Using the MX8 Research Platform, all your results in cross-tabs or insights are automatically weighted to make up for any skews in the sample. The weighting is set up to match the audience that you've set up to field against.

Weighting is always set up every time you specify an audience. For example, if you set up to survey our standard US Genpop audience, we defined the demographics of the country as follows:

In this example, we've specified that 30% of the audience should be aged 18-24, and this means that even if only 29% of the respondents who entered your survey are aged 18-24, we'll weight them up so that in any reporting, they match the 30% that you've specified. The same applies to all the other groups.

The weighting is run automatically when the survey closes. It includes all respondents who completed the study or were terminated as not qualifying for some secondary reason. This way, the results are always weighted to be representative of the audience.

The method is an IPFN (Iterative Proportional Fitting) algorithm. The algorithm adjusts the weights of each survey response iteratively. It starts by ensuring that the survey matches one target (like age distribution), then moves on to match another target (like region), and so on.

This process repeats multiple times, adjusting the weights in each step until the survey data closely matches all your targets.

After enough iterations, you have a set of weights for each survey response. These weights make your survey data more representative of the overall population, ensuring that your analysis and conclusions are more accurate.

In essence, IPFN helps you "correct" your survey data to reflect the real-world population better, making your findings more reliable.

You can see the distribution of your weights by downloading the raw data in Excel or SPSS format and creating your histogram there.

Boost audiences are typically provided to get more samples for an additional group without weighing it separately. To have your audience included with another audience for weighting purposes, you should specify a weighting audience when you set it up:

With this set, the respondents from this audience will be weighted as part of the primary audience, not as a separate sample.

The weighting will run automatically for any nested quotas you set up for your audience.

You can't turn off the weighting in the platform, as it might lead to inaccurate results. If your sample precisely reflects your target population, then the weights will all be 1.0, and the results will be the same as unweighted.

If you want to use unweighted results, you can download the raw data and run the analysis yourself, but remember the principle of caveat emptor—without proper weighting, your results will not accurately reflect the actual population, leading to misleading conclusions.

All the respondents who complete your survey or are terminated are included in the weighting.

Because the reports only include respondents who completed your survey, these will often not match the percentages in the audience, but they will ensure that your results are nationally representative.

Consider an example of a survey that 1000 subscribers of ESPN take. Surveying a genpop audience, you might have 500 men enter the experience and 500 women. You terminate 80% of women in women because they don't subscribe to ESPN, giving you a total of 700 respondents, 500 men and 200 women.

When we weigh, we also include the 300 women terminated from the survey, so we end up with even weighting for both men and women and report that 500 / 700, or 71.4% of ESPN subscribers, are women.

That's not consistent with the audience you specified upfront, but it is nationally representative, and that's what's important.