” options – a methodology to maintain design consistency across modes. In this method, the basic structure of the question and the scale or response categories usually can be retained, with minor changes.

The modal shift to online provides an opportunity to obtain the rich data and ideation possible through open-ended questions – an exception to the general “keep it short” online rule. In many interviewer-administered studies, open-ended questions were routinely cut from questionnaires, particularly those done via telephone where every minute of interviewer time is added to costs.

In telephone studies, the quality of open-ended data is hit-or-miss depending on the patience and skill of the interviewer. Whereas in online studies, many clients are surprised to find open-ended questions can yield voluminous results – often two or three times that of offline. We can help you utilize text-mining technology and other coding software to make the most of verbatim responses.

Naturally, online surveys should not over-use qualitative methods. These should be used only as a supplemental form of data, in consultation with your service representative.

“Split Administration” and other Sampling Suggestions

What if the survey you’re migrating is very long, but you need to retain all of the questions? “Split administration” techniques reduce the burden on any one group of respondents by splitting the overall sample into several matched sample groups. The questionnaire is similarly split – into modules containing sections of the lengthy offline survey.

A few modules are administered to all sample groups, but most are administered only to a single group. Each sample group, as well as the overall sample size, is sufficiently large to meet the objectives for the intended use of the results. Without sacrificing questions, you’ll reduce the respondent burden and improve data quality.

Lightspeed Research has a series of parallel tests and demonstrations on the split administration method currently underway. Please contact us for more information on this technique.

Another important sampling consideration involves your specifications. Are they appropriate for the new sampling frame? For example, you may have needed to keep your study relatively small when done in a mall environment, reducing costs by limiting the number of interviews. But given the cost and time efficiencies of online, and the much wider access to respondents, you should be able to greatly expand your sample specifications – along with your ability to meaningfully analyze the resulting survey data with a larger sample size.

Not all studies need to be moved online all at once, and it may be better to move only part of the study. For example, you may want to consider migrating the sub-sample of a study dedicated to mobile-phone-only respondents – obviously unavailable through traditional telephone sampling frames but readily available through our online or mobile phone panels. This would be an appropriately “blended” methodology.

Some Data Shifting May Occur

After migrating a study online, having taken into account the modal shift in other respects, any data shifts you may observe in the initial wave may be attributable to the interviewer’s absence. We often forget that using interviewers raises the potential for “social desirability” bias, or the tendency of respondents to give answers they think the interviewer wants to hear – instead of what they really think.

For more consultation about migrating your studies online, please contact your Lightspeed Research representative.

REFERENCES

Curtin, Richard, Stanley Presser, and Eleanor Singer. 2005. “Changes in Telephone Survey Nonresponse over the Past Quarter Century.” Public Opinion Quarterly 69:87-98.

Dillman, Donald A. 2007. Mail and Internet Surveys: The Tailored Design Method. Wiley.

Dillman, DA.;Sangster, RL.;Tarnai, J.; Rockwood, T. (1996). Understanding differences in people’s answers to telephone and mail surveys. In: Braverman MT, Slater JK. , editors. Current issues in survey research, New directions for program evaluation series. San Francisco: Jossey-Bass; pp. 45-62.

Groves, Robert M.; Peytcheva, Emilia. 2008. The Impact of Nonresponse Rates on Nonresponse Bias. Public Opinion Quarterly, 72:2, 167-189.

Groves, Robert M., Don A. Dillman, John L. Eltinge, and Roderick J. A. Little, eds. 2002. Survey Nonresponse. New York: Wiley.

Oldendick, Robert W., and Michael W. Link. 1994. “The Answering Machine Generation: Who Are They and What Problem Do They Pose for Survey Research?” Public Opinion Quarterly 58: 264-73.

Olsen, Kristen, 2006. Survey Participation, Nonresponse Bias, Measurement Error Bias, and Total Bias. Public Opinion Quarterly, 70:5, 737-758.

Why Make the Move Now?

Since respondents have become increasingly difficult to reach through other modalities, many marketers are considering moving their relatively small proportion of offline studies to online. For example, in 1975, the telephone sampling frame was robust enough to deliver about 70% of the US population – in those pre-cell phone, pre-answering machine days. By 1995, combined noncoverage/nonresponse of the telephone sample had shifted to about 62% (only 38% were available to interview). Today, combined noncoverage/nonresponse of telephone has grown to about 85%, while combined noncoverage/nonresponse for online stands at about 60%.

Rich Research History

In 1937, TNS – then known as NFO – established the first 1,000 member survey-by-mail panel. TNS conducted their first face-to-face interviews in 1962, their first phone interviews in 1970, and in 1996 became an established provider of online research. TNS has over half a century of honing and developing panel research skills.

“TNS has a rich heritage of working in research – and particularly in research panels,” said Frank Kelly, Lightspeed Research’s Global Panel Director. “Panels are very different than other research methodologies. You have to understand how to recruit, care for, develop, maintain, and incentivize panels.”

Continued Focus on Data Quality

TNS’s lengthy history of proven panel management techniques complements Lightspeed Research’s dedication to panel management – particularly our investment in our Data Quality Program, which includes Lightspeed RealRespondents, Lightspeed RealResults, and Machine ID. “I believe there are a lot of questions in the industry and a lot of interest in just what panel and research companies are doing to ensure that the panel research is of the highest quality,” said Kelly.

“During the integration process, it was quickly apparent that Lightspeed Research and TNS 6thdimension shared a belief that data quality was an essential element of our business,” said Lightspeed Research President and Global CEO David Day. “We have a wonderful opportunity to merge the best of what both businesses have to offer – specifically in regards to panel management and data quality. By uniting the best-of-breed capabilities of each organization – along with the sheer size of our global panels – we are in a position to be the industry leader.”

Unparalleled Global Reach

Along with establishing a combined organizational structure and establishing common work processes, an important early priority is the consolidation of our production platforms. The integration of major market panels will take place in 2010, but clients have access to our unparalleled global sample today.

The merged businesses will establish a best-in-class Kantar-standard approach to sample management and selection, panel recruiting and incentive programs, and panel quality methodologies.

Please know that the merger will not interrupt our services in any way. We will continue to provide updates as the business moves forward. If you have any questions, please do not hesitate to Contact Us.