Civis web surveys: designed by you, powered by data science

June 7, 2016 Masahiko Aida

It’s no secret that it’s becoming harder to field surveys in a way that’s both accurate and cost effective. Fewer people have landline phones, and surveying cell phones comes with high costs and regulatory issues. Because of this, market researchers and pollsters are turning to online surveys. But online surveys don’t come without challenges: it’s hard to know where exactly your sample is coming from and whether you can trust it. We’re changing that. Today, Civis is enabling you to conduct your own web surveys using our powerful scientific methodology, developed by our data and survey scientists.

It’s important to keep in mind that not all web surveys are created equal. In a recent report, the Pew Research Center compared the accuracy of nine web panel surveys against traditional benchmarks, and their analysis shows that results vary quite a bit. The most accurate results came from YouGov, which uses a method of matched sampling and weighting using data from the census and other federal government sources. What this demonstrates is that it’s possible to reap the benefits of the web as a means of conducting surveys while maintaining a high standard of accuracy through rigorous application of sampling and weighting techniques. This is no surprise to us: Civis uses a similar approach in our web surveys.

At a glance:

  • We provide a way to do surveys over the web by randomly sampling from a third party anonymized list mapped onto a known sampling universe so that we can be certain that it’s methodologically sound, yet also cost effective. Essentially, we are applying our list-based sampling approach (most recently highlighted in WIRED) and bringing it to web panels.

  • We leverage the power of big data to extrapolate from an online sample to the full population by stratifying and weighting using modeled variables. Typical web panels only using weighting to ensure that responses are representative according to the demographic characteristics of the population of interest. With modeled scores, we can go further, making sure the sampled respondents are representative according to a tailored set of actual attitudes and attributes most closely related to the questions you want answered.

As we work to advance the state of the art in survey research, we’ll also continue to work with the best tried and true methods. That means using phones still makes sense in a lot of cases and we continue to rely on them along with web surveys for our research, but now you can use that same methodology to conduct your own web surveys.

The post Civis web surveys: designed by you, powered by data science appeared first on Civis Analytics.

Previous Article
Using Sumo Logic for authentication auditing
Using Sumo Logic for authentication auditing

Behind the scenes at Civis, we take protecting client data and privacy very seriously. We invest substantia...

Next Article
Unexpected insights from repurposed data
Unexpected insights from repurposed data

At Civis, we’re always looking for new ways to use existing data. Even though I primarily spend my time wor...