SOI Tax Stats - Split-Interest Trust Study Data Sources and Limitations


Return to the main Split-Interest Trust Study Metadata page

This page contains information about data sources and limitations for SOI's Split-Interest Trust Study.

Please visit Split-Interest Trust Statistics to access articles and tables from the study.

The data presented here were collected from a sample of Forms 5227, Split-Interest Trust Information Returns, selected during a specific filing year.  A filing year includes returns received by the IRS for processing between January 1 and December 31 of a given year and primarily comprises returns for the tax year immediately prior.  However, it may include late-filed returns for numerous other tax years.  For Filing Year 2011, approximately 97.6 percent of returns included in the sample were for Tax Year 2010, while Tax Year 2009 returns comprised 1.6 percent of the sampled returns.  Partial-year returns, for either initial or final reporting periods, were included in the SOI sample.  All returns included in the sample were computer-designated at the IRS Ogden Submission Processing Center after posting to the IRS Master File.

For Filing Year 2011, a sample of 11,487 returns was drawn from an estimated population of 118,278.  This sample count includes returns that were selected for the sample but later rejected.  Returns were rejected if they were not one of the four types of trusts included in the study, or if no money amounts were reported.  The sample was stratified by the type of the trust (charitable remainder annuity trust, charitable remainder unitrust, charitable lead trust, or pooled income fund) and the reported end-of-year book value of total assets.  This figurePDF details the sampling strata and rates.  The data entry process revealed some trusts with incorrect type classifications.  In these cases, the trust information was corrected to reflect the correct type.  However, the weights used for these trusts were based on the original sample selection classification.  The magnitude of sampling error for selected items, measured by coefficients of variation, is shown in this figurePDF.

All samples were designed to provide reliable estimates of financial activity.  All data were collected from original returns as they were filed.  All edited returns were subjected to comprehensive testing and data verification procedures to ensure the highest quality of data.  Changes that were made to the return after filing, either by the taxpayer (on an amended return), or during IRS processing, generally were not incorporated.  A complete discussion of the reliability of estimates based on samples, methods for evaluating the magnitude of both sampling and nonsampling error, and the precision of the sample estimates can be found in the Appendix in each issue of the SOI Bulletin or online herePDF