DATA WERE TAKEN FROM THE NATIONAL LONGITUDINAL SURVEY OF WORK EXPERIENCE IN YOUNG WOMEN (MOTT ET AL., 1977). IN THIS PANEL STUDY, 5,159 WOMEN AGED 14 TO 24 IN 1968 WERE INTERVIEWED EACH YEAR FROM 1968 TO 1973 IN PERSON AND BY TELEPHONE IN 1975. THE SAMPLE WAS REPRESENTATIVE OF THE NATIONAL POPULATION OF YOUNG WOMEN, EXCEPT FOR ITS OVERSAMPLING OF BLACK WOMEN. A SUBSTANTIAL DEGREE OF INCONSISTENCY WAS FOUND BETWEEN THE 1968 AND 1972 ADMINISTRATIONS REGARDING RESPONSES TO THE QUESTION ABOUT THE PERSON OR PERSONS WITH WHOM THE RESPONDENT WAS LIVING AT THE AGE OF 14. THIS INCONSISTENCY IS SHOWN TO BE RELATED TO WHETHER OR NOT THE RESPONDENT SAID IN 1968 THAT AT AGE 14 SHE WAS LIVING WITH BOTH PARENTS. POSSIBLE EXPLANATIONS FOR THIS INCONSISTENCY ARE THAT SOME YOUNG PEOPLE LIVE IN HOUSEHOLDS WITH A FREQUENTLY CHANGING COMPOSITION, SO THAT THERE MAY BE NO SINGLE, TRUE ANSWER TO THE QUESTION FOR THEM, OR SOME RESPONDENTS MAY CHANGE THEIR RESPONSES TO A MORE SOCIAL DESIRABLE CATEGORY AS TIME PASSES. DESPITE THE INCONSISTENCY, HOWEVER, THE 1968 AND 1972 QUESTIONS YIELD SIMILAR CONCLUSIONS IN MULTIVARIATE RESEARCH WHEN THE RESPONSES ARE COLLAPSED INTO A FEW CATEGORIES. WHILE THE QUESTION IS STILL USEFUL AS A CRUDE INDICATOR OF PAST FAMILY LIFE, MORE REFINED MEASURES ARE NEEDED TO EXPLAIN THE INFLUENCE OF CHILDHOOD FAMILY COMPOSITION ON EVENTS IN ADULTHOOD. TABULAR DATA, NOTES, AND 15 REFERENCES ARE PROVIDED. (AUTHOR ABSTRACT MODIFIED)