Pluriel uses what we know to be the best method for making sure the data we collect is representative of the population. Our proprietary weighting algorithm relies on a combination of marginal and joint distributions to get you the most accurate data every time—we don’t ignore the fact that certain demographic factors are interconnected.

In the world of market research, the devil is in the details. One such detail that significantly affects the quality of survey insights is how research firms try to ensure the data they’ve collected is representative of the population—a process known as
“survey weighting.”

As we’ve written about before, we’ve developed online-first sampling methodologies to collect the most representative and unbiased samples on the market today.

But if there are still “imbalances” in our sample—that is, if certain demographic groups are over- or under-represented—we apply a proprietary algorithm to generate survey weights so the data is representative of the population.

What sets us apart is how we do that.

The common practice: “Raking” based on marginal distributions

The common practice for most research firms is to use census data to calculate the percentage of the population—or the marginal distribution—of different demographic groups. They then take these percentages and employ a procedure called “raking” (also known as iterative proportional fitting) to generate survey weights for the data they’ve collected. This can go a decent way in making the data more representative.

A problem might arise, however, when the procedure assumes that subgroups of the population (say, for example, women, immigrants, or retirees) have the same marginal distribution of a characteristic (for example, % who have a Bachelor’s degree) as other subgroups. Does it make sense for a weighting algorithm to assume that the same percentage of people 18-25, many of whom are still in school, have a BA as people over the age of 25?

At Pluriel, we believe that survey weights should not neglect the interconnectedness of demographic factors.

Pluriel takes a more refined, and more accurate, approach    

When it comes to survey weighting, not all techniques are created equal. Drawing on our expertise in quantitative public opinion research, Pluriel employs a more refined approach.

We use a proprietary algorithm that accounts not just for the marginal, but also for the joint distribution of certain demographic data. While that sounds technical, it simply means that our weighting algorithm accounts for the fact that certain demographic factors are interconnected and shouldn’t be thought of in isolation if you want data analysis that is truly representative of the population.

With this approach, we can capture interdependency and complexity, offering insights that are more reflective of reality.

Cost and speed benefits

While accounting for joint distribution may sound computationally expensive, our proprietary algorithm is optimized for efficiency. The result is a system that provides superior data without causing delays or increasing project costs the way “custom” methodological work might at other firms.

In a field where accuracy is paramount, Pluriel’s unique approach to survey weighting ensures we deliver nothing less than exceptional insights that are accurate, cost-effective, and fast. By going beyond the traditional approach to survey weighting, we offer our clients a level of granularity and realism that is at the forefront of the market research industry.