JMIR Publications

ABSTRACT

Background:

Background:

Web-based surveys dominate health data collection among young adults, yet validation studies rely on mandatory participation or in-person verification, conditions absent from real-world digital surveillance. Whether voluntary web-based surveys produce systematically different estimates than mandatory objective assessment is unknown.

Objective:

Objective:

We compared BMI from a voluntary, anonymous web-based survey with objectively measured BMI from a mandatory fitness assessment in the same university population.

Methods:

Methods:

We paired a voluntary web-based survey (n=7,465; Wenjuanxing platform) with the mandatory Chinese National Student Physical Fitness Standards assessment (n=14,166) at a Chinese engineering university. Under full anonymity, individual matching was infeasible. We constructed six gender-by-grade strata, computed stratum-level discrepancies, and used quantile mapping and counterfactual bounding to distinguish selection from reporting effects. Bootstrap 95% CIs quantified uncertainty.

Results:

Results:

Voluntary survey BMI exceeded mandatory assessment BMI in all six strata (+0.61 kg/m2 weighted mean). The discrepancy was driven by weight (+0.7 to +3.0 kg), not height (+0.5 to +1.0 cm). Bootstrap CIs crossed zero in the two largest strata. Self-reported obesity prevalence was 10.3% versus 8.2% measured. Treating the discrepancy as measurement error reduced obesity prevalence to 9.2%.

Conclusions:

Conclusions:

The pattern, weight-driven, concentrated in smaller strata, indistinguishable from zero in largest strata, is consistent with heavier individuals being more likely to respond to voluntary health surveys, not with systematic reporting error. The distinction between reporting bias and selection bias determines whether the remedy is better instructions or better sampling design. Clinical Trial: no