Abstract
This study explores family formation–work fit (i.e., beliefs that work demands fit with desired family size) and its relation to achievement of desired family size, work–family facilitation and conflict, and marital satisfaction using a subsample (N = 690) drawn from the nationally representative Singapore National Study of Work–Life Harmony. Results indicate that family formation–work fit was positively related to desired family size achievement. Importantly, those with desired family size achievement had more than twice as many children as those who had not achieved their desired family size. In addition, workplace schedule flexibility, work-to-family facilitation, and marital satisfaction were associated with higher family formation–work fit, while family-to-work conflict and work-to-family conflict were associated with lower family formation–work fit. These findings indicate the importance of considering family formation–work fit when studying family fertility decisions, particularly in nations with low fertility rates.
Many countries in the developed world are dealing with potentially serious problems associated with declining fertility. More than half of the world’s nations have total fertility rates (TFRs; number of births per woman over her lifetime) below replacement level (CIA World Factbook, 2012). In particular, Singapore has one of the lowest fertility rates in the world (1.20 children per woman; Department of Statistics Singapore, 2012)—well below the population replacement rate of 2.10 children per woman. This concerns policy makers because prolonged subreplacement fertility rates lead to depopulated nations with lower economic output (Matsutani, 2006) and potential intergenerational family conflict over increased economic and care demands (Meyer, 2009). Fertility decisions in developed nations, where health standards are high and working hours are long, are a matter of choice and timing rather than personal health and ability to have children. This choice and timing is likely influenced by not only the home environment but the work environment as well (Sun, 2009). Thus, the study of work–family interactions has the potential for expanding our understanding of fertility decisions made by working couples in developed nations.
The work–family literature is immense. Thousands of articles are published annually from a variety of perspectives and fields, including “psychology, sociology, family studies, economics, communication, organizational behavior, law, political science, social work, and public policy” (Perry-Jenkins & MacDermid, 2013, p. 381). These perspectives consider topics ranging from “the effects of daily work stressors on later family behavior to macro-level studies that examine the ways in which different social norms across countries shape workers commitment to family and employment” and address “societal problems such as unemployment, underemployment, recruitment and retention, work-life stress, and productivity, as well as related topics such as healthcare and childcare” (p. 381). For the purpose of this study, work–family perspectives that specifically focus on how work environments affect family decisions (i.e., a work socialization perspective) and perspectives that focus on how the work and family spheres positively and negatively interact with each other (i.e., a role theory perspective) provide a clear line of research for studying how family fertility decisions are influenced by family interactions with the workplace (see Perry-Jenkins & MacDermid, 2013). One particular aspect of work–family role theory, the person–environment (P-E) fit literature, is particularly important when studying fertility decisions. P-E fit stipulates that a family experiencing a workplace environment will assess whether that environment is beneficial or detrimental to the family’s desires and values (Edwards & Rothbard, 2005). A beneficial assessment (e.g., good fit) of the work environment facilitates family functioning and decision making, while a detrimental assessment (e.g., bad fit) fosters stress that families must cope with or suffer from.
Work socialization, role theory, and P-E fit perspectives provide a theoretical basis for how family fertility choices are influenced by the working environment, but these perspectives have yet to be empirically tested in the area of family fertility decisions. We use these three perspectives to guide the study of how one’s employment situation fits with one’s desired family size, a concept we define as family formation–work fit. Since children are often viewed as economic and career liabilities (Day, 2003), a couple’s sense of fit between workplace demands and family desires may influence decisions to delay or even avoid childbearing in order to cope with the stress associated with this lack of fit. Because of its history of low fertility rates and long work hours for both spouses (Hill, 2005; Jones et al., 2008), and more recent government policies to encourage pronatalist work environments (Sun, 2009), Singapore provides an ideal context for exploring the concept of family formation–work fit and its connection to fertility outcomes.
Accordingly, the purpose of this study is to reify the construct of family formation–work fit and explore its relation to achievement of desired family size in Singapore. To study this phenomenon, we examine 690 dual-earner individuals drawn from a nationally representative sample of Singaporean workers to test a conceptual model based on three work–family perspectives (work socialization, role theory, and P-E fit). The study of family formation–work fit extends scholarship on work–family interactions to include family formation decisions and has the potential to further inform the literature on fertility rates in developed nations who face the prospect of depopulation. A specific focus on Singapore also answers calls to explore greater sociocultural contexts in order to expand and adjust work–family theory as needed (Perry-Jenkins & MacDermid, 2013).
Review of Literature
National Context of Singapore
Singapore provides a compelling context in which to explore work–family interactions related to fertility. It is an exemplar of the prosperity to which many nations aspire but may also be a harbinger of work–life challenges that successful globalized economies might expect (Sassen, 2001). Despite a diverse society (i.e., 74% Chinese, 13% Malays, 9% Indians) with high health and income standards, Singapore currently has an alarming “ultra-low” fertility rate (Department of Statistics Singapore, 2012). Most households are dual-earner households, report long work hours, and experience little workplace flexibility (Hill, 2007).
Singapore offers a unique policy and cultural history that demonstrates the influence of public mandates, culture, and social location on family size decisions. In 1973, population control laws were enacted to promote lower fertility rates (Anderson, 2004). Because the birthrate plummeted far below replacement levels, Singapore has since undertaken a pronatalist approach in order to avoid an economically deleterious dependency ratio (i.e., the ratio of productive workers to dependent youth or elders; Harwood, Sayer, & Hirschfeld, 2004). Presently, Singaporeans are encouraged to have more children through financial incentives, paid paternity and maternity leave, flexible work hours, and unpaid childcare leave (Sun, 2009). In his review of East Asian fertility research, Jones (2012) suggests that fertility policies in Singapore have promoted fertility rates that are at least higher than in comparable East Asian countries (e.g., Japan, South Korea, Taiwan, etc.), but overarching trends in increased singlehood, delays in marriage, and less fertile marriages remain obstacles to further improving Singaporean fertility rates. The current study evaluates Jones’s assertions and expands critical thought on the potential impact of public (workplace) policies on fertility outcomes.
Though supportive work–family policies may exist, they may not be used; the P-E fit literature suggests this could be explained by a lack of fit between these workplace environments and the fertility needs of families (Edwards & Rothbard, 2005). Sun’s (2009) qualitative investigation supports this with findings on workers feeling anxious about using fertilely-friendly policies, and some even fearing pregnancy would lead to job termination. Moreover, Sun (2012) reports difficulty in policy makers’ attempts to frame, communicate, distribute, and otherwise promote pronatalist policies and programs. Policy makers and workers’ experiences highlight the importance of understanding family formation–work fit in Singaporean workers.
Development of a Conceptual Model for Family Formation–Work Fit
Work–family research is so broad and diverse that it is better conceptualized as a collection of theories and perspectives rather than one grand theory (Perry-Jenkins & MacDermid, 2013). Role theory, with its focus on work-to-family (WTF) and family-to-work (FTW) conflict (and to a lesser extent facilitation) paired with the P-E fit framework lend themselves well to our interest in how the work and family environments interact with family fertility decisions. Additionally, with workplace policy being so central to the promotion of fertility (Sun, 2009), the work socialization perspective can add to this investigation with its focus on how the work environment influences family life. We used these three perspectives to develop a conceptual model that incorporates the construct of family formation–work fit (see Figure 1).

Hypothesized model.
Role Theory and P-E Fit
Both role theory and P-E fit focus on how interactions between work and family spheres can have facilitative (i.e., positive) or conflictual (i.e., negative) implications for families. The central concept of work–family role theory is that the strain of one role (or the benefits obtained from one role) can spillover (i.e., influence) the other role (Frone, 2003). Role theory proposes that interactions between work and family are not only bidirectional (i.e., work effects family and family effects work) but that the outcomes of these interactions are independently positive or negative (Voydanoff, 2008). In other words, conflict and facilitation are independent constructs (i.e., limiting conflict does not necessary result in increased facilitation). Models based on role theory account for this bidirectionality by creating independent global assessments of conflict and facilitation for both WTF and FTW effects (see Galinsky, Aumann, & Bond, 2009). The P-E fit framework expands role theory to incorporate the concept of fit. Work–family fit is found when workers feel that they can simultaneously achieve the professional and the family life they desire. Fit is influenced by both the conflict and facilitation fostered from work–family interactions; in turn, degree of fit influences family well-being (Edwards & Rothbard, 2005; Voydanoff, 2008). When fit is high, families desires are met. When fit is low, families will often use coping mechanisms to compensate for the stress created by the lack of fit (Voydanoff, 2008). For the purpose of our study, it is possible that work–family conflict as well as facilitation may influence perceptions of family formation–work fit and the degree of fit is likely to be related to achievement of desired family size. Those with good fit are likely to achieve the number of children they desire, while those with a lack of fit are likely to delay or forgo childbearing in order compensate for the lack of fit.
Work Socialization Perspective
The central concept of the work socialization perspective is that work life has a direct effect on family decisions. While this perspective is less used than those of role theory and P-E fit, it is particularly important when studying fertility decisions in regions with explicit fertility policy. For example, the government of Singapore has implemented multiple policies in order to promote fertility (e.g., workplace flexibility), in the hope that they will increase fertility rates (Sun, 2009). However, overtime hours are common in the working context of Singapore. Overtime hours have been found to promote a variety of positive and negative outcomes in Singaporean families (Hill, 2007). This combination of workplace policy and expectations are important to consider when building a model that tests family formation–work fit in Singaporean workers.
Conceptual Framework of Family Formation–Work Fit
Voydanoff’s (2007) work provides an empirical example of combining of work socialization, role theory, and P-E fit perspectives. In Voydanoff’s model, work factors influence the role theory concepts of conflict and facilitation. This conflict or facilitation is then modeled to influence the P-E perspective concept of fit, but not the concept of family–formation work fit specifically. Guided by Voydanoff’s model and informed by the relevant work socialization, role theory, and P-E fit literature, we identified three research questions to test hypothesized relations on how work–family interactions influence family formation decisions. The first question focuses on variables identified by the workplace socialization perspective: (1) How do workplace characteristics, such as schedule flexibility and overtime, influence FTW and WTF conflict and facilitation, and workers’ sense of family formation–work fit in Singapore? Our second question focuses on the role theory and P-E fit perspectives: (2) How do FTW and WTF conflict and facilitation influence family formation–work fit? Our last question focuses on the validity of the family formation–work fit as part of the work–family literature and eventual relation to family formation: (3) What is the relation between FTW and WTF conflict and facilitation, family formation–work fit, and the actual family formation of workers in Singapore? The identifying letters for each hypothesis presented below correspond to specific paths in our hypothesized model (see Figure 1).
Research Question 1: Work Socialization
Our first research question focuses on the theoretical influence of workplace characteristics (e.g., overtime hours, schedule flexibility) on FTW and WTF conflict and facilitation in Singaporean workers. The amount of time a worker spends in the workplace, particularly the amount of noncontracted hours (i.e., overtime), has a complicated relation with work–family interactions. Overtime involves working beyond what is expected and can be a source of WTF conflict in some families (see Hill, 2005); however, other studies have found overtime to foster greater FTW facilitation (Karimi & Nouri, 2009). This is possibly because of the benefits of greater income that extra hours provide.
In addition to overtime, work socialization suggests that schedule flexibility should be related to WTF/FTW conflict and facilitation as well. In previous work, schedule flexibility was related to both FTW and WTF facilitation and conflict (Hill, 2005; Hill, Erickson, Holmes, & Ferris, 2010); however, recent work with international samples has found regional differences. Masuda et al. (2011) found that schedule flexibility was related to WTF conflict only for Anglos and not for Asians. Thus, for a Singaporean sample, we are unsure whether WTF or FTW conflict will be related to workplace schedule flexibility.
Based on the work socialization perspective and relevant research, we hypothesize (see corresponding paths in Figure 1) that (a) overtime will be related to an increase in both WTF/FTW conflict and facilitation and (b) schedule flexibility will be related to an increase in WTF/FTW facilitation and possibly reduced WTF/FTW conflict.
Research Question 2: Role Conflict, Role Facilitation, and Family Formation–Work Fit
In our next research question, we sought to understand how FTW/WTF conflict and facilitation influence family formation–work fit. Conceptually, role theory suggests that all work–family interaction variables should influence family formation–work fit. Additionally, workplace schedule flexibility, a workplace variable, has been found to be a powerful predictor of family life in previous work and has the potential to be associated with family formation–work fit (Hill, 2005).
Our particular interest in family formation–work fit requires a slight modification to Voydanoff’s (2007) model. Specifically, life satisfaction has been linked the desire to have more children (Li, Patel, Balliet, Tov, & Scollon, 2010). The desire to have more children is an important element of family formation–work fit. Thus, marital satisfaction, a family sphere variable, likely conceptually affects family formation–work fit rather than family formation–work fit affecting marital satisfaction. For the particular type of fit we investigate, work–family interaction variables are more likely to directly and independently affect both marital satisfaction and fit. Therefore, we conceptualized marital satisfaction (an element of life satisfaction) as being directly affected by WTF and FTW conflict and facilitation. Similar to Li et al. (2010)—and given the particular type of work–family fit we are exploring—we conceptualize marital satisfaction as affecting family formation–work fit rather than the reverse.
Multiple meta-analyses of American and international workers link FTW conflict and facilitation with marital satisfaction (Amstad, Meier, Fasel, Elfering, & Semmer, 2011; cf. Galovan et al., 2010). Studies connecting work–family facilitation to marital satisfaction are sparse; however, in a U.S. sample, Hill (2005) found that FTW facilitation was significantly and positively associated with marital satisfaction but WTF facilitation was not. To our knowledge, no other studies have examined this relation and so we are unsure if this effect will be replicated with an international sample. Additionally, important Singaporean family and workplace characteristics that have established relations in role theory (specifically FTW/WTF conflict and facilitation) and need to be controlled for include age, income, gender, and number of children living in the home (Galovan et al., 2010; Hill, 2005; Hill et al., 2004).The most important control variable, however, might be for depressive symptoms. Depressive symptoms have been associated with most family and workplace variables (see Repetti, Taylor, & Seeman, 2002) and have also been previously controlled for in Singaporean samples (Galovan et al., 2010).
In considering the potential association between work–family interactions, marital satisfaction, and family formation–work fit, we offer the following hypotheses (see Figure 1): (c) FTW and WTF conflict will be negatively related to family formation–work fit while FTW and WTF facilitation and workplace schedule flexibility will be positively related to family formation–work fit; (d) FTW and WTF conflict will be related to a decrease in marital satisfaction while FTW and WTF facilitation may be related to increases in marital satisfaction; and (e) Marital satisfaction will have a positive relation with family formation–work fit and achievement of desired family size.
Research Question 3: The Construct of Family Formation–Work Fit
In our final research question, we sought to determine how our concept of family formation–work fit is related to the actual family formation of workers in Singapore. While an attempt has been made to identify discrete predictors of desired and actual family size in Singapore (see Call, Sheffield, Trail, Yoshida, & Hill, 2008), the present study is the first known attempt to explore those relations in a comprehensive model that includes the concept of family formation–work fit, desired family size achievement, and individuals’ plans to have (more) children. In considering potential influences on family formation–work fit and actual family formation, we posit the following hypothesis (see Figure 1): (f) Family formation–work fit will be positively related to desired family size achievement, which in turn will influence individuals’ plans to have children.
Method
Sample and Procedures
For this study, we used a nationally representative sample of employed persons drawn from the 2006 Singapore National Study of Work–Life Harmony (see Hill, 2007). The original sample consisted of 1,601 employed persons in Singapore selected from a pool of respondents previously recruited by a respected independent survey company contracted by the government of Singapore. Previous to this survey, the company recruited a diverse pool of respondents who had agreed to participate in a number of surveys. For the current project, they selected 1,601 respondents from this pool as a representative sample of workers in Singapore (according to the Singapore Department of Statistics; see Hill, 2007) based on gender, age, race, and house type (their proxy for socioeconomic status [SES]). Because participants were already available and had previously agreed to participate in surveys, the response rate for this study was 100% (see Sandberg, Yorgason, Miller, & Hill, 2012). As our focus was to determine how workers balance their work and family life and the connection to attitudes about fertility—and because work–family issues are particularly salient in dual-earner households—only workers who were married, employed, and whose spouses were employed were included in the analysis. Those not married were excluded from our analysis because, in Singapore, having children outside of marriage is not generally culturally acceptable (Teo, 2005), and including their data would confound the meaning of the responses for this particular analysis. In focusing only on dual-earner couples, we acknowledge that work–family dynamics may have influenced partners who were previously in the workforce to exit the workforce following the birth of a child.
The final sample consisted of 690 employed and married men (n = 264) and women (n = 426). In narrowing the sample, 495 never married, 10 separated, 42 divorced, and 6 widowed individuals were excluded. Another 313 married and 4 remarried individuals whose spouse did not work were also excluded. Additionally—as the focus of this study was to determine influences on achievement of desired family size—we removed 28 individuals who had more children then they desired. Compared with the full sample, the final dual-earner sample reported higher household income and had more children (MFull = 1.16, SDFull = 1.19; MDE = 1.56, SDDE = 1.02). The final sample did not differ significantly from the full sample in age, education, or individual respondent income.
The surveys were administered as a one-time face-to-face interview (see Galovan et al., 2010, for information regarding survey development). During the questionnaire, 155 potential items were read aloud to participants in English, the official language of Singapore. Each interview lasted 35 to 45 minutes. The participants included in this study ranged in age from 21 to 65 years, with a mean age of 39.01 years (SD = 8.68 years). A plurality of participants (34.8%) reported secondary education (equivalent to high school education). Participants’ median education was upper secondary education (equivalent to some post–high school education). Just over one eighth of the sample (14.1%) had a college or graduate degree. The mean annual family income was US$35,640 (SD = $18,413, median $31,658), measured in categorical ranges from $1,000 to $10,000 or more per month (Singapore Dollars). Similar to the full sample and the Singaporean population, the ethnicity of the final sample was 70% Chinese, 22% Malay, 7% Indian, and 2% other.
Measures
For multi-item scales, we report the omega (ω) reliability coefficient. The omega reliability coefficient is superior to Cronbach’s alpha, as it does not assume that individual items contribute equally to the construct (i.e., tau-equivalence), and is interpreted in the familiar 0 to 1 range (McDonald, 1999). It is derived by squaring the correlation between the scale and the true score.
Family Formation Achievement and Plans for Child Bearing
We used two items to measure aspects of participants’ family formation. The first item was desired family size achievement. Participants responded to a question about desired family size (i.e., “What is your best guess for how many children you will have when your family is complete?”). Achievement of desired family size was derived by subtracting the number of children respondents indicated they had from the number they indicated they desired. Responses for those who had achieved their desired family size were coded as 1 and those who had not were coded as zero. The second item, plans to have children, was given in response to the question, “Do you plan to have children or more children in the future?” Responses were also given dichotomously, with a response of yes scored as 1 and no scored as zero.
Family Formation–Work Fit
As a proxy for cognitions of family formation–work fit, we used a single item (i.e., “My job fits well with my desired number of children”) with responses given on a 4-point Likert-type scale from strongly disagree to strongly agree. This is a particularly descriptive single item that fits with the specific cognition we are attempting to measure. While single-item measures are a source of debate in the field, previous work has found that job satisfaction can be validly assed using single-item measures when multi-item scales are not feasible (Dolbier, Webster, McCalister, Mallon, & Steinhardt, 2005). Our measure corresponds to a specific aspect of job satisfaction and has been validated in previous studies of work–family fit (Call et al., 2008). Due to the exploratory nature of our work and the valid previous use of this item, we believe a single item is acceptable until a better scale becomes available.
Work–Family Spillover Variables
The four work–family spillover scales were developed for the 1992 National Study of the Changing Workforce (NSCW; see Galinsky et al., 2009). For each of the scales, respondents were asked to consider their experience “during the past three months.” For each of the facilitation scales and for each of the conflict scales, similarly worded items were used. FTW facilitation is a 4-item measure with questions such as “How often have you been in a better mood at work because of your family life?” and “How often have you had more energy to do your job because of your family life?” WTF facilitation is a 4-item measure with questions such as “How often has your job helped you gain insights that have helped your family life?” and “How often has work helped you do your family-related activities better at home than you would have otherwise?” FTW conflict is a 5-item measure with questions such as “How often have you not had enough time for your job because of your family life?” and “How often has your family life drained you of the energy you needed to do your job?” WTF conflict is a 5-item measure with questions such as “How often has work kept you from doing as good a job at home as you could?” and “How often has your job kept you from concentrating on important things in your personal or family life?” Each of the spillover scales used a 4-point Likert-type scale (rarely/never, sometimes, often, very often), and each of the scales demonstrated adequate reliability (ω ranging from .88 to .92). Items were coded such that higher scores denote more conflict or facilitation.
Workplace, Family, and Individual Context Variables
Workplace schedule flexibility
Workplace flexibility was measured using a single-item measure (i.e., “Overall, how much control [flexibility] would you say you have in scheduling when you work [your work hours]?”) using a 4-point Likert-type scale (very little/none, some, a lot, complete) that has been used previously (Jones et al., 2008).
Marital satisfaction
Marital satisfaction was measured with a 4-item measure. Respondents were asked to rate how satisfied they were with various aspects of their marriage (e.g., “The quality of your communication together,” “The fun you have together (e.g., dates together, recreation),” and “Your overall relationship with your spouse”) with responses given on a 4-point Likert-type scale (not too satisfied to extremely satisfied). To avoid inflating any associations with marital satisfaction, a fifth item related to work and family role division was removed from the scale. The scale demonstrated adequate reliability (ω = .93) and has been used previously (Hill et al., 2008).
Based on previous literature and theory, overtime hours and hours worked per week were included as single-item variables (Karimi & Nouri, 2009; Voydanoff, 2007). Overtime is much more valuable to families with fewer work hours (Karimi & Nouri, 2009); thus, hours worked per week was correlated with overtime hours to control for this impact. We also included the following control variables: depressive symptoms, a 7-item measure based on the Center for Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977; reliability ω = .92 for the current sample); household income (9 income groups ranging from less than $1,000 to over $10,000 Singapore dollars per month); gender of respondent; and number of children in the household (Galovan et al., 2010).
Results
Descriptive Results
As seen in Table 1, on average, participants worked 46.85 hours per week, while working 3.30 hours of overtime. They had between none and some schedule flexibility (M = 1.78) and sometimes experienced depressive symptoms (M = 2.19). Participants sometimes experienced FTW facilitation (M = 2.09), while average scores for WTF facilitation (M = 1.87), WTF conflict (M = 1.84), and FTW conflict (M = 1.46) were between rarely and sometimes. They were somewhat to very satisfied with their marriage (M = 2.66). Participants scored in the disagree to agree range in family formation–work fit (M = 2.85). Just more than half (55%) had achieved their desired family size, while about a third (34%) planned to have more children.
Descriptive Statistics and Correlations Between Study Variables.
Note. N = 690.
p < .10. *p < .05. **p < .01.
Structural Equation Model Analysis
We constructed a structural equation model (SEM) to examine influences on marital satisfaction, family formation–work fit, achievement of desired family size, and plans to have (more) children (see Figure 1). We employed full information maximum likelihood (FIML) estimation to account for missing data. All correlation coefficients (see Table 1) were less than the standard multicollinearity cutoff of .85 (Kline, 2005). In the initial model, we included paths from all endogenous variables to the work–family spillover variables and both marital satisfaction and family formation–work fit; we also included paths from each of the spillover variables to both marital satisfaction and family formation–work fit. We controlled for age, household income, depressive symptoms, gender of respondent, number of children, and weekly work hours. After constructing the model, we examined the critical ratios for the regression weights of each path to ascertain which paths, if any, were not significant. There were several paths that were not significant. The final model is shown in Figure 2 and the results are discussed below. Based on fit indexes, the model fit the data well: χ2 = 934.748, df = 465, p < .01; Tucker–Lewis index = .958; comparative fit index = .965; root mean square error of approximation = .038 (Byrne, 2010).

Final model with standardized coefficients to family formation–work fit and fertility outcomes.
As significance tests for indirect and total effects cannot be conducted with missing data, we exported the FIML estimated correlation matrix, means, and standard deviations and reran the model (Macho & Ledermann, 2011). As was expected, model results were identical. We then employed maximum likelihood Monte Carlo bootstrapping to extract 1,000 bootstrap samples to obtain the bias-corrected significance levels for the indirect, direct, and total effects (see Table 2).
Decomposition of Effects From Structural Equation Model on Mediating and Outcome Variables.
Note. N = 690. Bootstrap bias corrected p values: †p < .10. *p < .05. **p < .01. Only variables with a total effect of .05 or greater are included in the table. Variables with no significant direct, indirect, or total effect were removed from the table. Direct and indirect effects may not sum to total because of rounding.
Predictors of WTF Spillover
Evaluating the total effects (see Table 2), for both FTW and WTF conflict, the strongest predictor was depressive symptoms (β = .50 and .42, respectively). Being female were predictive of higher FTW conflict (β = .11). Household income (β = .15) and overtime hours (β = .25) were predictive of higher levels of WTF conflict. The greatest indicator of TF facilitation was schedule flexibility (β = .20). Schedule flexibility was also the strongest predictor of FTW facilitation (β = .22). Interestingly, overtime hours (β = .11) and depressive symptoms (β = .14) were also predictive of greater FTW facilitation.
Predictors of Outcome Variables
The final model (see Figure 2) accounted for 15% of the variance in marital satisfaction scores. Considering the total effects (see Table 2), depressive symptoms had a significant effect on both marital satisfaction and fit, with more depressive symptoms associated with lower marital satisfaction (β = −.20) and less family formation-work fit (β = −.22). FTW facilitation was positively associated with marital satisfaction (β = .28), while FTW conflict was predictive of lower satisfaction (β = −.25). The one finding contrary to our hypotheses was that the direct effect of FTW facilitation on family formation–work fit was negative (β = −.12), though the total effect was not significant. The final model also explained 25% of the variance in family formation–work fit. Schedule flexibility (β = .21), WTF facilitation (β = .24), and marital satisfaction (β = .17) were predictive of higher family formation–work fit, while FTW conflict (β = −.28) and WTF conflict (β = −.16) were predictive of lower family formation–work fit. Above and beyond the effect of control variables (see Table 2), family formation–work fit was predictive of desired family size achievement (β = .06), while marital satisfaction (β = −.08) was negatively related to desired family size achievement. Finally, in addition to the control variable effects, desired family size achievement was a strong predictor of plans to have (more) children (β = −.56).
Nonlinear Regression Analysis
We were intrigued that the SEM analysis showed only a weak relation between family formation–work fit and achievement of desired family size. We concluded that this relation may not be linear. In predicting fertility outcomes, the degree to which an individual agrees or disagrees that his or her work fits with his or her desired number of children may not be as important as whether he or she scored in the agree or disagree range. To explore this nonlinear relation, we conducted a hierarchical regression analysis. In hierarchical regression—also known as sequential regression—groups of variables are entered in separate blocks. An F test is then used to evaluate significant improvement in the R2 value from one block to the next and indicates if the predictor(s) at each step significantly improve the prediction from earlier blocks (Tabachnick & Fidell, 2013).
In the first block, we used stepwise regression and entered current number of children, age, education, household income, depressive symptoms, marital satisfaction, and gender as covariates. Only number of children (β = .50, p < .001), age (β = .17, p < .001), and marital satisfaction (β = −.07, p < .05) were significant. We next entered in family formation–work fit as a linear variable. Similar to the SEM analysis, the linear relation was weak (β = .06, p < .10). Given the expected relation of family formation–work fit with desired family size achievement, testing a cubic effect was the most appropriate. To do so, we mean centered family formation–work fit and computed both a quadratic and cubic term and entered them into the regression equation in separate blocks. The quadratic effect was not significant (β = .03, ns); however, consistent with our expectations, the cubic term was significant (β = −.20, p < .01). This final model significantly accounted for approximately a third of the variability in scores for desired family size achievement (adjusted R2 = .34; F(6, 608) = 54.13, p < .001) and was a significant improvement over the model with only the covariates (F(3, 608) = 4.31, p < .01).
After conducting the regression, we examined the regression plots and noted that desired family size achievement scores for those in the disagree range on family formation–work fit were nearly identical while scores for those in the agree range were also very similar. However, there was a noticeable difference in desired family size achievement score between those in the disagree and those in the agree range of family formation–work fit. Noting these differences in the plots, we conducted mean comparison analyses (ANOVA, not shown). Mean desired family size achievement scores were not significantly different between the strongly disagree and disagree levels, nor were mean scores different for the strongly agree and agree levels. However, there was a significant mean difference (Mdiff = .18) in the desired family size achievement scores between those in the Disagree range and those in the Agree range. Thus, approximately 18% more of individuals who scored in the agree range said they had achieved their desired family size than those in the disagree range.
Post Hoc Analysis
Because individuals who reported family formation–work fit were more likely to indicate achievement of desired family size, we were curious about the number of children of those who had achieved their desired family size. We also were curious about differences in plans to have children based on achievement of desired family size. To explore these post hoc questions, we conducted a multivariate analysis of covariance (MANCOVA) to compare the mean scores for number of children and plans to have more children by desired family size achievement group after controlling for age, respondent gender, education, household income, and depressive symptoms. The MANCOVA was significant (Wilks’s Λ = .471, F(2, 682) = 383.19, p < .001), indicating there were differences between groups, with a large effect for desired family size achievement (partial η2 = .53). The age covariate was significant (Wilks’s Λ = .839, F(2, 682) = 65.24, p < .001), as was gender (Wilks’s Λ = .972, F(2, 682) = 9.99, p < .001), albeit the effects were small (partial η2 = .16, .03 respectively). The effect for depressive symptoms (Wilks’s Λ = .988, F(2, 682) = 4.02, p < .05, partial η2 = .01) was also significant. Household income and education were not significant. As follow-up tests to the MANCOVA, we conducted a univariate analysis of covariance (ANCOVA). We employed the Bonferroni method to control for Type I error. Compared with those with desired family size (M = 2.11, SD = 0.80), those with less than desired family size had significantly fewer children (M = 0.89, SD = 0.84; F(1, 683) = 281.41, p < .001; partial η2 = .29). A significantly smaller percentage of those who reported desired family size achievement (M = 0.03, SD = 0.17) reported plans to have more children than those who had not achieved desired family size (M = 0.71, SD = 0.46; F(1, 683) = 557.19, p < .001; partial η2 = .45).
Discussion
Guided by three perspectives from work–family research, we developed a model to better understand how work and family spheres interact and influence the fertility decisions of Singaporean workers in dual-earner families. Family formation–work fit, the primary variable of interest, was negatively related to WTF/FTW conflict, positively related to WTF facilitation, and to a lesser extent, FTW facilitation. While only our central hypothesis (f) was fully supported, all our other hypotheses were partially supported. Results that support our central hypothesis and partially support our other hypotheses provide evidence that refines the construct of family formation–work fit, as well as work–family research with Singaporean dual-earning couples.
Family Formation–Work Fit and Family Fertility
Family formation–work fit was positively related to desired family size achievement, which in turn was related to individuals’ plans to have children. Post hoc analysis revealed that those who had achieved ideal family size had 2.11 children in the household, which is well above the current Singaporean fertility rate of 1.20 (Department of Statistics Singapore, 2012). These findings support other work that suggests the intention of workers to expand or delay their families is related to their perceptions of how well work life aligns with ideals of family life (Gerson, 2010). The support for this hypothesis provides early evidence validating the concept of family formation–work fit. Further empirical and theoretical exploration of the construct is warranted and could be of particular importance to nations facing diminishing fertility rates.
Work–Family Perspectives and Family Formation-Work Fit
Work Socialization Perspective
The partial support of Hypotheses a and b replicate previous work in the work socialization literature on overtime and schedule flexibility. Schedule flexibility was positively related to both WTF/FTW facilitation and overtime hours was positively related to WTF conflict and FTW facilitation. Schedule flexibility’s relation with FTW/WTF facilitation, but not WTF/FTW conflict, reinforces that facilitation and conflict are independent constructs that require separate study (Voydanoff, 2008). Masuda et al.’s (2011) work also found no relation between WTF conflict and schedule flexibility when analyzing data from Asian samples and points to the need for research that further examines cultural differences (cf. Galovan et al., 2010). Theoretical work by Frone (2003) on domain specificity helps explain the lack of relations between FTW conflict and workplace variables. Frone’s work suggests the relation between workplace variables such as overtime and FTW conflict is theoretically questionable, and our work supports this claim. Finally, overtime was not associated with increases in WTF facilitation, meaning that for our sample, the increased work hours associated with overtime was related to greater FTW facilitation, but not WTF facilitation, once again providing evidence of the independence of the spillover constructs. In sum, our testing of the work socialization perspective successfully found evidence of the workplace affecting families; however, our work also shows that refinement testing of the perspective for cultural differences is still needed.
Conflict, Facilitation, and Fit Perspectives
Conflict and facilitation, as well as workplace schedule flexibility, were nearly all related to family formation work–fit as expected. However, FTW facilitation’s relation with family formation–work fit was unexpectedly negative, possibly indicating that family spillover (conflict or facilitation) into work is generally not viewed positively in Singaporean workers. Thus, though family formation–work fit seems to have a theoretically clear path toward family fertility decisions in our model, the interaction the family and work spheres were particularly complicated when considering the role of marital satisfaction in our model.
Though both WTF conflict and facilitation were related to family formation–work fit, WTF spillover was not associated with marital satisfaction, meaning that the workplace had no measured impact on marital satisfaction. Martial satisfaction had other surprising results. For example, marital satisfaction’s relation with achievement of desired family size was negative—possibly meaning that the more satisfied a worker is with his or her marriage, the less likely that he or she is to be foreclosed to having more children. Also, marital satisfaction in our study was related to higher rates of family formation–work fit and lower rates of achieving desired family size, which has multiple possible implications. Further qualitative work on these findings has the potential to provide better evidence of the processes involved in the relation between family formation–work fit and marital satisfaction. However, these complex findings point to the reality that individuals do not make family decisions solely because of job circumstances and that many elements of the ecological system influence family processes, including having more children. Thus, family formation–work fit should not be viewed as a panacea for understanding and improving nations’ fertility rates. However, these findings do point to the importance of understanding family formation–work fit as one important component of family fertility decisions and as a potential guidepost for policy makers.
Research and Theory Implications
The concept of family formation–work fit is in need of further development. This was an exploratory study to test the relation of this particular fit construct in a region that is particularly interested in family fertility decisions. Despite the need to further understand the construct, evidence from this study suggests the incorporation of family formation–work fit will benefit research on family fertility decisions.
Although this study provides evidence of the importance of workplace flexibility and FTW/WTF conflict and facilitation on marital satisfaction and family formation–work fit, it does not analyze all possible causes of facilitation or conflict. While workplace flexibility was related to WTF/FTW facilitation, it is likely that there are other variables that affect facilitation. In terms of FTW/WTF conflict, few relations were found, which suggests the need for more research exploring the antecedents of work–family conflict. Moreover, the current study reinforces that cultural differences produce slight changes in how work and family interact. Greater research and theory building should explore other globalized economic contexts in order to assess applicability and relevance of westernized theories of the work–family theory in other contexts (Galovan et al., 2010). Finally, by focusing on dual-earner couples, we excluded single-earner spouses. Future research should explore if, and how, work–family dynamics influence individuals’ decisions to exit the workforce leading up to (or following) the birth of a child.
Policy Implications
Family expansion, although not without caveats, may be a key to long-term economic and social stability in the globalized nation (Matsutani, 2006)—stability that may only be assured through a vibrant workforce and manageable dependency ratios (Meyer, 2009; United Nations, 2002). Our results suggest that in rapidly aging nations experiencing ultra-low fertility rates, promoting a balance between work–family demands and resources may lead to greater family formation. One key to this promotion may be greater workplace schedule flexibility. Flexibility in the workplace has been demonstrated to be related to reduced worker stress and burnout (Grzywacz, Carlson, & Shulkin, 2008), which may lead to a better balance between family resources and work demands. Our findings suggest flexibility may help decrease FTW conflict and increase WTF and FTW facilitation. Despite workplace schedule flexibility having one small, unexpected, indirect negative effect on family formation–work fit (through FTW facilitation), the total effect of workplace flexibility on family–formation work fit is moderately strong and positive (β = .21).
Workplace policies that promote flexibility (e.g., parental leave, childcare, flextime) should be researched for potential implementation in regions seeking to increase facilitation, decrease WTF conflict, and increase feelings of family formation–work fit. One benefit of workplace flexibility is that policies implemented through employers may be more visible than tax credits and government programs that require broad public awareness campaigns. The availability of these programs and resources could be missed by large groups of individuals, while employers have presumably more direct lines of communication with their workers (Sun, 2012).
Limitations
This research has several limitations. First, all findings were derived from a cross-sectional design that employed multiple single-item measures. While our analysis was able to provide a theoretical basis for causation, actual causation can be better accessed through longitudinal studies. Future studies should also incorporate multiple-item measures to better account for measurement error and the multidimensional nature of many of the constructs. Furthermore, because of selection criteria for inclusion in this study (e.g., married, dual-earner couples), the subsample may not be fully representative of all dual-earner couples in Singapore. However, given the relative similarity of the subsample with the full sample in demographic characteristics, this is a minor concern. Also, this is a study on a select group of people from a select context (i.e., married, dual-earner Singaporeans) requiring caution when generalizing findings from this context to another.
Conclusion
This study provides provocative findings highlighting the relation of the work–family interface to family formation and the achievement of desired family size. Like Singapore, there are many prosperous countries in the world that struggle with ultra-low fertility rates. Our findings suggest that workplace policies that promote family formation–work fit may be an important part of the solution to increase national fertility rates. While there are no simple solutions for complicated processes, such as declining fertility rates, our model provides a useful exploration of whether the work environment supports or interferes with a family’s desired formation or expansion. Understanding antecedents to and improving family formation–work fit may lead to achievement of workers’ desired family size.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: We thank the Family Studies Center of the Brigham Young University School of Family Life and the Ministry of Social and Family Development in the Republic of Singapore for their support of this project.
