Stratified Sampling Framework for Household Travel Survey: A Monte Carlo Simulation Approach for Sample Size Optimization

A Micro-Analysis Zone (MAZ) for localized transport planning creates challenges in cost and logistics from large sample sizes, delaying household travel surveys. Thus, this study developed a stratified sampling framework with a lower sampling rate for MAZ based on household size as a demographic stratum. Using census data from a barangay in Davao City, various sampling rates from 20% to 1% were examined using Monte Carlo simulation with one million iterations across two stratification approaches: proportional and Neyman’s allocation. The simulations revealed that a sampling rate as low as 8% can estimate trip frequencies within 10% error (MAPE ≤ 10%) at around 95% probability. Nevertheless, there is a trade-off between these metrics, with the probability dropping to 68% to maintain an error within 5% (MAPE ≤ 5%) under both stratification approaches. Furthermore, trip generation modeling revealed that R-squared values for estimating the trip frequency have a 95% probability of being 0.7 or higher. Despite this, caution is warranted, as some secondary predictors lost statistical significance at lower sampling rates. Ultimately, this study recognizes the performance of lower sampling for a small urban barangay, relieving transport planners from challenges in conducting household travel surveys.