«Demographic Research a free, expedited, online journal of peer-reviewed research and commentary in the population sciences published by the Max ...»
In 2004, the ratio of the number of students in academic and vocational high schools was about 7:3 (KEDI, 2005). Therefore, selecting the same number of students (2,000) from each of academic and vocational high schools for the ﬁnal sample of KEEP resulted in oversampling students in vocational high schools. To infer parameters for population of high school seniors (12th graders), the appropriate weight variable should be used to take into account the oversampling of vocational high school students.
The interview with a selected student was administered using PDA (Personal DigiThe overall response rate was quite high, as much as 93.8 percent for the student survey. The KEEP website (http://keep.nhrd.net/jsp/index.jsp) provides more detailed information on the survey method, sampling framework, and ﬁeldwork.
tal Assistant), which is an advanced version of CATI (Computerized Assisted Telephone Interview), in order to collect a large set of information including student’s school life, educational outcomes, and educational and occupational aspirations. A household member of the student (usually a mother or a father) completed a household questionnaire, which provided information on the household’s cultural, economic, educational, and social environments. As a panel survey, KEEP will be tracking sampled students’ educational and occupation trajectories over coming years. At the time when this manuscript was prepared, only data of the baseline year (2004) were available.
Two outcome variables Educational attainment is a major outcome on which most studies of family structure have focused in order to examine the effects of single parenthood. Given that the KEEP study has conducted only the baseline-year survey at this time, the current study is not able to assess the extent to which children from single-parent families are disadvantaged, for example, in actually attending or completing four-year colleges, compared to their counterparts from two-parent families.4 Instead, I examine two other outcomes related to education: educational aspiration and student disengagement.
A large body of research based on status attainment models has shown that a student’s educational aspiration is a critical factor predicting her or his subsequent educational and occupational attainment (Sewell, Haller, & Portes, 1969; Sewell & Hauser, 1975). Numerous studies in various countries have examined the variation in educational aspiration by family social status, gender, and race/ethnicity (Marjoribanks, 2005; Johnson, Crosnoe, & Elder, 2001). In the KEEP study, respondents were asked to indicate the highest level of educational attainment they would like to achieve. I constructed the outcome variable of educational aspiration as coded 1 if students wanted to complete four-year university education and 0 otherwise.
In addition to educational attainment or aspiration, researchers have been interested in student disengagement indicated by school attendance and punishment as an important dimension of educational outcomes, exploring individual-level and school-level determinants of student disengagement (Pellerin, 2005; National Research Council and the Institute of Medicine, 2004). Punishment is a straightforward measure of students’ misbehavior in schools. As a good measure indicating the degree of commitment to study and school (Astone & McLanahan, 1991), unexcused absence has been seriously considered as one of the early warning signs of educational failure, often leading to suspension, expulsion, truancy, and dropping out (CFFC 2002). I operationalize student disengagement as a dichotomous variable having 1 if respondents were absent from school at least
one day during the last school year or they received any kinds of punishment from school during their middle school or high school years. Note that 79 percent of total respondents in KEEP reported no absence at all during the last school year.5 Family structure Using information on whether the student lived with a biological father and/or a biological mother at the time of survey and on the reason for not living with
both, obtained from the household survey, I distinguish ﬁve types of family conﬁguration:
respondents living with (1) both biological parents, (2) a father only due to parental divorce, (3) a father only due to the death of the mother, (4) a mother only due to parental divorce, and (5) a mother only due to the father’s death. In this study, I exclude students living in stepfamilies. Reﬂecting the very small proportion of stepfamilies in the Korean family structure, the current data have only very small numbers of respondents who live in stepfamilies, which makes difﬁcult a reasonable analysis of this group (see Table 1 below). Therefore, this study limits its focus to comparisons between single-parent and two-(biological)-parent families with the distinction of single parenthood by parent’s gender and the causes of single parenthood.
Control variables The effects of single parenthood are estimated after controlling for parental education, gender of the respondent, and number of siblings. Parental education is the higher level of educational attainment of the parents (or educational attainment of a single parent), and it is included in the models as a continuous variable of the years of education completed. I recognize potential variation by gender in the relationship between family structure and educational outcomes. I ﬁrst examined the relationship separately for female and male students, and the results showed a similar pattern between the two groups in the ways in which each type of single-parent family is associated with the educational outcomes. Thus, I decided to include both female and male students in the same model, treating gender as a control variable.6 Household income To measure the economic condition of the family, I include household income. This information is obtained from the household survey. Household income 5 In KEEP, students were asked to indicate how many days they were absent from school without specifying reasons. Therefore, some of the absence might be health-related. However, literature in the United States has suggested that much of the absence is caused by disengagement (CFFC 2002). Moreover, in preliminary analysis of predicting school disengagement, I included student’s health status to somewhat control for the absence caused by health-related issues. The results were basically the same what are reported in the current study.
6 It is notable, however, that the relative advantage of students living with a single parent due to the death of a parent compared to those living with a single parent due to divorce was more apparent among female than male students.
indicates the average monthly income from all household members in the previous year (i.e., 2003). I classify students into three groups depending on their locations at the distribution of the household income: low (the ﬁrst quartile) household income, middle (the second quartile), and high (the third and the fourth quartile). I used different categorizations of household income and also a continuous form of it but the results were very robust to different speciﬁcations of household income.
Parent-child interaction The major indictors of parent-child interaction, which are particularly relevant for children’s educational outcomes, are parental involvement and interest in children’s education. I use three indicators that have been widely used in literature of parental involvement and children’s education (Scott, 2004; Kim, 2002; Astone & McLanahan, 1991). The ﬁrst measure is whether parents know the respondents’ future plan after high school graduation (i.e., whether respondents would proceed to higher education or whether they would enter labor market). This measure indicates the extent to which parents are interested in children’s progress and plan. The variable is coded 1 if parents know the plan of their children and 0 otherwise. The second measure of parental involvement is the degree to which parents know their children’s grade and school work, which intends to measure the extent to which parents monitor children’s progress in school. This variable has a scale that goes from 1 (don’t know at all) to 5 (know very well) and is used as a continuous measure in the analysis. The third variable is whether the student’s family has a rule regarding either TV program or hours of watching TV.
This variable indicates the overall degree of parents’ supervision on children’s day-to-day activities. Finally, I include a measure of student’s overall satisfaction with family life as another indicator of parent-child interaction. Although the variable does not explicitly measure student’s satisfaction in terms of the relationship with parents, it is reasonable to assume that the relationship with parents is a major aspect of family life. The variable has a scale that goes from 1 (not satisﬁed at all) to 5 (satisﬁed very much).
Table 1 presents the distributions of middle school and high school seniors by family type. In the column showing the distribution of middle school and high school seniors combined, 89 percent live with both biological parents, while 9 percent live with a single parent. Obvious is the very low prevalence of stepfamily in the Korean context: students in stepfamilies account for only 2 percent. Of those growing up with a single parent, twothirds live with a mother, while one-third live with a father. About half of those students living with a single mother experienced parental divorce, while the other half live with 390 http://www.demographic-research.org Demographic Research: Volume 18, Article 13 a single mother due to the death of the father. Most of single-father families are due to parental divorce. Note that because the sample size in the data is extremely small, I exclude children living in stepparent families from the analysis and focus on differences between children from single-parent and two-parent families.
Examining the distribution separately between middle school and high school seniors shows that single parenthood is slightly more prevalent among high school seniors (11 %) than middle school seniors (8 %). However, the relative share of each type of single parenthood is similar. Among both middle school and high school seniors who live with a single parent, parental divorce is the dominant cause of single parenthood among singlefather families, while divorce and widowhood equally account for single motherhood.
Socioeconomic conditions and parenting practices by family types It is useful to look at socioeconomic conditions and parent-child interaction of the four types of singleparent families in order to assess the extent to which differences in the two factors account for variation in educational outcomes across students from different types of family. Table 2 presents descriptive statistics on parental education, household income, and four measures of parent-child interaction across two-parent families and the four types of single-parent families. Apparently, students from two-parent families enjoy socioeconomic advantages compared to their counterparts from single-parent families, indicated by higher levels of both parental education and household income. About half of students from two-parent families belong to the category of “high” household income (3rd and 4th quartiles of the overall distribution), while the corresponding proportion is 15 percent among divorced-father families. The proportion is even less than 10 percent among the other three groups of single-parent families. Among the single-parent families, students from single-father families report relatively higher household income than do those from single-mother families. This pattern reﬂects the Korean employment and wage structure that favor men workers over women as discussed earlier.
The relative advantage of students from two-parent families is also obvious in regard to parent-child interaction. The extent to which parents know their child’s grade/schooling and their child’s plan after high school graduation is greater among students from twoparent families than those from single-parent families. The percentage of students whose families have a TV rule is highest among students with two parents. The level of student’s satisfaction with family life is also highest among students with two parents.
Students with a single mother are better off than those with a single father with respect to parent-child interaction. The degree to which parents know their child’s grade and schooling is greater among students with a single mother than those with a single father.
The higher percentage of students from single-mother families than those from singlefather families report that parents know their plan after high school graduation and also
that their families have a TV rule. Although students with a divorced mother report the similar level of life satisfaction as students with a widowed father, evident is the relatively high level of life satisfaction among students with a widowed mother.
In the bottom of the table, 85 percent of students from two-parent families aspire to four-year university education, whereas less than 60 percent of students from divorced single-father families do so. The degree to which students from other types of singleparent families aspire to four-year university education is between. Students with two parents are less likely to be disengaged than their counterparts who grow up with a single parent. In addition to the lowest level of educational aspiration, students from divorced single- father families show the highest level of disengagement. Evident from the table is that students with two parents are better off in both educational outcomes. Next, I turn to multivariate analyses to address the extent two which differences in two educational outcomes between students from two-parent and single-parent families remain after controlling for socioeconomic conditions, parent-child interaction, and other characteristics of students and their families.