The Survey of Consumer Finances is the gold standard for researchers interested in the finances of American families, yet it is a complex dataset. One methodological choice regards the use of the respondent’s demographics versus those of the reference person. When the respondent and reference person are not the same (40.4% of couples in 2022), many researchers use the demographics of both individuals (e.g., the race of the respondent and the education of the reference person). This mixed usage can introduce misclassification and androcentric bias, and it can shift results significantly (e.g., net worth estimates by education vary by 5.0%–35.2%). We argue that the optimal solution is to always use the demographics of the respondent. This is especially critical when the analyses include race and ethnicity variables (which from 1989 to 2019 were only collected from the respondent), or when they include sex and marital status because the reference person is almost always male in coupled households.
Read full abstract