Start Date
26-6-2025 3:15 PM
End Date
26-6-2025 4:15 PM
Description
To ensure the accuracy and consistency of mass appraisals, BC Assessment employs various ratio study statistics, including the Median Assessment to Sales Ratio (Med ASR), Coefficient of Dispersion (COD), and Price Related Bias (PRB). These metrics compare sale prices to assessed values across jurisdictions, identifying those that do not meet the International Association of Assessing Officers (IAAO) quality standards. However, the reliability of these statistics diminishes as the number of sales (sample size) decreases, reducing their validity in jurisdictions with limited sales data. This presentation introduces a method to determine whether the sample size is adequate for the metrics to be statistically meaningful. We utilize bootstrapped confidence intervals to assess reliability: a wide confidence interval indicates low reliability and signals the need for a larger sample size. Unlike the traditional approach of setting a uniform minimum sample size, our method evaluates each jurisdiction individually. For example, a jurisdiction with uniform sale records (many strata sales) requires a smaller sample size compared to one with more heterogeneous sale records (numerous detached sales), making a single minimum sample size inappropriate for both. Our approach de-emphasizes ratio studies in jurisdictions with limited sales data, thereby enhancing the overall quality assessment process and ensuring that metrics are only relied upon when supported by sufficient and reliable data.
Recommended Citation
Kuramoto, Mitchell and Helisaz, Hamed, "Reliability of assessment quality measures: Using bootstrapped confidence intervals to verify adequate sample size" (2025). Mass Appraisal Valuation Symposium. 23.
https://researchexchange.iaao.org/mavs/mavs2025/sessions/23
Reliability of assessment quality measures: Using bootstrapped confidence intervals to verify adequate sample size
To ensure the accuracy and consistency of mass appraisals, BC Assessment employs various ratio study statistics, including the Median Assessment to Sales Ratio (Med ASR), Coefficient of Dispersion (COD), and Price Related Bias (PRB). These metrics compare sale prices to assessed values across jurisdictions, identifying those that do not meet the International Association of Assessing Officers (IAAO) quality standards. However, the reliability of these statistics diminishes as the number of sales (sample size) decreases, reducing their validity in jurisdictions with limited sales data. This presentation introduces a method to determine whether the sample size is adequate for the metrics to be statistically meaningful. We utilize bootstrapped confidence intervals to assess reliability: a wide confidence interval indicates low reliability and signals the need for a larger sample size. Unlike the traditional approach of setting a uniform minimum sample size, our method evaluates each jurisdiction individually. For example, a jurisdiction with uniform sale records (many strata sales) requires a smaller sample size compared to one with more heterogeneous sale records (numerous detached sales), making a single minimum sample size inappropriate for both. Our approach de-emphasizes ratio studies in jurisdictions with limited sales data, thereby enhancing the overall quality assessment process and ensuring that metrics are only relied upon when supported by sufficient and reliable data.