Compare how weighting is calculated in standard reports and modern reports, and understand why your weighted results stay consistent when you move to modern reports.
Modern reports use an updated weighting engine with stronger statistical safeguards, including Random Iterative Method (RIM) weighting, stricter handling of missing data, weight capping, and higher-precision calculations. In practice, these improvements rarely change your final numbers. For the majority of reports, weighted results are effectively identical to standard reports. Minor differences in counts or bases are uncommon and are usually limited to edge cases, such as participants with missing demographic data or extreme weight outliers.
The following table summarizes how weighting is calculated in each reporting experience.
| Dimension | Standard reports | Modern reports |
|---|---|---|
| Weighting method | Cell weighting (interlocking). A single-pass calculation that applies one static multiplier to each combined demographic cell, such as each gender-and-age combination. |
Random Iterative Method (RIM) weighting, also known as Raking or Iterative Proportional Fitting. It balances each weighting variable independently, adjusting the weights one variable at a time and iterating until the sample matches all of your targets. For more information, see Weighting calculations in modern reports. |
| Missing data handling | Includes all completed live responses, even when a participant skipped a weighting question. | Includes completed live responses that also have a value for every weighting variable. Participants with missing data in a weighting field are excluded, so the weighting is based only on complete data. |
| Convergence | Uses broad convergence limits and runs a high number of iterations to reach a result. | Continues iterating until successive passes change the weights by less than a strict tolerance, producing a tightly converged result. |
| Outlier protection | No weight capping. A severely under-sampled subgroup can produce extreme weights that overinflate a small number of responses. | Caps extreme weight factors and redistributes the excess weight across the sample, limiting the influence of outliers. |
| Numerical precision | Passes raw values downstream, which can carry floating-point noise into later calculations. | Rounds all weight factors and bases to a consistent precision, removing floating-point noise from downstream calculations. |
| Base metrics | Calculates the effective base from raw, uncapped weight variances. |
Calculates the effective base (Kish's effective sample size) from cleaned, capped, and rescaled weights, giving a more reliable measure of your weighted sample size. For more information, see Interpreting weight scheme results in modern reports. |
When results can differ
Because modern reports apply stricter data handling and outlier protection, a small number of reports show slightly different counts or bases than the equivalent standard report. This is most likely when:
- Some participants are missing data for one or more weighting variables. These participants are excluded from weighting in modern reports but included in standard reports.
- A subgroup is severely under-sampled, which can produce extreme weights. Modern reports cap and redistribute these weights, while standard reports apply them without limits.
In both cases, the modern result reflects the additional safeguards designed to keep your weighted data representative and reliable.