This note examines the task of summarizing statistics by percentile-defined groups in Survey of Consumer Finances (SCF) microdata. Percentile-based grouping is a commonplace analytical operation, but its use in SCF analysis is complicated by the data's complex design. This note reviews three methods for making such comparisons: a naive threshold approach, implicate-wise estimation using Rubin's Rules, or the Federal Reserve's official "stacking" method. The function scf::scf_pctile_sum() offers a structured and transparent means for comparisons by percentile-based grouping variables in SCF analysis. The note demonstrates the function by comparing top-decile mean net worth estimates across SCF survey years to Federal Reserve published estimates and to the Federal Reserve-provided nwcat grouping variable.

@techreport{cohen2026pctile,

author      = {Cohen, Joseph N.},

  title       = {Percentile-Based Grouping in {SCF} Analysis: Introducing the Percentile Cut Function},

  year        = {2026},

  month       = {5},

  institution = {CUNY Queens College, Publications and Research},

  type        = {Working Paper},

  url         = {https://academicworks.cuny.edu/qc_pubs/705}

}