You asked how the U. S. Census Bureau calculated the number of same-sex, unmarried partner households in the 2000 Census. This report is a follow-up to the OLR Backgrounder "Same-Sex Partners in Connecticut (2002-R-0834) which cites Census 2000 data counting 7,386 same-sex, unmarried partner households in Connecticut. " (This number represents less than . 5% of all households in the state. )

The information below comes from two U. S. Census Bureau reports: "Households and Families: 2000" and "Technical Note on Same-Sex Unmarried Partner Data From the 1990 and 2000 Censuses. " These documents are on-line at http: //www. census. gov/population/www/cen2000/samesex. html and http: //www. census. gov/prod/2001pubs/c2kbr01-8. pdf respectively. We enclose copies.

The Census Bureau's figures were primarily derived from what it refers to as the "relationship item," which asks how each member of the household is related to the householder. (The "householder" is the person in whose name the housing unit is owned, being bought, or rented and is referred to throughout the questionnaire as "Person 1. ") Figure 1, below, reproduces the question from the census short-form questionnaire. All households were asked this question.

The Census Bureau got the data it used to calculate the number of unmarried partner households and the partners' genders by correlating questionnaires where the "Unmarried partner" box was checked with responses to other questions asking the sex of the householder and the partner.

Analysts made one more adjustment when the "husband/wife" relationship box was checked but other information in the questionnaire indicated that the person designated as such was the same sex as the householder. In that case, the "husband/wife" response was, in most cases, changed to an "unmarried partner" response.

The bureau calls this change "data allocation," which is a standard statistical practice that most data collection agencies use when survey respondents leave out information or give answers to survey questions which are internally inconsistent. The allocation formula for same-sex partners described above is different than the method used in the 1990 Census, and for this reason the Census Bureau cautions that data from the two censuses are not comparable.

The Technical Note referred to above describes the allocation formula used for 1990 Census data and the basis for the agency's conclusion that the new methodology presents a more accurate picture of the number of same-sex households.

SP-L-ts