Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of...

27
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=wmfr20 Download by: [The Education University of Hong Kong (EdUHK)], [Adam Cheung] Date: 04 October 2016, At: 23:59 Marriage & Family Review ISSN: 0149-4929 (Print) 1540-9635 (Online) Journal homepage: http://www.tandfonline.com/loi/wmfr20 Single Parenthood, Parental Involvement and Students’ Educational Outcomes in Hong Kong Adam Ka-Lok Cheung & Hyunjoon Park To cite this article: Adam Ka-Lok Cheung & Hyunjoon Park (2016) Single Parenthood, Parental Involvement and Students’ Educational Outcomes in Hong Kong, Marriage & Family Review, 52:1-2, 15-40, DOI: 10.1080/01494929.2015.1073650 To link to this article: http://dx.doi.org/10.1080/01494929.2015.1073650 Accepted author version posted online: 01 Sep 2015. Published online: 01 Sep 2015. Submit your article to this journal Article views: 287 View related articles View Crossmark data

Transcript of Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of...

Page 1: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=wmfr20

Download by: [The Education University of Hong Kong (EdUHK)], [Adam Cheung] Date: 04 October 2016, At: 23:59

Marriage & Family Review

ISSN: 0149-4929 (Print) 1540-9635 (Online) Journal homepage: http://www.tandfonline.com/loi/wmfr20

Single Parenthood, Parental Involvement andStudents’ Educational Outcomes in Hong Kong

Adam Ka-Lok Cheung & Hyunjoon Park

To cite this article: Adam Ka-Lok Cheung & Hyunjoon Park (2016) Single Parenthood, ParentalInvolvement and Students’ Educational Outcomes in Hong Kong, Marriage & Family Review,52:1-2, 15-40, DOI: 10.1080/01494929.2015.1073650

To link to this article: http://dx.doi.org/10.1080/01494929.2015.1073650

Accepted author version posted online: 01Sep 2015.Published online: 01 Sep 2015.

Submit your article to this journal

Article views: 287

View related articles

View Crossmark data

Page 2: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

Single Parenthood, Parental Involvement andStudents’ Educational Outcomes in

Hong Kong

ADAM KA-LOK CHEUNGDepartment of Social Sciences, The Hong Kong Institute of Education,

Tai Po, Hong Kong

HYUNJOON PARKDepartment of Sociology, University of Pennsylvania, Philadelphia, Pennsylvania, USA

Analyzing data from the Programme for International StudentAssessment Hong Kong 2009 and 2012 (n¼ 7,669), we examinedthe differences in socioeconomic characteristics of fathers andmothers and levels of parental involvement between two-parent,single-mother, and single-father families in Hong Kong. We foundthat parents from single-mother and single-father families faredifferently in terms of sociodemographic background and parentalinvolvement at home. We also investigated the differences instudents’ academic performance among these families. Past studiesfailed to find any significant effect of single parenthood on stu-dents’ academic performance in Hong Kong. We found negativeeffects of single fatherhood, but not single motherhood, on educa-tional outcomes. The disadvantages of single fatherhood arepartially explained by the poorer sociodemographic backgroundand lower levels of parental involvement.

KEYWORDS academic achievement, Hong Kong, parentalinvolvement, single-father families, single-parent families

Address correspondence to Adam Ka-Lok Cheung, Department of Social Sciences, TheHong Kong Institute of Education, D3-1=F-58, 10 Lo Ping Road, Tai Po, N.T., Hong Kong.E-mail: [email protected]

Color versions of one or more of the figures in the article can be found online atwww.tandfonline.com/wmfr.

Marriage & Family Review, 52:15–40, 2016Copyright # Taylor & Francis Group, LLCISSN: 0149-4929 print=1540-9635 onlineDOI: 10.1080/01494929.2015.1073650

15

Page 3: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

INTRODUCTION

Like other East Asian societies such as South Korea, Japan, Taiwan, and theMainland China, a steady increase in the divorce rate was evident in Hong Kongin the past few decades, particularly after the 1990s. In 1985 there were only 0.8divorces per 1,000 people in Hong Kong. The crude divorce rate was muchlower than the same figures in other OECD member countries and economiesat the time. In 2006 the crude divorce rate in Hong Kong rose to 2.5, which wasalready higher than most OECD countries (Dommaraju & Jones, 2011; Jones,2012). The rise of divorce rate has also increased the number of children livingwith a single parent. From 2001 to 2011 the annual growth rate of single parentswas 2.8% in Hong Kong. Currently, in Hong Kong, there are 81,680 singleparents who have children under age 18. Most children living with a single parentare children of divorce (Hong Kong Census and Statistics Department, 2013).

Cross-national studies in the Western context have found that divorce isconcentrated among lower class couples, and the differentials are increasingover time, which has important implications for disparities in resourcesavailable between children from advantaged families and their counterpartsfrom disadvantaged families (McLanahan, 2004). Similar patterns of divorceare found in some East Asian countries such as Korea (Park & Raymo,2013) and Japan (Ono, 2009; Raymo, Iwasawa, & Bumpass, 2004). Althoughlittle research has examined trends in educational differentials in divorce inHong Kong, studies have shown that compared with parents in two-parentfamilies, single fathers and single mothers in Hong Kong are more likely tobe less educated (Hong Kong Census and Statistics Department, 2013).In addition, single-parent families in Hong Kong are poorer, in terms ofincome level and housing conditions, than two-parent families (Hong KongCensus and Statistics Department, 2013).

Another important consideration in regard to well-being of singleparents and children in Hong Kong is strong family norms. Despite growingdiversity of family structure, cultural norms and preferences for traditionaltwo-parent family are still prevalent in Hong Kong (Chow & Lum, 2008).Compared with Western societies where divorce is more socially accepted,a larger proportion of the Hong Kong population still agree that couplesshould remain in an unhappy marriage when children are involved (Chow& Lum, 2008; Yodanis, 2005). In this context of strong family tradition, thepostdivorce experience may be particularly stressful to divorcees becauseof the formal and informal social sanctions against them (Lee & Law, 1994;Kung, Hung, & Chan, 2004). The well-being of children growing up insingle-parent families in this social context is worthy of further investigation.

Although evidence in the Western context consistently shows thatchildren from single-parent families have poorer academic performance thanthe children from intact families (Astone & McLanahan, 1991; Dronkers, 1994;

16 A. K.-L. Cheung and H. Park

Page 4: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

McLanahan & Sandefur, 1994), cross-national comparative studies by Pong,Dronkers, and Hampden-Thompson (2003), Marks (2006), and Hampden-Thompson (2013) have also found that the negative effects of single parent-hood on children’s academic performance vary across societal contexts.For example, Park (2007) found comparatively weak effects of single parent-hood on students’ reading performance in some Asian societies. In particular,he found that the effects of single parenthood are negligible or even some-what positive in Hong Kong, Korea, Indonesia, and Thailand, which heattributed to strong family systems in those countries. Similarly, Chiu andHo (2006) could not find any significant effect of single parenthood onstudents’ academic performance in Hong Kong.

Although past studies attempted to explain the weak and negligibleeffect of single parenthood in those societies with welfare and educationalpolicies and a tradition of strong family ties (Hampden-Thompson, 2013;Park, 2007; Pong et al., 2003), the potential role of gender of a single parenthas not been thoroughly examined. Many past studies combinedsingle-father and single-mother families into one category of single-parentfamilies, but relatively little is known about whether the absence of negativeeffects of single parenthood on students’ educational outcomes applies toboth single-father and single-mother families in these Asian societies (forexception, see Park, 2008a, 2014).

Past research has shown that familial economic conditions and parentalinvolvement are important factors associated with differences in students’educational outcomes by family structure (Chiu & Ho, 2006; Hampden-Thompson, 2013; McLanahan & Sandefur, 1994). Specifically, Lee and Law(1994) found that single parenthood is related to less parental involvementin Hong Kong. Park (2007) also showed students with a single parentwere more likely to live in poor socioeconomic conditions than students withtwo parents in five Asian countries, although the socioeconomic differencesbetween single-parent and two-parent families were generally smaller inAsia than in the United States. However, without distinction betweensingle-mother and single-father families, we do not know much about moredetailed differences in socioeconomic conditions and parental involvementamong two-parent, single-mother, and single-father families in Asia.

In this article, we extend the current literature of single parenthood andchildren’s educational outcomes in Asia by further separating single-parentfamilies into single-mother and single-father families. Analyzing combined datafrom the Programme for International Student Assessment (PISA) Hong Kong2009 and 2012 (n¼ 7,669), we examined the differences in socioeconomiccharacteristics of fathers and mothers and levels of parental involvementbetween two-parent families, single-mother families, and single-father familiesin Hong Kong. We also investigated the differences in students’ academicperformance among these families and how these differences can be explainedby the familial sociodemographic background and levels of parental involvement.

Single Parenthood in Hong Kong 17

Page 5: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

PAST LITERATURE AND THE CURRENT STUDY

Under the context where out-of-wedlock childbearing is rare and mortalityrate for nonelderly is low, divorce is the leading cause of single parenthoodin Hong Kong (Hong Kong Census and Statistics Department, 2013). Unlikein the United States where shared custody is increasingly prevalent (Cancian& Meyer, 1998), shared custody is rare in Hong Kong (Sullivan, 2005). Similarto many other countries, mother sole custody is often granted, and single-mother family is the dominant group of single-parent families in Hong Kong(Kung et al., 2004). According to the latest official figure, there were only 276single fathers per 1,000 single mothers in 2011 (Hong Kong Census andStatistics Department, 2013). Because single-father families are relatively rarein many contexts, most past studies solely focused on children’s educationaloutcomes in either single-mother families (e.g., Creighton, Park, & Teruel,2009; Hampden-Thompson, 2013) or combined single-father and single-mother families into one category—single-parent families (e.g., Astone &McLanahan, 1991; Chiu & Ho, 2006; Marks, 2006; Pong et al., 2003).

However, combining single-father and single-mother families canbe problematic in this area of research because it fails to consider potentialdifferences between single-father families and single-mother families.Previous studies have already showed that the parent’s gender in single-parent families matters in terms of parenting (Biblarz & Stacey, 2010). Notsurprisingly, parenting alone is difficult. Single parents often have moredifficulties in childcare than parents in two-parent families (Lee & Law,1994). Single fathers hold more negative attitudes in parenting and are lessinvolved in parenting and in parent–school activities than single mothers(Downey, 1994; Dufur, Howell, Downey, Ainsworth, & Lapray, 2010;Hawkins, Amato, & King, 2006). Single fathers are also less likely to monitorchildren’s behaviors than single mothers, and that may lead to moreantisocial behaviors of children from single-father families than those fromsingle-mother families (Breivik, Olweus, & Endresen, 2009).

The differences in parenting between fathers and mothers may be exacer-bated by the Asian context. Choi et al. (2012b) found that men in Hong Kongare constantly stressed by their male gender roles. As prescribed by the culturalnorms, most fathers prioritize the responsibility of providing financial supportto the family over parenting. In particular, traditional cultural norms may pre-vent them from expressing their feelings and affection (Choi et al., 2012b).Meanwhile, over half of the adult respondents in a representative survey agreedthat women ‘‘should focus more on family than work’’ (Women’s Commission,2011, p. 7). With reference to the past studies, we may expect that single fathersmay be less involved than single mothers in parenting and in school activities.However, little is known about the patterns of parental involvement betweenthe two types of single-parent families in Hong Kong.

18 A. K.-L. Cheung and H. Park

Page 6: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

Second, single fathers and single mothers fare differently in terms ofeconomic well-being and have different demographic characteristics. Singlefathers and single mothers differ in their work and wealth status (Hong KongCensus and Statistics Department, 2013). It is noteworthy that the female laborforce participation rate is significantly lower than the male labor force partici-pation rate in Hong Kong and many other Asian societies. Labor market isgender-segregated, and the traditional cultural norm against women workingoutside the home is still well-rooted. For instance, in 2009, 96.5% of men aged30 to 34 in Hong Kong were in the labor force, whereas the respective figurefor women was only 79.7%. Single mothers are less likely to work and not aswealthy as single fathers. The median monthly employment income for singlefathers in Hong Kong was HK$12,000, whereas the same figure for singlemothers was HK$9,000 (Hong Kong Census and Statistics Department,2013). In addition, although women are often granted custody of the children,child custody after parental divorce is also determined with reference to theparents’ capability in providing childcare. To be granted child custody bythe court, divorced mothers sometimes have to give up their time-demandingjobs to provide close supervision to the child and therefore sacrifice theircareers (Kung et al., 2004). In short, fathers and mothers from single-fatherfamilies, single-mother families, and two-parent families are socioeconomicallydifferent. Nevertheless, little research in Hong Kong has addressed the differ-ences in parental socioeconomic characteristics among different types of fam-ily structure. To have a better understanding of the conditions under whichchildren are growing up in single-father and single-mother families in HongKong, we need to examine in detail heterogeneity in socioeconomic con-ditions and parents’ characteristics between single fathers and single mothers.

Because parental socioeconomic background and parental involvementare related to students’ academic performance (Park, 2008b), the differencesin parental background and involvement between children with a singleparent and their peers with two parents may at least partly account for thedifferences in academic performance between students from single-parentfamilies and their counterparts from two-parent families. Several studies inother countries have revealed that students from single-mother families farebetter than their counterparts from single-father families, although the differ-ences are not always consistently significant across samples (Downey, 1994;Downey, Ainsworth-Darnell, & Dufur, 1998; Park, 2014). Park (2008a, 2014)found that students from single-father families fare worse than students fromsingle-mother families and two-parent families in Korea. Studying U.S.families, Downey (1994) argued that the mechanisms leading to poorerperformance for children from single-father and single-mother families, ascompared with two-parent families, are different. Specifically, socioeconomicdeprivation primarily explains the vulnerability of single-mother families,whereas the lack of parental involvement is a particularly important factorfor explaining the problems of single-father families.

Single Parenthood in Hong Kong 19

Page 7: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

Regarding the literature of single parenthood and academic perfor-mance, past studies have argued that the negative effect of single parenthoodis negligible in contexts where the family and education policies provideadequate and relatively equal support to all students (Chiu & Ho, 2006). Itis then an interesting question whether the absence of an effect of singleparenthood applies to both single-father and single-mother families. Giventhat single-mother families are the dominant group in single-parent familiesin Hong Kong, the absence of an effect of single parenthood in the currentpolicy context may indeed indicate the absence of an effect of single mother-hood only but may not necessarily indicate the absence of an effect of singlefatherhood. For instance, Chiu and Ho (2006) argued that the educational pol-icy in Hong Kong, which provides equal funding to each student in public andgovernment-aided schools, reduces the adverse effect of family socioeco-nomic disadvantages and may explain the absence of negative effect of singleparenthood on students’ academic performance. If single-mother families aredisadvantaged in terms of socioeconomic background and single-father fam-ilies have less parental involvement, we may still expect a negative effect ofsingle fatherhood despite educational policies to help families with socioeco-nomic hardship. This is subjected to further empirical investigation.

In this study, we attempt to answer three research questions: (1) Areparents from single-father families and single-mother families more likely tobe economically disadvantaged, and less likely to be involved in children’seducation, compared with parents from two-parent families? (2) Do studentsfrom single-father families and single-mother families have lower testscores than students from two-parent families? (3) How do socioeconomicconditions and parental involvement mediate the relationship between singlemotherhood, single fatherhood, and students’ academic performance?

METHODS

Data

Because single-father families constitute a relatively small group in thepopulation, a large sample size is required for making any statistically mean-ingful comparison between single-father families and other types of families.To examine the relationships among single parenthood, family’s socio-economic conditions, parental involvement, and academic test scores, weanalyzed data from a combined sample of PISA Hong Kong 2009 and 2012.The combined data from both years provide enough cases for children livingwith a single father as well as a single mother. PISA is a triennial internationalsurvey that assessed the skills and knowledge (including reading, mathe-matics, and science literacy) of 15-year-old students in OECD members andsome non-OECD countries and economies. PISA Hong Kong is a territory-widerepresentative sample, drawing with a multistage sampling procedure.

20 A. K.-L. Cheung and H. Park

Page 8: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

A stratified sample of schools within Hong Kong was first drawn.Students aged 15 were then sampled within each sampled school. The HongKong PISA reports provide the detailed description of sampling procedureand data collection approach (HKPISA Centre, 2011; HKPISA Centre, 2013).In total, 9,507 students from 299 schools in the combined data of PISA 2009and 2012 completed the survey. Information collected from parental andstudent questionnaires were used in this study. Listwise deletion strategywas used to handle missing data in this study. Excluding the cases withmissing values, our analytic sample consists of 7,669 cases. We used thestudent weight variable provided by the dataset in our statistical analysis.

Measures

FAMILY STRUCTURE

Family structure is the main independent variable in this study. In the studentquestionnaire, students were asked with whom they were usually living.The questionnaire was identical for PISA 2009 and 2012. Based on responses,four categories were constructed: two-parent families (students living withboth father and mother), single-father families (students living with fatheronly), single-mother families (students living with mother only), and others(students living with neither father nor mother). The ‘‘others’’ (as a residualcategory) in this study are included in the regression analysis but are omittedin the tables.

Note that with these PISA measures of living arrangement, we are notable to further distinguish single-parent families by reasons (e.g., whetherdue to divorce, widowhood, or other). Although we acknowledge this limi-tation of our single-parent family measure, we have already mentionedabove that divorce is the major cause of single parenthood in Hong Kong.Another limitation of our family structure measure is that we cannot dis-tinguish between a biological parent and a stepparent as the wording of‘‘mother’’(or ‘‘father’’) in the PISA questionnaire included a stepmother(father) or a foster mother(father) as well as a biological mother(father).However, in Hong Kong the number of children living with a stepmother(father) or a foster mother(father) is relatively small (Wang, 2004), and there-fore this limitation would not seriously affect our study.

READING, MATH, AND SCIENCE LITERACY

Our main dependent variables are students’ scores on reading, mathematics,and science literacy tests. All students who participated in PISA were assessedfor skills and knowledge on the three mentioned domains. Item ResponseTheory was used to generate each of the three literacy scores. Instead ofone single score for each test, five plausible values are provided, whichshould be analyzed simultaneously. In PISA, each test score is standardized

Single Parenthood in Hong Kong 21

Page 9: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

to have a mean of 500 points and a standard deviation of 100 points acrossOECD students. To see whether the result may vary across different domainsof literacy, we analyzed reading, math, and science scores separately.

PARENTAL INVOLVEMENT AT HOME

Parents in Hong Kong filled in the parent questionnaire where the samequestion of ‘‘how often do you or someone else in your home do the follow-ing things with your children,’’ was asked for each of different parentalinvolvement activities separately. Although these questions were identicalbetween PISA 2009 and 2012, parents were required to respond on the basisof four ordinal categories in PISA 2009 but five ordinal categories in PISA2012. By combining two lowest categories in PISA 2012, we were able tocreate the same scale for both PISA 2009 and 2012 questions on the basisof four ordinal categories: less than once a month, once or twice a month,once or twice a week, and every day or almost every day.1

Literature on parental involvement has highlighted the relevance of vari-ous activities parents do at home for their children’s educational outcomes(Downey, 2002; Park, Byun, & Kim, 2011). Among various kinds of parentalinvolvement activities at home, we examined four different activities: discuss-ing well-being at school with the child, eating the main meal together, spend-ing time talking to the child, and providing homework help to the child.2

Discussing with the child either generally or specifically about schooling isone of the major involvement measures that previous studies have paid atten-tion to and often have found to be most significantly related to better edu-cational outcomes of students among other types of parental involvement(Ho & Willms, 1996; McNeal, 1999; Park, 2008b). Recent studies are alsointerested in potential benefits of eating meals together for children’swell-being and examine how children in different types of families differin the extent to which they eat meals with their parents (Musick & Meier,2012; Raymo, Park, Iwasawa, & Zhou, 2014). Although it is difficult todetermine the direction of the relationship between helping in homeworkand children’s educational outcomes, helping in (or checking and monitor-ing) children’s homework has also been widely examined in the literatureof parental involvement (Ho & Willms, 1996).

PARENTAL INVOLVEMENT IN SCHOOL

In addition to the four variables of parental involvement at home, we alsoexamined patterns of parental involvement in school. Literature, especiallyin the United States, has focused on the extent to which parents communi-cate with teachers=schools and participate in PTO meetings and other schoolactivities, although the relationship between various kinds of parentalinvolvement in school and children’s educational outcomes is often tenuous

22 A. K.-L. Cheung and H. Park

Page 10: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

or even reverse (Ho & Willms, 1996; McNeal, 1999). In the current studywe examined parents’ participation in any of the following six items: (1)volunteer in physical activities, (2) volunteer in extracurricular activities,(3) volunteer in the library, (4) assist a teacher in the school, (5) appear asa guest speaker, and (6) participate in local school governance. For eachitem parents indicated yes or no. Considering that the share of parents inHong Kong who participate in each school activity is fairly small, we didnot examine each activity separately. Instead, we created a dichotomousvariable that indicates participation in any of the six activities (1) versusparticipation in none of these school-related activities in the last academicyear (0).

PARENTS’ SOCIODEMOGRAPHIC CHARACTERISTICS

We took into account fathers’ and mothers’ full-time employment status.In addition, we included the highest parental education level and householdwealth index. For the highest parental education level, the education level ofmothers in single-mother families and the education level of fathers insingle-father families were used, whereas the education level of the parentwith the higher education level was used for the two-parent and other fam-ilies. Household wealth index is included in the original PISA data, and it wascreated as a composite measure to indicate the household possessions,including the availability of the child’s own room, a dishwasher, educationalsoftware and Internet connection, and the number of cars, computers,televisions, mobile phones, and bathrooms.

OTHER INDIVIDUAL AND HOUSEHOLD COVARIATES

In addition to the above socioeconomic and demographic backgrounds ofstudents that likely affect students’ academic performance, we also includedother individual-level variables such as students’ gender, birthplace, andschool grade as controls. Student’s birthplace variable indicates whethera student was born in Hong Kong. In addition, we also controlled for parents’immigration status for which we distinguish students whose fathers andmothers were not locally born. Previous studies suggested that a parent’simmigration status is related to family well-being and children’s academicperformance in Hong Kong (Choi, Cheung, & Cheung, 2012a; Zhu & Leung,2011). In addition, we took into account whether a student lives with agrandparent. Because a co-residing grandparent can play an important rolein childrearing and in providing emotional support for children, especiallyin single-parent families, it is necessary to control for co-residence with agrandparent to precisely estimate the effect of single parenthood on chil-dren’s education. Finally, we also take into account whether a student livestogether with siblings.3

Single Parenthood in Hong Kong 23

Page 11: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

SCHOOL-LEVEL COVARIATES

In addition to individual-level variables, we consider differences in schoolcharacteristics among students from different family types. We considerschool types and school-level mean household wealth index. School typesare used as sampling stratum in the PISA Hong Kong. Four major types ofschools are included in Hong Kong: government schools, aided and caputschools, private schools, and schools under the Direct Subsidy Scheme.The first two types of schools have rigid guidelines to follow regardingcurriculum design, fees, and entrance requirements set by the EducationDepartment, whereas the latter two types have more flexibility inthese aspects of school management. In our analysis we grouped the formertwo types of schools as Government=Aided Schools and the latter two typesof schools as Private=Direct Subsidy Scheme Schools. We created the variableof school mean household wealth index by averaging student’s householdwealth index values among students attending the same school.

ANALYTICAL STRATEGY

Differences in Parental Involvement by Family Types

We first compared the degree of parental involvement at home and in schoolfor different types of family structure. Specifically, we estimated a series ofnested ordinal and binary logistic regression models that predict the degreeof parental involvement at home and in school, respectively. In the baselinemodel we only controlled for the survey year, students’ characteristics (age,gender, and birth place), parents’ birthplaces, and co-residence with a grand-parent and sibling. In Model 2 we additionally controlled for socioeconomicstatus (SES) background (parents’ educational level, mother’s and father’sfull-time employment status, and household wealth). Then, in our finalmodel (Model 3), we took into account potential differences in schoolcharacteristics (school types and school-level mean household wealth).These models examined whether there are significant differences in parentalinvolvement among different family types and net of the impacts of thestudent, parent, and school characteristics. Because logit coefficientsbetween nested models cannot be compared directly due to unobservedheterogeneity (Mood, 2010), we additionally calculated y�-standardized logitcoefficients for comparison (for a more technical discussion of this method,please see Winship & Mare, 1984). Taking the complex sample design of PISAdata into account, all analyses are weighted and variances are cluster-adjusted using the balanced repeated replication methods. Because of thecomplication of y-standardization procedure and weighting=variance issues,we did not estimate multilevel models for the final models, which includethe school characteristics. As sensitivity analysis, we additionally estimated

24 A. K.-L. Cheung and H. Park

Page 12: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

models with both binary logistic regression (with collapsed categories) andordinal logistic regression for each dependent variable in this study. Ourfindings are robust to the choice of modeling.

Explaining Differences in Test Scores Among Students fromDifferent Family Types

To examine the relationship between family structure and students’academic performance, we ran a series of nested linear regression modelsfor each of reading, mathematics, and science test scores. In the baselinemodel (Model 1), we only controlled for the survey year in addition tofamily structure. In Model 2 we additionally controlled for the students’,parents’, and household sociodemographic characteristics and schoolcharacteristics. In Model 3, in addition to the covariates included in Model2, we controlled for the parental involvement at home variables. In Model 4we controlled for the covariates in Model 2 and also parental involvementin school but without parental involvement at home variables from Model3. In the final model (Model 5), we included both background characteris-tics in Model 2 (including student, parent, and school characteristics) andparental involvement at home and school in Model 3 and Model 4. Thesemodels account for the extent to which differences in test scores amongstudents from different family types are explained by the backgroundcharacteristics and parental involvement.

RESULTS

Descriptive Statistics

Table 1 shows the distribution (percentages) of parental involvementactivities by family structure in the analytic sample. At the top of Table 1,the distribution of students in regard to their family structure is presented.Most 15-year-old students (87.25%) in Hong Kong lived in two-parentfamilies; 8.86% of students lived with a single mother and only 2.01% of stu-dents lived in single-father families. Only 1.87% of students were not livingwith any parent.

Turning to parental involvement measures, we found that for slightlymore than two-thirds of 15-year-old students (68.6%), parents discussed withtheir children about their well-being at school at least once a week. Familystructure is significantly associated with the frequency of parent–childdiscussion over well-being at school. Parents from two-parent familiesdiscussed well-being at school with students most frequently. Comparatively,parents from single-parent families discussed well-being at school withchildren significantly less frequently than parents from two-parent families(t-test for single mothers: p< .01; for single fathers: p< .001). Among

Single Parenthood in Hong Kong 25

Page 13: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

TA

BLE

1D

istrib

ution

ofPar

enta

lIn

volv

em

entVar

iable

sby

Fam

ily

Stru

cture

(n¼

7,6

69)a

Overa

llTw

o-P

arent

(87.2

5%

)Si

ngle

-Moth

er

(8.8

6%

)Si

ngle

-Fat

her

(2.0

1%

)O

thers

(1.8

7%

)P

Val

ue

b

Dis

cuss

well-b

ein

g=doin

gat

school

Less

than

once

am

onth

8.2

4%

7.9

2%

9.6

2%

14.4

4%

9.8

4%

t-te

st1:<

.01

Once

or

twic

ea

month

23.1

7%

22.5

9%

25.3

7%

27.8

2%

34.7

8%

t-te

st2:<

.001

Once

or

twic

ea

week

36.8

9%

37.0

6%

36.7

4%

39.4

4%

26.9

4%

t-te

st3:<

.05

Every

day

or

alm

ost

every

day

31.7

0%

32.4

3%

28.2

8%

18.3

0%

28.4

4%

Eat

mai

nm

eal

Less

than

once

am

onth

1.4

3%

1.4

0%

1.3

4%

2.2

3%

1.9

9%

t-te

st1:<

.001

Once

or

twic

ea

month

2.3

2%

1.9

4%

4.4

1%

4.7

7%

7.2

3%

t-te

st2:<

.001

Once

or

twic

ea

week

10.0

8%

8.8

0%

19.0

8%

18.1

4%

18.9

2%

t-te

st3:n.s

.Every

day

or

alm

ost

every

day

86.1

7%

87.8

6%

75.1

6%

74.8

6%

71.8

6%

Spendin

gtim

eta

lkin

gLe

ssth

anonce

am

onth

2.5

8%

2.4

7%

2.8

0%

5.9

0%

3.3

8%

t-te

st1:n.s

.O

nce

or

twic

ea

month

6.5

6%

6.2

7%

6.8

4%

9.5

3%

15.3

5%

t-te

st2:<

.001

Once

or

twic

ea

week

24.2

0%

23.5

9%

26.1

2%

34.2

1%

32.7

6%

t-te

st3:<

.01

Every

day

or

alm

ost

every

day

66.6

6%

67.6

7%

64.2

4%

50.3

6%

48.5

2%

Hom

ew

ork

help

Less

than

once

am

onth

51.0

2%

50.1

2%

58.8

7%

48.6

4%

58.6

2%

t-te

st1:<

.001

Once

or

twic

ea

month

24.7

7%

25.2

3%

22.2

5%

25.2

5%

14.8

3%

t-te

st2:n.s

.O

nce

or

twic

ea

week

16.6

0%

16.9

2%

13.0

9%

17.2

8%

17.5

4%

t-te

st3:<

.05

Every

day

or

alm

ost

every

day

7.6

1%

7.7

3%

5.7

8%

8.8

3%

9.0

1%

Involv

em

entin

school

No

84.4

8%

84.4

7%

84.7

0%

83.9

0%

84.8

4%

t-te

st1:n.s

.Y

es

(atle

ast1

item

)15.5

2%

15.5

3%

15.3

0%

16.1

0%

15.1

6%

t-te

st2:n.s

.t-te

st3:n.s

.

aCom

bin

ed

anal

ytic

sam

ple

,w

eig

hte

d%

,w

eightan

dvar

iance

adju

sted.

bt-Test

1co

mpar

es

two-p

arentan

dsi

ngle

-moth

er

fam

ilie

s;t-te

st2

com

par

es

two-p

arentan

dsi

ngle

-fat

her

fam

ilie

s;t-te

st3

com

par

es

single

-moth

er

and

single

-fat

her

fam

ilie

s.

26

Page 14: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

single-parent families, single mothers discussed well-being at school withchildren more frequently than single fathers (p< .05).

Not surprisingly, eating the main meal with children is the mostcommon item among all measures in parental involvement. Overall, over86% of students eat their main meal with parents almost every day. Less than4% of students only eat the main meal with parents less than once a month.The pattern, however, is also associated with family structure. Eating themain meal together is more common for two-parent families (87.86% eatthe main meal together almost every day). Eating the main meal togetheris significantly less frequent for single mothers (75.16% almost every day,p< .001) and single fathers (74.86% almost every day, p< .001). However,there is no significant difference in the frequency of eating the main mealwith the child between single mothers and single fathers.

For about two-thirds of students (66.7%), parents spent time talking withthe students almost every day. However, parents from single-father familiestalked with their children less frequently than parents in two-parent families(p< .001) and single-mother families (p< .01). Only 50.4% of them talkedwith the students almost every day as compared with two-parent families(67.7%) and single-mother families (64.2%).

Compared with other parental involvement measures, providinghomework help is not as common in Hong Kong. Slightly more than halfof students (51.0%) received help from parents less than once a month. Lessthan one-fourth of students (24.2%) received help from parents at least oncea week. Parents from two-parent families (p< .001) and single-father families(p< .05) provided more frequent help to their children, compared withsingle mothers. It is interesting to see more homework help received amongstudents in single-father than in single-mother families.

Parents’ involvement in school activities was relatively uncommonin Hong Kong. Only about 15% of parents were involved in any of the sixschool activities. Although using a different dataset, Park et al. (2011) alsoshowed a considerably low level of school contact by Korean parents,attributing it to the impact of the highly standardized Korean educationsystem. Moreover, there is no significant bivariate association between familystructure and parental involvement in school (p¼ .99).

In summary, significantly less frequent parental involvement at home, butnot in school, was found among single-parent families than among two-parentfamilies. Single-father families had less parental involvement than single-mother families for the items that require more verbal communication andemotional support (discussing well-being at school and spending time talking)but had more involvement in providing homework help than single mothers.

Figure 1 shows the students’ average test scores in each subject by familystructure. Family structure is significantly associated with all three test scores at.001 level. Students from two-parent families performed the best on allreading, mathematics, and science scores. Data present no statistical evidence

Single Parenthood in Hong Kong 27

Page 15: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

that students from single-mother families performed worse than studentsfrom two-parent families. Yet, students from single-father families performedsignificantly worse than students from two-parent families (p< .001) andsingle-mother families (p< .01) in all three domains. Students from single-father families on average got about 35 points and 31 points less than studentsfrom two-parent families and single-mother families, respectively.

Socioeconomic Characteristics of Single-Parent Families

However, the differences in parental involvement at home and academicscores may be attributable to the differences in other background factors,including parents’ SES. Table 2 shows the distribution of these backgroundcharacteristics among the four types of families.

Bivariate analyses show that family structure was associated with stu-dent characteristics, although student gender was not significantly associatedwith family structure (p¼ .163). Students from two-parent families were morelikely locally born as compared with students from single-mother families(p< .01) and single-father families (p< .05). Most 15-year-old students arein grade 10 (Secondary 4). Students from single-mother families (p< .05)and single-father families (p< .01) were more likely to be at lower schoolgrades. However, student characteristics were not significantly differentbetween single-mother and single-father families.

As expected, most parental and household characteristics were alsoassociated with single parenthood. Cross-border and migrant marriages are

FIGURE 1 Academic ability scores (readings, mathematics and science) by family structure.

28 A. K.-L. Cheung and H. Park

Page 16: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

TA

BLE

2St

udent,

Fam

ily,an

dSc

hoolChar

acte

rist

ics

by

Fam

ily

Stru

cture

a

Overa

llTw

o-P

arent

Single

-Moth

er

Single

-Fat

her

Oth

ers

PVal

ue

b

Studentch

arac

terist

ics

Gra

de

71.2

8%

1.2

4%

1.0

5%

2.2

3%

3.4

0%

t-te

st1:<

.05

86.0

6%

5.7

1%

7.0

1%

12.3

8%

11.0

2%

t-te

st2:<

.01

924.8

0%

24.2

9%

27.7

0%

28.7

4%

30.5

8%

t-te

st3:n.s

.10

66.9

7%

67.8

3%

64.0

0%

54.3

3%

54.4

9%

11

0.8

9%

0.9

3%

0.2

3%

2.3

2%

0.5

1%

Gender

Mal

e47.8

9%

47.5

9%

48.8

8%

47.6

5%

57.5

0%

t-te

st1:n.s

.Fem

ale

52.1

1%

52.4

1%

51.1

2%

52.3

5%

42.5

0%

t-te

st2:n.s

.t-te

st3:n.s

.St

udent’s

birth

pla

ceBorn

inH

ong

Kong

77.8

4%

78.6

1%

73.4

8%

69.4

4%

71.8

5%

t-te

st1:<

.01

Born

outs

ide

Hong

Kong

22.1

6%

21.3

9%

26.5

2%

30.5

6%

28.1

5%

t-te

st2:<

.05

t-te

st3:n.s

.Fam

ily

char

acte

rist

ics

Moth

er’s

birth

pla

ceBorn

inH

ong

Kong

48.2

8%

49.1

2%

44.1

1%

40.0

3%

37.6

8%

t-te

st1:<

.05

Born

outs

ide

Hong

Kong

51.7

2%

50.8

8%

55.8

9%

59.9

7%

62.3

2%

t-te

st2:<

.05

t-te

st3:n.s

.M

oth

er

em

plo

ym

ent

Full-tim

e49.8

2%

50.4

1%

45.4

8%

48.5

3%

43.9

5%

t-te

st1:<

.05

Notfu

ll-tim

e50.1

8%

49.5

9%

54.5

2%

51.4

7%

56.0

5%

t-te

st2:n.s

.t-te

st3:n.s

.Fat

her’s

birth

pla

ceBorn

inH

ong

Kong

57.3

5%

57.3

5%

56.0

5%

65.3

0%

54.8

8%

t-te

st1:n.s

.Born

outs

ide

Hong

Kong

42.6

5%

42.6

5%

43.9

5%

34.7

0%

45.1

2%

t-te

st2:n.s

.t-te

st3:<

.05

Fat

her

em

plo

ym

ent

Full-tim

e81.4

5%

83.8

9%

59.8

1%

73.0

4%

78.8

1%

t-te

st1:<

.001

Notfu

ll-tim

e18.5

5%

16.1

1%

40.1

9%

26.9

6%

21.1

9%

t-te

st2:<

.001

t-te

st3:<

.01

(Con

tin

ued

)

29

Page 17: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

TA

BLE

2Continued

Overa

llTw

o-P

arent

Single

-Moth

er

Single

-Fat

her

Oth

ers

PVal

ue

b

Par

ent’s

educa

tion

level

Low

er

seco

ndar

yor

belo

w33.6

5%

32.1

0%

44.9

6%

50.1

4%

24.7

0%

t-te

st1:<

.001

Upper

seco

ndar

y47.3

4%

47.8

7%

42.9

1%

36.7

7%

54.8

6%

t-te

st2:<

.001

Tertia

ry19.0

1%

20.0

3%

12.1

3%

13.0

9%

10.4

4%

t-te

st3:n.s

.

House

hold

weal

thin

dex

�0.9

9�

0.9

5�

1.3

0�

1.2

4�

1.2

2t-te

st1:<

.001

t-te

st2:<

.001

t-te

st3:n.s

.Li

vin

gw

ith

gra

ndpar

ents

No

85.3

3%

86.4

8%

87.6

6%

70.7

0%

36.1

8%

t-te

st1:n.s

.Y

es

14.6

7%

13.5

2%

12.3

4%

29.3

0%

63.8

2%

t-te

st2:<

.001

t-te

st3:<

.001

Livin

gw

ith

siblings

No

25.2

1%

22.3

9%

37.8

4%

51.5

4%

68.4

2%

t-te

st1:<

.001

Yes

74.7

9%

77.6

1%

62.1

6%

48.4

6%

31.5

8%

t-te

st2:<

.001

t-te

st3:<

.01

Schoolch

arac

terist

ics

Schoolty

pes

Aid

ed=govern

mentsc

hool

88.0

8%

88.0

6%

87.1

7%

91.2

9%

89.6

0%

t-te

st1:n.s

.D

irect

subsi

diz

ed=privat

e11.9

2%

11.9

4%

12.8

3%

8.7

1%

10.4

0%

t-te

st2:n.s

.t-te

st3:n.s

.Sc

hoolm

ean

weal

thin

dex

�1.0

0�

1.0

0�

1.0

3�

1.1

1�

1.0

7t-te

st1:<

.05

t-te

st2:<

.001

t-te

st3:<

.01

aCom

bin

ed

sam

ple

,w

eig

hte

d%

,w

eig

htan

dvar

iance

adju

sted.

bt-Test

1co

mpar

es

two-p

arentan

dsi

ngle

-moth

er

fam

ilie

s;t-te

st2

com

par

es

two-p

arentan

dsi

ngle

-fat

her

fam

ilie

s;t-te

st3

com

par

es

single

-moth

er

and

single

-fat

her

fam

ilie

s.

30

Page 18: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

increasingly prevalent in Hong Kong. Studies suggest that compared withmarriage between native-born couples, couples in cross-border marriageare more likely to encounter marital conflict in Hong Kong (Choi et al.,2012a). It is possible that couples of cross-border marriage are more likelythan those in local marriage to end their marriages. Compared with intactfamilies, mothers of students from single-father families (p< .05) and single-mother families (p< .05) were more likely to have been born outside HongKong. However, father’s birthplace was not significantly different betweentwo-parent families and single-parent families. Yet, fathers from single-fatherfamilies were more likely to have been locally born than fathers fromsingle-mother families (p< .05). Parents in two-parent families were morelikely to work full-time. Mothers in single-mother families were less likelyto work full-time than the mothers from two-parent families (p< .05). Fathersin single-mother and single-father families were also less likely to workfull-time (p< .001). Comparatively speaking, fathers from single-fatherfamilies were more likely to work full-time than fathers of students fromsingle-mother families (p< .01). Parents’ education level was also higher inintact families (p< .001) as compared with single-mother and single-fatherfamilies. In addition, single-mother and single-father families were notas wealthy as two-parent families (p< .001). In general, single-parentfamilies were more socioeconomically disadvantaged. Among single-parentfamilies, only limited evidence in our sample showed that single-mother familieswere more disadvantaged than single-father families.

Students from single-father families were more likely to live with grand-parents than students from single-mother families (p< .001) and two-parentfamilies (p< .001). Students from two-parent families, on the other hand,were more likely to live with a sibling than those living with single mothers(p< .001) and single fathers (p< .001).

Finally, we also examined the bivariate association between students’family structure and the characteristics of the schools they attended. Schooltypes were not significantly associated with family structure (p¼ .646).However, students from two-parent families were more likely to attendschools with a higher school-level mean household wealth index than thosefrom single-mother (p< .05) and single-father families (p< .001). Compara-tively speaking, students with single mothers were also more likely to attendschools with a higher school-level mean household wealth index than thosewith single fathers (p< .01).

REGRESSION ANALYSES

In this section, we first report the results of binary=ordinal logistic regressionmodels to examine the differences in the parental involvement at homeand in school among the four types of families while controlling for

Single Parenthood in Hong Kong 31

Page 19: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

blocks of students, parents, and school characteristics. Table 3 shows theodds ratios and y-standardized logit coefficients of single-parent families(compared with two-parent families as the reference category) on parentalinvolvement measures in the nested models. Results of additional t-testscomparing the coefficients of single-mother and single-father families arealso reported in Table 3.

Parents from single-mother and single-father families discussed well-being in school with children significantly less than parents from two-parentfamilies, even after controlling for students’ characteristics, parents’ birth-place, and co-residence with sibling and grandparents. When parental SESwere additionally controlled in Model 2, parents from single-mother families

TABLE 3 Binary=Ordinal Logistic Regression Coefficients of Single-Parent Families onParental Involvement at Home and in School

Model 1a Model 2 Model 3

Odds Ratio Y-std. coef. Odds Ratio Y-std. coef. Odds Ratio Y-std. coef.

Family structureb Discuss well-being in school (Ordinal)Two-parent Ref Ref RefSingle-mother 0.80�� �0.12 0.94 �0.03 0.92 �0.04Single-father 0.52��� �0.35 0.59��� �0.28 0.60�� �0.27Ad-hoc testc p< .01 p< .01 p< .01

Family structure Eating main meal (Ordinal)Two-parent Ref Ref RefSingle-mother 0.44��� �0.44 0.47��� �0.41 0.47��� �0.41Single-father 0.49��� �0.39 0.50��� �0.38 0.50��� �0.37Ad-hoc test n.s. n.s. n.s.

Family structure Spending time talking (Ordinal)Two-parent Ref Ref RefSingle-mother 0.85 �0.09 0.95 �0.03 0.94 �0.03Single-father 0.49��� �0.39 0.52��� �0.35 0.53��� �0.35Ad-hoc test p< .01 p< .01 p< .01

Family structure Homework help (Ordinal)Two-parent Ref Ref RefSingle-mother 0.73��� �0.17 0.83� �0.10 0.83� �0.10Single-father 1.12 0.06 1.29 0.14 1.29 0.14Ad-hoc test p< .05 p< .05 p< .05

Family structure Involvement at school (Binary)Two-parent Ref Ref RefSingle-mother 1.00 0.00 1.07 0.04 1.06 0.03Single-father 1.05 0.03 1.10 0.05 1.08 0.04Ad-hoc test n.s. n.s. n.s.

aModel 1: survey yearþ student’s gender, grade and birthplaceþmother’s and father’s birthplaceþco-residence with sibling and grandparents; Model 2: Model 1þparent’s education levelþmother’s and

father’s employment statusþ household wealth score; Model 3: Model 2þprivate=government aid

schoolsþ school-average household wealth score.bThe ‘‘others’’ category in family structure is included in the models but not shown in the table.cTest for the difference between coefficients of single-mother and single-father families.���p< .001; ��p< .01; �p< .05 (two-tailed).

32 A. K.-L. Cheung and H. Park

Page 20: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

were no longer significantly different in the frequency of discussing well-being with children than the parents from two-parent families. However,even after controlling for all students’ and parents’ covariates, single fatherswere still significantly less involved in discussing well-being with theirchildren, as compared with parents from two-parent families (p< .001) andsingle-mother families (p< .01). Results from Model 3 showed that singlefathers were still less involved than parents from two-parent families(p< .01) and single-mother families (p< .01), even after school characteris-tics were additionally controlled. Decomposition of y-standardized coeffici-ents showed that parental SES partially explains the differences in this itemof parental involvement between parents among types of families, whereasschool characteristics do not explain the difference.

Parents from two-parent families ate main meals with their childrenmore often than parents from single-mother and single-father families, evenafter controlling for all sets of factors. There was no significant difference inthe frequency of eating the main meal together between the single-motherand single-father families, with or without controlling the covariates.Decomposition of y-standardized coefficients showed that neither students’characteristics, household socioeconomic background, nor school character-istics substantially explained the differences between two-parent families andsingle-parent families. The disadvantages of single parenthood in this aspectof parental involvement may come from limited time-availability or otherunobserved characteristics instead.

Parents from two-parent families spent more time talking with theirchildren, as compared with parents from single-father families in all threemodels (p< .001). However, single mothers were not significantly differentfrom parents from two-parent families in spending time talking withchildren. Single fathers less frequently talked with their children as comparedwith single mothers (p< .01). Decomposition of y-standardized coefficientsshowed that students, parents, household, and school characteristics didnot substantially explain the differences in this aspect of parental involve-ment between two-parent and single-father families.

Different from other items of parental involvement at home, singlefathers were not significantly different from the parents from two-parentfamilies in providing homework help to their children. In contrast, parentsfrom single-mother families were less involved in providing homework helpthan parents from two-parent and single-father families in all three models(p< .05). Results of decomposing y-standardized coefficients showed thatthe difference in providing homework help between single-mother familiesand two-parent families was partially explained (41%) by the lower SES.

Compared with parental involvement activities at home, our results fromlogistic regression models for parental involvement in school showed thatfamily structure was not a significant factor for parental involvement inschool, regardless of whether controlling for other factors. Parents from

Single Parenthood in Hong Kong 33

Page 21: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

single-mother, single-father, and two-parent families were not significantlydifferent in this aspect of parental involvement in all three models. The cov-ariates in this study also did not explain parental involvement in school. Insummary, parents from two-parent families were more involved in somebut not all aspects of parental involvement.

We now discuss the results of differences in test scores by family type.Table 4 reports the regression coefficients of single-parent families on thethree academic scores. For all three tests, students from single-mother famil-ies were not significantly different from students from two-parent families.This finding is consistent with those of Chiu and Ho (2006) that students fromsingle-parent families and two-parent families fare similarly. However, stu-dents from single-father families performed significantly worse than studentsfrom two-parent families, with and without controlling for background char-acteristics and parental involvement. For students’ mathematics performance,students from single-father families scored 38 points and 30 points less thanstudents from two-parent and single-mother families, respectively, in thebaseline model (controlling for the survey year only). Roughly about 38%and 25% of the differences in mathematics test score between single-fatherfamilies and two-parent families and between single-father families andsingle-mother families, respectively, were explained by such backgroundcharacteristics as student, parent, household, and school (derived by com-paring the coefficients between Model 1 and Model 2: the coefficient forsingle-father families shrank from �38.03 to �23.56, a 31% reduction inthe coefficient). After controlling for the background covariates, about 18%and 19% of differences in mathematics test scores between single-father fam-ilies and two-parent families and between single-father and single-motherfamilies, respectively, were explained by the frequency of parental involve-ment at home (comparing the coefficients between Model 2 and Model 3;the coefficient for single-father families shrank from �23.56 to �19.36, an18% reduction in the coefficient). However, parental involvement in schooldid not explain the differences in mathematics test scores between single-father families, two-parent families, and single-mother families, controllingthe background characteristics (comparing coefficients between Model 2and Model 4).

Because parental involvement in school did not explain the differencesin mathematics test scores, the coefficients in Model 5 are similar to thosein Model 3. After controlling for all relevant covariates in the final model(Model 5), about 18% of the disadvantages in mathematics test scores for stu-dents from single-father families were attributable to the difference in par-ental involvement as compared with single-mother families and two-parentfamilies (comparing coefficients between Model 2 and Model 5). It is note-worthy that students from single-father families still scored about 19 pointsless than students from two-parent and single-mother families, even aftercontrolling for both background characteristics and parental involvement.

34 A. K.-L. Cheung and H. Park

Page 22: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

TA

BLE

4Regre

ssio

nCoeffic

ients

ofSi

ngle

-Par

entFam

ilie

son

Aca

dem

icSc

ore

s(M

athem

atic

s,Read

ing,an

dSc

ience

)

Model1

aM

odel2

Model3

Model4

Model5

Coeffic

ient

(SE)

Coeffic

ient

(SE)

Coeffic

ient

(SE)

Coeffic

ient

(SE)

Coeffic

ient

(SE)

Fam

ily

stru

cture

bM

athem

atic

ssc

ore

Tw

o-p

arent

Ref

Ref

Ref

Ref

Ref

Single

-moth

er

�7.6

8(4

.31)

�0.7

7(4

.40)

�0.8

2(4

.33)

�0.6

4(4

.38)

�0.6

9(4

.33)

Single

-fat

her

�38.0

3���

(8.0

8)

�23.5

6��

(7.7

1)

�19.3

6�

(7.7

9)

�23.4

2���

(7.6

3)

�19.3

1�

(7.7

1)

Ad-h

oc

test

cp<

.01

p<

.05

p<

.05

p<

.05

p<

.05

Fam

ily

stru

cture

Read

ing

score

Tw

o-p

arent

Ref

Ref

Ref

Ref

Ref

Single

-moth

er

�3.9

6(4

.06)

�0.7

6(4

.00)

�0.9

0(3

.93)

�0.6

3(3

.97)

�0.7

7(3

.92)

Single

-fat

her

�39.5

4���

(7.9

1)

�29.3

5���

(7.2

5)

�25.4

2���

(7.3

3)

�29.2

1���

(7.1

6)

�25.3

6���

(7.2

3)

Ad-h

oc

test

p<

.001

p<

.01

p<

.01

p<

.01

p<

.01

Fam

ily

stru

cture

Scie

nce

score

Tw

o-p

arent

Ref

Ref

Ref

Ref

Ref

Single

-moth

er

�5.6

0(3

.90)

0.1

8(4

.13)

0.3

6(4

.07)

0.3

0(4

.12)

0.4

7(4

.07)

Single

-fat

her

�36.6

7���

(7.9

9)

�26.4

3���

(7.5

7)

�22.2

7��

(7.6

9)

�26.3

0���

(7.5

0)

�22.2

2��

(7.6

2)

Ad-h

oc

test

p<

.01

p<

.01

p<

.05

p<

.01

p<

.05

aM

odel

1:fa

mily

stru

ctureþ

surv

ey

year

;M

odel

2:M

odel

students

,par

ents

,house

hold

and

schoolch

arac

terist

ics;

Model3

:Model2þ

par

enta

linvolv

em

entat

hom

e;

Model4:M

odel2þ

par

enta

lin

volv

em

entat

school;

Model

5:M

odel

par

enta

lin

volv

em

entat

hom

ean

dsc

hool.

bThe

‘‘oth

ers

’’ca

tegory

infa

mily

stru

cture

isin

cluded

inth

em

odel

sbutnotsh

ow

nin

the

table

.cTest

for

the

diffe

rence

betw

een

single

-moth

er

and

single

-fat

her

fam

ilie

s.��� p<

.001

;��

p<

.01;� p<

.05

(tw

o-tai

led).

35

Page 23: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

The patterns on the differences in test scores by family structure were almostthe same for the reading and science performance.

DISCUSSION

We found that parents in single-parent families are more disadvantaged insocioeconomic conditions and are less involved in some aspects of parentalinvolvement at home than parents in two-parent families. Single mothers didnot show much difference from parents in two-parent families in two aspectsof parental involvement that require more verbal communication andemotional attachment—discussing well-being and spending time talking. Incontrast, single fathers were less involved in these two aspects of parentalinvolvement than single mothers and parents in two-parent families. Thismay be related to the gendered aspects of these two items of parentalinvolvement. In contrast, single mothers were less involved in providinghomework help than other parents, partly because of parental SES back-ground. Both single mothers and single fathers were less involved in eatingmain meals with children regardless of SES background. This may be due tothe limited opportunities because of the absence of one parent in these fam-ilies. Compared with parental involvement activities at home that differ byfamily structure, two-parent families did not have advantages over single-parent families in parental involvement in school.

Consistent with Chiu and Ho’s (2006) finding, we did not find asignificant negative effect of single motherhood on students’ academicperformance, even before controlling for parental involvement and socio-economic background. However, we found that the negative effect of singlefatherhood on students’ academic performance was significant for all threetest scores. In addition, the effects were partially mediated through socio-economic background and parental involvement. Hence, single parenthoodstill matters in students’ educational outcome in Hong Kong, but the genderof the single parents has to be considered. Although single-father families arestill a relatively small group among Hong Kong families—38 points less in thetest scores— this indicates about 40% of the standard deviation in theinternational distribution of the test scores and therefore presents a moderateto strong disadvantage that requires attention from policymaking and thescholarly community.

Taking the context into account, it is not very surprising that singlefathers are less involved in communicating with their children than singlemothers. Traditional gender roles are still expected and performed in HongKong (Choi & Ting, 2009). Men’s self-evaluation is still largely based ontheir ability to provide financial support to the home (Choi et al., 2012b).They may have less time available and are more self-restrained to talk andexpress affection to their children. Therefore, in terms of more instrumental

36 A. K.-L. Cheung and H. Park

Page 24: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

interactions such as providing homework help, single fathers are no worsethan parents from two-parent families. On the other hand, single motherscommunicate with children as much as parents from two-parent families,possibly because providing emotional care to the children is the expectedresponsibility of mothers. These differences in parenting behaviorsbetween the types of families have implications on students’ educationaloutcomes.

Previously, when Chiu and Ho (2006) discussed the possible reasons forthe absence of the negative effect of single parenthood on students’ academ-ic performance, they suggested that the equal pupil funding policy mayweaken the effect of family socioeconomic background. In addition, theyalso suggested that the strong family ties in Hong Kong and close geographi-cal proximity with other family members may reduce the negative impacts ofthe broken family structure. Although these explanations are reasonable,more detailed arguments are needed to explain why there is only anabsence of effect on single motherhood but presence of effect of singlefatherhood. In addition, the lower levels of parental involvement forsingle-father families only provide a partial answer. Indeed, the percentageof difference explained by parental involvement is relatively small. Otherthan the parents’ socioeconomic background and parental involvement,about two-thirds of the differences in test scores between single-fatherfamilies and two-parent families remained unexplained by the covariatesincluded in this study.

In regard to the differential effects of single motherhood and singlefatherhood, we consider two possible reasons. First, in addition to thefrequency of parental involvement measured in this study, perhaps parentingstyles and quality of parent–child relations are also gendered and single-father families are more disadvantaged in these aspects of parent–childinteraction, therefore leading to poorer educational outcomes for children.Second, single-father families are rare and therefore are a more selectivegroup of families. Considering that custody is usually granted to mothersin Hong Kong, single-father families may have gone through a severe conflictprocess during the divorce or the mothers may have given up the opport-unity to get custody, which may have caused negative emotions for thechildren, hence affecting their educational outcomes negatively. Furtherempirical investigations are required to confirm these speculations. Nonethe-less, more attention has to be paid to the distinction between single-fatherand single-mother families in the future.

FUNDING

Hyunjoon Park acknowledges support from the Academy of Korean StudiesGrant funded by the Korean Government (MEST; AKS-2010-DZZ-2101).

Single Parenthood in Hong Kong 37

Page 25: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

NOTES

1. Specifically, compared with PISA 2009, PISA 2012 had one more category of ‘‘once or twice a year’’

between the category of ‘‘never or hardly ever’’ and the category of ‘‘once or twice a month.’’ We com-

bined these two lowest categories in PISA 2012 to produce a single category of ‘‘less than once a month.’’

We compared the distribution of students across the four categories between PISA 2009 and 2012 data and

found the distribution to be similar between PISA 2009 and 2012.

2. Notably, there is a slight discrepancy of wording on the item ‘‘homework help’’ in the two waves of

PISA. In 2009 the item refers to ‘‘help your child with his=her homework,’’ whereas the question wording

changed to ‘‘help my child with his=her mathematics homework.’’ We examined the items separately for

the two waves. The general pattern of parental involvement in homework is similar between the two

waves. In this article, we show the results with the combined sample.

3. The number of siblings in the household is not available for PISA 2012 data. Therefore, we used the

variable of whether the student lived together with any sibling, which is available for both PISA 2009 and 2012.

REFERENCES

Astone, N. M., & McLanahan, S. S. (1991). Family structure, parental practices andhigh school completion. American Sociological Review, 56, 309–320.

Biblarz, T. J., & Stacey, J. (2010). How does the gender of parents matter? Journal ofMarriage and Family, 72, 3–22.

Breivik, K., Olweus, D., & Endresen, I. (2009). Does the quality of parent-childrelationships mediate the increased risk for antisocial behvaior and substanceuse among adolescents in single-mother and single-father families? Journalof Divorce and Remarriage, 50, 400–426.

Cancian, M., & Meyer, D. R. (1998). Who gets custody. Demography, 35, 147–157.Chiu, M. M., & Ho, E. S. C. (2006). Family effects on student achievement in Hong

Kong. Asia-Pacific Journal of Education, 26, 21–35.Choi, S. Y. P., Cheung, Y. W., & Cheung, A. K. L. (2012a). Social isolation and

spousal violence: Comparing female marriage migrants with local women.Journal of Marriage and Family, 74, 444–461.

Choi, S. Y. P., Au, W. W. T, Wong, A. W. C., Liong, M. C. C., Wong, M. F. Y., Lo, S. K.W., & Chao, K. C. (2012b). Research report: Exploratory study on genderstereotyping and its impacts on male gender. Hong Kong: Equal OpportunitiesCommission.

Choi, S. Y. P., & Ting, K. F. (2009). A gender perspective on families in Hong Kong.In F. M. Cheung & E. Holroyd (Eds.), Mainstreaming gender in Hong Kongsociety (pp. 159–180). Hong Kong: The Chinese University Press.

Chow, N., & Lum, T. (2008). Trends in family attitudes and values in Hong Kong.Hong Kong: The University of Hong Kong.

Creighton, M. J., Park, H., & Teruel, G. M. (2009). The role of migration and singlemotherhood in upper secondary education in Mexico. Journal of Marriage andFamily, 71, 1325–1339.

Dommaraju, P., & Jones, G. W. (2011). Divorce trends in Asia. Asian Journal ofSocial Science, 39, 725–750.

Downey, D. B. (1994). The school performance of children from single-mother andsingle-father families: Economic or interpersonal deprivation. Journal of FamilyIssues, 15, 129–147.

38 A. K.-L. Cheung and H. Park

Page 26: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

Downey, D. B. (2002). Parental and family involvement in education. In A. Molnar(Ed.), School reform proposals: The research evidence (pp. 113–134).Greenwich, CT: Information Age.

Downey, D. B., Ainsworth-Darnell, J. W., & Dufur, M. J. (1998). Sex of parent andchildren’s well-being in single-parent households. Journal of Marriage andFamily, 60, 878–893.

Dronkers, J. (1994). The changing effects of lone parent families on the educationalattainment of their children in a European welfare state. Sociology, 28,171–191.

Dufur, M. J., Howell, N. C., Downey, D. B., Ainsworth, J. W., & Lapray, A. J. (2010).Sex differences in parenting behaviors in single-mother and single-fatherhouseholds. Journal of Marriage and Family, 72, 1092–1106.

Hampden-Thompson, G. (2013). Family policy, family structure, and children’seducational achievement. Social Science Research, 42, 804–817.

Hawkins, D. N., Amato, P. R., & King, V. (2006). Parent-adolescent involvement: Therelative influence of parent gender and residence. Journal of Marriage andFamily, 68, 125–136.

HKPISA Centre. (2011). The fourth HKPISA report PISA 2009. Hong Kong: HKPISACentre, The Chinese University of Hong Kong.

HKPISA Centre. (2013). HKPISA 2012 preliminary report. Hong Kong: HKPISACentre, The Chinese University of Hong Kong.

Ho, E. S. C., & Willms, J. D. (1996). Effects of parental involvement on eighth-gradeachievement. Sociology of Education, 69, 126–141.

Hong Kong Census and Statistics Department. (2013). 2011 population censusthematic report: Single parents. Hong Kong: Hong Kong SAR Government.

Jones, G. W. (2012). Changing family sizes, structures and functions in Asia.Asia-Pacific Population Journal, 27, 83–102.

Kung, W. W., Hung, S. L., & Chan, C. L. W. (2004). How the socio-cultural contextshapes women’s divorce experience in Hong Kong. Journal of ComparativeFamily Studies, 35, 33–50.

Lee, M. Y., & Law, C. K. (1994). Child care practice of single parent families in HongKong. International Journal of Sociology of the Family, 24, 45–59.

Marks, G. N. (2006). Family size, family type and student achievement:Cross-national differences and the role of socioecnomic and school factors.Journal of Comparative Family Studies, 37, 1–24.

McLanahan, S. (2004). Diverging destines: How children are faring under the seconddemographic transition. Demography, 41, 607–627.

McLanahan, S., & Sandefur, G. (1994). Growing up with a single parent: What helps,what hurts. Cambridge, MA: Harvard University Press.

McNeal, R. B., Jr. (1999). Parental involvement as social capital: Differential effective-ness on science achievement, truancy, and dropping out. Social Forces, 78,117–144.

Mood, C. (2010). Logistic regression: Why we cannot do what we think we can do,and what we can do about it. European Sociological Review, 26, 67–82.

Musick, K., & Meier, A. (2012). Assessing causality and persistence in associationsbetween family dinners and adolescent well-being. Journal of Marriage andFamily, 74, 476–493.

Single Parenthood in Hong Kong 39

Page 27: Adam Ka-Lok Cheung & Hyunjoon Park...single-father and single-mother families into one category of single-parent families, but relatively little is known about whether the absence

Ono, H. (2009). Husbands’ and wives’ education and divorce in the United Statesand Japan. Journal of Family History, 34, 292–322.

Park, H. (2007). Single parenthood and children’s reading performance in Asia.Journal of Marriage and Family, 69, 863–877.

Park, H. (2008a). Effects of single parenthood on educational aspiration and studentdisengagement in Korea. Demographic Research, 18, 377–408.

Park, H. (2008b). The varied educational effects of parent-child communication:A comparative study of fourteen countries. Comparative Education Review,52, 219–243.

Park, H. (2014). Single parenthood and children’s education in Republic of Korea:An update. In H. Park & K. Kim (Ed.), Korean education in changing economicand demographic contexts (pp. 153–171). Dordrecht, Netherlands: Springer.

Park, H., Byun, S. Y., & Kim, K. K. (2011). Parental involvement and students’cognitive outcomes in Korea: Focusing on private tutoring. Sociology ofEducation, 84, 3–22.

Park, H., & Raymo, J. M. (2013). Divorce in Korea: Trends and educational differen-tials. Journal of Marriage and Family, 75, 110–126.

Pong, S. L., Dronkers, J., & Hampden-Thompson, G. (2003). Family policies andchildren’s school achievement in single- versus two-parent families. Journalof Marriage and Family, 65, 681–699.

Raymo, J. M., Iwasawa, M., & Bumpass, L. (2004). Marital dissolution in Japan:Recent trends and patterns. Demographic Research, 11, 395–420.

Raymo, J. M., Park, H., Iwasawa, M., & Zhou, Y. (2014). Single motherhood, livingarrangements, and time with children in Japan. Journal of Marriage and Family,76, 843–861.

Sullivan, P. L. (2005). Culture, divorce, and family mediation in Hong Kong. FamilyCourt Review, 43, 109–123.

Wang, D. B. (2004). Family background factors and mathematics success: A compari-son of Chinese and US students. International Journal of Educational Research,41, 40–54.

Winship, C., & Mare, R. D. (1984). Regression models with ordinal variables.American Sociological Review, 49, 512–525.

Women’s Commission. (2011). What do women and men in Hong Kong think aboutthe status of women at home, work and in social environments? Highlights ofsurvey findings. Hong Kong: Women’s Commission.

Yodanis, C. (2005). Divorce culture and marital gender equality: A cross-nationalstudy. Gender & Society, 19, 644–659.

Zhu, Y., & Leung, F. K. S. (2011). Mathematics achievement of mainland immigrantstudents in Hong Kong. Asia Pacific Journal of Education, 31, 471–485.

40 A. K.-L. Cheung and H. Park