21.06.2014 Views

Questionnaire Dwelling Unit-Level and Person Pair-Level Sampling ...

Questionnaire Dwelling Unit-Level and Person Pair-Level Sampling ...

Questionnaire Dwelling Unit-Level and Person Pair-Level Sampling ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

different ways depending upon the age group pair. For age group pairs where the only pair<br />

relationships of interest involved were child-parent pairs, two classes were required: both<br />

respondents never married <strong>and</strong> one respondent never married. Three classes were required for<br />

age group pairs where the only pair relationships of interest involved were spouse-spouse pairs:<br />

both respondents not currently married, one respondent not currently married, <strong>and</strong> both<br />

respondents currently married. Finally, six classes were attempted if both the spouse-spouse <strong>and</strong><br />

child-parent pair relationships were possible: (1) both respondents never married; (2) one<br />

respondent never married, the other formerly married (widowed or divorced); (3) one respondent<br />

never married, the other currently married; (4) both respondents formerly married (widowed or<br />

divorced); (5) one respondent formerly married, the other currently married; <strong>and</strong> (6) both<br />

respondents currently married. It should be noted that not all of these classes would need donor<br />

pairs if no recipient pairs were within the class. It also should be noted that marital status could<br />

not have been considered a logical constraint where spouse-spouse pairs were involved, since<br />

many live-in partners (who were considered spouse-spouse pairs) answered the marital status<br />

question as "never married."<br />

Gender makeup of pair constraint. For donors who formed a spouse-spouse pair, the vast<br />

majority were male-female. Hence, in those cases where a spouse-spouse pair was possible, the<br />

gender likeness constraint required that the donor pair <strong>and</strong> recipient pair be either both of the<br />

same gender or both of a different gender. This meant that the likelihood of same-sex spousespouse<br />

pair relationships were equally likely (more or less, depending upon the model) among<br />

donors <strong>and</strong> recipients.<br />

Age constraint on 15- to 17-year-old pair members. For the 15-to-17 age group, the<br />

likelihood of being in a spouse-spouse relationship was very small. Nevertheless, the likelihood<br />

that a 17-year-old was married was considerably greater than the likelihood for a 15-year-old.<br />

Hence, for the age group pairs where at least one pair member was between 15 <strong>and</strong> 17, the<br />

younger pair member of both the donor pair <strong>and</strong> recipient pair had to be of the exact same age.<br />

Number of children constraints. In Section 6.2.3.2, a covariate was defined for the<br />

number of children in the household younger than 12, AGE011, <strong>and</strong> one was defined for children<br />

in the household between 12 <strong>and</strong> 17, AGE1217. If there was disagreement between pair<br />

members on the values of these covariates, the pair member with information agreeing with the<br />

screener was used if possible. For the imputation of spouse-spouse relationships with <strong>and</strong><br />

without children, these covariates were used to restrict donor pairs, where AGE011 was used for<br />

potential parents younger than 18, <strong>and</strong> AGE011+AGE1217 was used for potential parents 18 or<br />

older. If the recipient pair had no children according to the relevant covariate or covariates,<br />

donor pairs also did not have children. If the recipient pair had children, the same was true for<br />

the donor pair. In almost all cases, when there was disagreement between pair members<br />

regarding whether the pair had children in the household or not, the imputation used information<br />

that was closer to the screener. 19<br />

The likeness constraints were loosened in the following order (where applicable): (1) for<br />

the age group pairs where six marital status classes were used, collapse to two classes (the same<br />

19 This will not always be true, because it is not always possible that the screener can be used to determine<br />

the value for AGE011 <strong>and</strong> AGE1217 when the pair members' information disagrees.<br />

43

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!