Doc: Household energy insecurity data dictionary
Status: Source-complete selected comparisons
Owner: Rene
Last verified: 2026-11-15
Depends on: EIA RECS 2024 HC 11.1 and analysis.py

group is the source classification; label is the source subgroup. These classifications overlap with one another. Do not sum across classifications. measure is the hardship column: any_insecurity(C), food_medicine_tradeoff(D), unhealthy_temperature(E), disconnect_or_delivery_notice(F), unable_use_heating(G), unable_use_cooling(H). all_households(B) is the same-row denominator.

households_millions is the outcome count at producer precision 0.01 million. denominator_millions is the matching group household population. point_estimate_percent=100*outcome/denominator. Counts refer to primary occupied homes, not people or utility accounts. source_cell and denominator_cell identify the original worksheet cells. raw_workbook_value preserves OOXML numerical representation; source-rounded values, not binary serialization artifacts, inform the calculations.

numerator_rse_percent and denominator_rse_percent come from the same-row rseworksheet. P means a poststratification control total, not missing or zero error. Q means withheld because RSE>50% or sample support<10. The selected 49 rows have complete numeric outcome counts. The analytical selection excludes the bill-payment classification containing a suppressed category. No missing value is replaced by 0.

count_standard_error_millions=count*RSE/100. Ratio confidence intervals are not calculated without joint covariance. difference_percentage_points=leftpercent-rightpercent. point_estimate_ratio=leftpercent/rightpercent. No significance test or causal claim follows.

source-rounded-partition-audit retains small sum residuals against the published national row. All source estimates remain unchanged. Units are nominal household counts and descriptive percentages, with no expenditure, emissions or current-winter interpretation.
