Many of us in marketing research have been deploying web surveys for over ten years, and web surveys are, by far, the dominant mode of data collection in our industry nowadays.  But our techniques and methods are an amalgam of practices adapted from other data collection modes, learned in part through trial and error, and taught to others through channels more akin to oral traditions.  So it is helpful when our academic colleagues manage to document and codify the art and science of what we do.

A new Handbook of Web Surveys does just that.  Among other things, it reminds us that whatever the survey mode—mail surveys, phone surveys, in-person surveys, mobile surveys, or online surveys—the key to rigorous research is bringing together theory, logic, mathematics, and practicality.

The biggest challenges for web surveys are that (1) not all people have Internet access, introducing the potential for coverage bias, and (2) most web surveys rely on volunteer respondents, introducing the possibility for self-selection bias and non-response bias.

There are ways of correcting for these biases, primarily through careful adjustment of the data through weighting.  A highlight of this handbook is that it reviews the complex ways in which weighting can and should be done for web surveys, including the use of regression estimates, raking (also known as rim weighting) and propensity scores.  Indeed, as one recent reviewer of the handbook noted:

The chapter on sampling and the later chapters on self-selection (chapter 9), weighting adjustment (chapter 10) and response propensities are central to statistical analysis of Web survey data, and the concepts treated in these chapters are at the core of debates on the scientific use of Web surveys. The authors should be complemented on the accessible way they introduce and describe these topics.

If you do any kind of survey work, you need to understand these issues.  You need to understand them at a conceptual level, and you need guidelines on how to implement them at a practical level.  This handbook will help.  Versta Research can also help.  We have expertise in complex quantitative methods, including the use web surveys for scientific and market research as well as for public opinion polling.  Please feel free to give us a call.

–Joe Hopper, Ph.D.

]]> http://www.verstaresearch.com/blog/just-published-handbook-of-web-surveys/feed/ 0 http://www.verstaresearch.com/blog/your-margin-of-error-is-probably-wrong/ http://www.verstaresearch.com/blog/your-margin-of-error-is-probably-wrong/#comments Thu, 29 Dec 2011 15:32:13 +0000 Joe Hopper http://www.verstaresearch.com/blog/?p=1847

Even if you are not involved in political polling, it is worth paying attention to the methods and best practices of political pollsters.  One reason is that few other areas of research offer a way to completely validate one’s methods.  Pollsters are using sampling and survey methods to predict the behaviors of a much larger population.  Then in just one day that population behaves, we get a near-perfect count of exactly how they behaved, and we know whether the methods worked.

Several industry colleagues have recently been debating the merits of calculating and reporting “margins of error” in political polling, and pointed us to some surprising data from The New York Times:

[The New York Times has compiled] a database consisting of thousands of primary and caucus polls dating back to the 1970s.  Each poll contains numbers for several candidates, so there are a total of about 17,000 observations. How often does a candidate’s actual vote total fall within the theoretical margin of error?  The answer is, not very often. In theory, a candidate’s actual vote total should fall outside the margin of error only 5 percent of the time [given that political polls report margins of error using a 95% confidence interval]. In reality, the candidate’s vote total was outside the margin of error 65 percent of the time! Part of this is because the database includes some polls conducted months before the actual voting took place. But even if you restrict the analysis to polls conducted within the final week of the campaign, about 40 percent of the vote totals fell outside the margin of error — eight times more often than is supposed to happen if you could take the margin of error at face value.

This does not mean that the polls were wrong, predicting wins for losing candidates and vice versa.  Rather, it means that the estimates were not as precise as the stated margins of error would have a reader believe.

The problem is that “margins of error” are based on a statistical theories that almost never line up with the messy reality of our world.  Margins of error make a number of assumptions which are rarely true in practice, including:

• Respondents are selected through simple random sampling

• All those sampled participate in the survey

• Sampling error is the only source of survey error

Indeed, Versta Research usually recommends to clients who publish survey research that they not report margins of error because the concept (and the calculations) are seriously misleading and flawed.

Calculating margins of error and looking at statistical significance should be used not because they give accurate or “scientific” predictions, but because they provide useful summary measures of how much variability there is in the data given the sample size and other critical factors that can affect one’s estimates.  At Versta Research, this helps us better interpret data and better assess what matters.  That, in turn, allows us to tell a story with the data that does not overreach or misrepresent what is going on.

–Joe Hopper, Ph.D.

But that doesn’t mean “anything goes” when it comes to fielding market research surveys and public opinion polls through online panels.  Many panels are poorly managed and overused, and some have high proportions of fraudulent respondents.  While conducting good research through online panels is possible, it requires a great deal of effort and oversight from smart people who know what they are doing.

I was reminded of this recently as we worked with a newer panel provider that recruits respondents through not-for-profit organizations. When respondents complete surveys, their sponsoring NFP organizations get donations.  Response rates are high because members are collectively motivated to participate.  But depending on your study, panelists may not represent the population you want to understand.  If your survey is geographically targeted at the local level, for example, chances are high that respondents are clustered into a limited number of social groups, because that is exactly how they were recruited.

It was a reminder, too, that while sampling through social media and social networking can leverage the amazing power of online social networks, it is critical to understand the effect of networks and clusters on sampling.  And it is critical to incorporate that understanding into your statistical analyses.

Before you commit to any type of online study that relies on sample from a panel, we recommend ongoing due diligence about how the panels are constructed and how respondents are deployed.  At the very least:

1.  Find out how respondents are recruited onto the panel.  As in the example above, different recruitment methods may affect your research design and analysis plan, and for some studies you may need to find an alternative.

2.  Find out how panelists are selected for your particular survey.  You need to ensure that survey respondents are broadly representative of the population of interest.  Quick convenience samples or fast polls using routers can mess that up, so be sure to understand the protocols.

3.  Ask for validation data.  Studies show that panel research can replicate the most rigorous methods used by agencies like the Census Bureau and the CDC.  Ask panel providers for evidence that they have benchmarked their techniques for sampling against data provided by these agencies.

For additional questions you might ask (23 more questions, to be exact) we recommend ESOMAR’s 26 Questions to Help Research Buyers of Online Samples.  Or, give us a call at Versta Research and we will  be happy to guide you through the process.

–Joe Hopper, Ph.D.

Is it true that the Associated Press refuses to carry stories from online surveys?  Yes, as odd as that seems nowadays.  But news media face a difficult problem given how easy it is to conduct biased public opinion polling, especially now with online panels and social networks.  So some news organizations like the Associated Press (AP), The New York Times, and ABC News have developed guidelines that specify for a survey or public opinion poll to be valid and reliable, it must be conducted by telephone.

It is worth understanding their rationale even if you do not pitch research stories to the media because the guidelines provide a fascinating glimpse into current methodological debates about probability sampling, inferential statistics, and the rapidly changing world of online data collection and analysis. If your organization does use research for public relations and marketing, then a deeper understanding will help you offer recommendations to your organization about conducting PR research.

In our Fall 2011 Newsletter, we outline media guidelines for reporting on survey research.  Whether your goal is getting research into the board room to influence top decision makers, or in front of the public to promote your brand, you need to know the standards of rigor against which the research will be judged.

Our commitment at Versta is to advise you on the best research mode for your campaign strategy, and to conduct rigorous research that can withstand the highest levels of scrutiny.

–Joe Hopper, Ph.D.

Monty Hall in Let's Make A Deal

The hardest part of quantitative market research is not that it involves numbers, math, or even statistics, but that it involves complex problems in probability.

Over the past several years, psychologists have been documenting how difficult it is for us humans to solve even “simple” probability problems.  One fascinating example is a puzzle known as the Monty Hall dilemma based on the 1960’s game show Let’s Make A Deal.  Monty would offer his contestants three doors to choose from, one of which had a valuable prize behind it.  After the contestant chose, Monty would open one of the other two doors, deliberately choosing one that had no prize behind it.  Then he offered the contestant an option of staying with the original choice, or switching to the other unopened door.  Which should the contestant do?

The contestant should always switch.  The odds of winning are two-thirds if she switches, and one-third if she stays.  Most contestants, however, stay with their original choice, believing that the odds of winning are the same whether they stay or switch.  And it turns out that pigeons do a better job solving this puzzle than humans.  In an article published last year in the Journal of Comparative Psychology, researchers showed that if a similar game is played with pigeons, they start to catch on and consistently choose to switch, which maximizes their winnings.  Humans, however, do not.  Not only do we have a hard time grasping the true probabilities conceptually, but even if we play the game over and over, we ignore our experience and learning.

What does this have to do with market research?  Well, behind all the numbers, charts, and percentages that we present to our clients, most of our methods and analyses are based on probabilistic reasoning.  We calculate the probabilities that our sample statistics represent true population values.  We build conjoint or MaxDiff models based on probabilities of certain responses occurring even if we did not measure them directly.  We ask respondents to assess the probabilities of their own behavior (“How likely are you to buy?”) and use those to calculate estimates of market potential. We are dealing with layers upon layers of probabilities.

It is no wonder that market research reports can be so impenetrable and difficult to untangle.  Behind nearly every chart or table is a probability puzzle, and for most of us there is certainly nothing intuitive about probabilities.

And it is no wonder that many research firms do not even try to go beyond giving you charts, data, and tabulations.  But that’s where a firm like Versta Research comes in.  We solve two of the most difficult challenges facing research professionals:  (1) grasping the complex nature of probabilistic reasoning, which may befuddle even the most accomplished mathematicians, and (2) turning mounds (or crumbs) of data and probabilistic reasoning into an effective and compelling story that you can use and that your clients can understand.

When you need help with either or both of these challenges, call us at (312) 348-6089.

—Joe Hopper, Ph.D.

]]> http://www.verstaresearch.com/blog/pigeons-beat-people-on-probability-problems/feed/ 0 http://www.verstaresearch.com/blog/smartphones-matter-more-than-cell-phones/ http://www.verstaresearch.com/blog/smartphones-matter-more-than-cell-phones/#comments Fri, 19 Aug 2011 12:47:28 +0000 Joe Hopper http://www.verstaresearch.com/blog/?p=1543 The most recent government estimates of cell phone usage among U.S. households were released a few weeks back, and the pace at which landline usage is disappearing is astonishing.  Here are just some of the numbers:

• Thirty percent of U.S. households do not have a landline telephone
• An additional 16% have a landline telephone, but never or rarely use it to receive calls
• The percentage of households without landlines is increasing by about five to six percentage points each year
• Half of young adults under age 30 have no landline in their homes
• Half of adult renters have no landline in their homes
• Nearly four out of ten Hispanic adults have no landline in their homes

For anyone doing random sample surveys of consumers, the implications are huge.  Methodological purists insist that only phone-based surveys are rigorous, but clearly the biases of phone-based research can be severe, and no doubt they often are.

Putting aside the issue of research methods, however, we believe a more significant issue over the next few years will be the explosive growth in smartphone usage.  Although not captured in government surveys that track phone usage and availability, the Pew center estimates that one-third of all U.S. adults currently own a smartphone, and industry analysts predict that by the end of 2011 half of all cell phones in the U.S. will be smartphones.

Why is this is so critical for researchers?  Because more than any other technology since the widespread adoption of the Internet, smartphones are changing how consumers behave.  Consumers get up in the morning and check media apps before brushing their teeth.  Then while commuting to work or waiting for their computer to boot up, they purchase coupons for goods and services they might never have bought otherwise.  They map and track their locations, and expect product features and services that can anticipate what they need.

Smartphones are important for reasons far beyond narrow methodological concerns about how to execute surveys.  They are important because they are changing the very nature of what we research: how people behave and think, and what they buy, believe, want, and aspire to.  Indeed, just as we nearly always ask our research respondents to tell us their gender, age, and income, we now often ask whether they have smartphones.

Beyond anything else, market research is about understanding people (the verstehen in Versta), which means our focus at Versta is always on the “what, how, and why” of what matters in peoples’ lives in a rapidly changing world.  Smartphones matter in a more profound way than cell phones ever did.

—Joe Hopper, Ph.D.

]]> http://www.verstaresearch.com/blog/smartphones-matter-more-than-cell-phones/feed/ 0 http://www.verstaresearch.com/blog/cell-phones-may-double-your-survey-costs/ http://www.verstaresearch.com/blog/cell-phones-may-double-your-survey-costs/#comments Tue, 09 Aug 2011 14:13:02 +0000 Joe Hopper http://www.verstaresearch.com/blog/?p=1528

These days most researchers agree that if you want to do a random sample phone survey of the U.S. population, you ought to include cell phones.  More than one-quarter of the population do not have landline telephones at home.  Those who do have landline telephones are less likely than ever to answer them, and less likely than ever to participate in surveys.

But it is not easy to include cell phones.  The sampling protocols and the post-stratification weighting become more complicated.  You need to account for a higher probability of cell phone owners being in your sample, because most of them also have landlines.  You can’t use automated or predictive dialing to call cell phone numbers.  You can’t target geography as well, because area codes and exchanges have become mobile.  And people get mad at you if they have to pay for incoming calls, so you need to offer cash.

What’s the bottom line effect on costs for a survey that includes cell phones?  A recent study sponsored by the American Association for Public Opinion Research (AAPOR) documents the following:

The cost of fieldwork for cell phone interviews is double if you don’t screen out those who also have landlines, and more than double (2.6 times higher) if you do screen them out.

Of course if you are surveying the whole population, only a portion of your sampling and interviews will be cell-phone based.  Currently, we recommend that 20% to 40% of interviews be cell-based.  But there are additional professional costs to remember as well, such as purchasing, managing, merging, weighting, and analyzing different types and sources of sample, and training interviewers to work with different sources and types of respondents.

There are still plenty of phone surveys being done that do not include cell-phones, and for many types of studies landline-only surveys produce information that is good enough for what needs to be learned.  But it is getting increasingly difficult for these surveys to achieve true representation and surely their days numbered.

Feel free to give us a call if you need help figuring out the best approach for your research.  We can advise you on the most cost-effective, feasible, and rigorous approaches to getting the data, stories, and level of understanding you need.

—Joe Hopper, Ph.D.

What does Versta Research recommend when it comes to sample size?  Well, the answer always depends on the type of study, the objectives of the study, the audience to whom it will be presented, and of course budget. We can make legitimate statistical calculations based on any sample size, but there are dramatic trade-offs in precision and cost no matter what sample size is chosen.

Here, however, are some general guidelines:

If you have the budget for a large sample, don’t even consider going beyond a couple thousand, as you gain so little. Beyond a sample size of 2,000 (which gives you a margin of error of about ±2%) you would have to pull an additional 4,700 into your sample (for a total of 6,700) to gain just one more percentage point in precision.  The benefit of doing this will almost never exceed the cost of including that many more people in your sample.

The exception: If you need to understand segments or subgroups within your sample, choose your sample size based on the precision you need for those smaller groups, not the sample overall.

With a small sample there is a substantial gain in precision for every random person you add to the sample. The difference between a sample of 1,000 and 1,075 is relatively small, decreasing the maximum margin of error by just a tenth of a percentage point.  But the difference between a sample of 50 and 125 is dramatic, decreasing the maximum margin of error by more than five percentage points.  Another twenty five, fifty, or one hundred respondents can make an important difference when you are looking at small samples.

Most sample sizes will range from about 100 and 1,200 and within this range, for each additional 50 or 100 people you include in your sample, you gain a decent improvement in the precision of your estimates.  So the questions to consider are always (1) how much precision do you really need, and (2) what is the cost of including each additional person?

In academic journals you will find studies with sample sizes as small as thirty to fifty people.  Some of the healthcare research we do relies on samples sizes of fifty to one-hundred.  Most other research we do, including that for publication in media outlets, relies on samples ranging from 300 to 1,200.

Unfortunately there is no magic number for sample size, whether it be thirty, one hundred, three hundred, or one thousand.  But there is a magic phone number worth trying:  (312) 348-6089.  Versta Research has a great deal of experience choosing sample sizes and consulting with research, marketing, and communications teams on the key questions that need to be answered within constraints of time and budget.  As always, the magic is not in any number itself, but in the highly skilled way in which it is deployed and then turned from data into stories.

—Joe Hopper, Ph.D.

]]> http://www.verstaresearch.com/blog/an-interactive-graph-for-choosing-sample-size/feed/ 0 http://www.verstaresearch.com/blog/online-surveys-have-same-accuracy-as-phone/ http://www.verstaresearch.com/blog/online-surveys-have-same-accuracy-as-phone/#comments Thu, 26 May 2011 13:24:37 +0000 Joe Hopper http://www.verstaresearch.com/blog/?p=1379 A new study presented by two professors from Harvard University and the University of Massachusetts at Amherst was probably one of the liveliest and potentially disruptive presentations at least week’s annual meeting of the American Association of Public Opinion Research (AAPOR) in Phoenix.

Why?  Because their research challenges the beliefs of many AAPOR-ites who disregard most online research as being theoretically indefensible since it is not based on probability sampling.  The research presented was based on parallel surveys conducted last year, designed to allow careful comparison of three survey modes:  a “gold standard” phone survey that included both landline and cell phone RDD sampling, an opt-in Internet panel survey that used careful weighting techniques to ensure a comparably representative sample of the U.S. population, and a traditional paper-based mail survey.  The questionnaires for each mode were nearly identical and elicited data on variety of topics including several that could be validated against the most rigorous U.S. Census and NIH data.

Here is how the authors summarized their results:

Comparing the findings from the modes to each other and the validated benchmarks, we demonstrate that a carefully executed opt-in Internet panel produces estimates that are as accurate as a telephone survey and that the two modes differ little in their estimates of other political indicators and their correlates. (download full paper)

Their research adds to a growing body of research showing that rigorously done online surveys can be as useful and effective as rigorously done phone surveys.

True, most online surveys are not rigorously executed (but they can be), and the practical successes of online surveys have outpaced our understanding of the statistical, theoretical, and social underpinnings of why they work.  But it is both intellectually closed-minded and pragmatically foolish to dismiss all online surveys as “invalid” and “not newsworthy” as some of our colleagues in AAPOR do.

Our approach at Versta Research is rooted in our training and continuing engagement in academic research, but we are also rigorously focused on doing what works to get the urgent answers and compelling business stories that our clients need.  Online survey research is one crucial piece of that.

—Joe Hopper, Ph.D.

This shift in telephone usage is critically important to survey research, because telephone surveys have set a gold standard for rigorous research over the last two decades.  Given the trends, most survey researchers would agree that cell phone sample must now be included in the most rigorous research designs for an RDD survey.  But doing so introduces new difficulties in design, sampling, weighting, and cost, and our collective experiences are not yet sufficient for the industry to  agree on a set of “best practices” when it comes to including cell phones in survey research.

Our leading industry association, AAPOR (The American Association of Public Opinion Research) recently published a comprehensive update from the AAPOR Cell Phone Task Force that offers a good overview of the issues, complications, guidelines, and considerations every researcher should take into account when designing a telephone survey of the general population.  Briefly, it covers and highlights the following issues:

Coverage and Sampling: It is increasingly difficult to reach young men and minorities via landlines, so cell phone coverage is critical.  But using dual and overlapping sampling frames (both landline and cell phone) introduces new methodological complications in sampling.

Nonresponse: Response rates are lower among cell phones users, though response rates among landline users continue to decline as well.

Measurement: Data gathered via cell phone interviews is generally not of lower quality (despite concerns about audio quality, talking in public places, and distractions from multitasking).

Weighting: If dual and overlapping sampling frames are used (to include both landlines and cell phones) then data need to be statistically weighted to account for the complicated differences in probabilities that each person in the population will be recruited into the survey.

Legal and Ethical Issues: Including cell phones in surveying introduces a new set of legal, regulatory, and ethical considerations regarding auto-dialers, texting, caller ID, and do-not-call lists.

Operational Issues: Cell phone interviewing requires a unique set of protocols for recruiting, call-backs, and interviewing, all of which require special training for the people conducting the interviews.

Cost Issues: Including cell phones at least doubles the cost of a survey, and many times may triple or quadruple the cost.

Should you include cell phones in your survey?  It depends on who you are trying to reach and for what purposes.  When it comes to effective surveying, there are no absolutely right answers about cell phones vs. landlines, just as there are no absolutely right answers about online vs. phone surveys.  Moreover, the rapidly changing landscape of how people communicate means that the best answer today may be different in six months.

Need some help thinking through the options and figuring out the best approach for your research?  Feel free to give us a call.  We have decades of experience and are committed to helping our clients take full advantage of important changes and innovations in research.

—Joe Hopper, Ph.D.