How Dating-Pool Filters Combine: Independence, Correlation, and Compounding

Suppose 50% of a reference population passes an age filter and 40% passes an income filter. Adding the percentages would make no sense because the calculator needs the overlap. If the filters were independent, the estimated overlap would be 0.50 × 0.40, or 20% of the original population.

Each additional requirement narrows the overlap again. This is why several moderate-looking filters can produce a small combined percentage even when no single filter is especially rare.

Independence means that knowing whether someone passes one filter gives no information about whether they pass another. Real demographic traits often violate that condition. Income distributions change with age. Marital-status patterns change with age. Some health measurements vary with age and sex.

Blindly multiplying national averages ignores those relationships. Depending on the selected combination, that can exaggerate or understate the overlap. There is no universal direction of error.

Version 1.0 divides the selected age range into bands. Inside each included band it applies the corresponding income, unmarried, and obesity assumptions, then weights and sums the band results. This means age is used as a conditioning variable rather than multiplied as one flat national percentage.

Height, income, marital status, obesity, and race or ethnicity are still treated as conditionally independent within each age band. That remaining assumption is why the output is labeled a model estimate rather than an official population statistic.

American Community Survey Public Use Microdata Sample records make it possible to count weighted people who jointly meet compatible variables such as age, sex, income, marital status, and race or ethnicity. Height and measured BMI come from different CDC survey systems, so integrating every filter still requires careful modeling rather than a single universal table.

A future model should publish variable definitions, survey weights, exclusions, transformations, and reproducible output tables. It should also show how the new joint estimate differs from the current age-conditioned approximation.

After calculating a scenario, loosen one filter and calculate again. The change reveals which measurable preference has the greatest effect in the current model. That is more useful than interpreting a rarity label as a judgment about you or your standards.

The result still says nothing about values, chemistry, mutual interest, location, or relationship quality. It describes an estimated demographic overlap under published assumptions.