Measurement matters: Non-random missing data and indicator divergence in energy poverty

Energy poverty is measured using a range of expenditure-based and experience-based indicators that differ in their data requirements and the dimensions of hardship they capture. We examined (i) the extent and social patterning of financial data missingness, (ii) how closely different indicators of energy poverty identified the same households, and (iii) the validity of a composite experience-based severity score. In a nationally representative online survey of 1,011 adults in Ireland, participants reported household income, energy expenditure, and six experience-based indicators. Expenditure-based energy poverty was defined as energy expenditure exceeding 10% of disposable household income. Experience-based energy poverty was defined primarily by inability to afford adequate warmth (IAAW), with five other experience-based indicators combined into a composite severity measure. More than four in ten participants lacked a point estimate for income, energy expenditure, or both, precluding calculation of the 10% indicator for these households. This missingness was not random. Lower educational attainment, lower occupational social grade, and unemployment were significantly associated with being unsure or unwilling to report income or expenditure. Among those with full data, agreement between the 10% and IAAW indicators was weak. The 10% measure identified twice as many households as the IAAW indicator and the households it identified were less socioeconomically disadvantaged. The composite severity score was sensitive to socioeconomic disadvantage and explained more variation in financial difficulty than IAAW alone. Financial non-response may systematically bias expenditure-based measures of energy poverty by underrepresenting disadvantaged households. Complementary indicators may be preferable for monitoring or policy targeting, rather than reliance on any single measure.