The survey research community has relied on the statistical machinery developed nearly a century ago to produce measurable inferences when only a relatively small sample of the target population can be observed . This groundbreaking development, however, was made possible under a set of conditions that include access to a complete sampling frame where units could be selected with known probabilities, and near-perfect response rates. Arguably, these requirements are becoming exceedingly difficult to satisfy for most surveys.
While we remain aware of the truism that organic representation is exclusive to probability-based samples, the fact is, such samples are often cost-prohibitive and infeasible. Researchers need more pragmatic alternatives that can accommodate their analytical needs in the 21st century digital age.
Many survey and market research applications have turned to online solutions, replacing traditional sampling frames with online panels for survey design and administration. While there are a handful of online panels that rely on quasi probability-based sampling techniques for recruitment purposes, the vast majority of online panels are compiled through various methods of intercept or targeted recruitments. With these so-called "opt-in panels," sample representation is never organic and at best can be approximated through sample balancing. Moreover, most opt-in panels are increasingly tormented with fraudulent respondents that are either automated (bots) or impersonated (click farms).
Dealing with these challenges directly with online panel suppliers is akin to wandering in a Turkish bazaar without necessary haggling skills. Here is where Marketing Systems Group comes to the rescue. Our decades of experience with sampling and access to dozens of panel providers allows us to secure the most representative samples that opt-in panels can deliver. PreVision is our latest umbrella of services for online surveys.
Our survey research experts provide a complete sequence of services, starting with securing samples from probability and nonprobability panels to providing weighting and calibrations solutions uniquely developed for samples with representational challenges.