The Longitudinal Study of Australian Children
Annual statistical report 2012

4 Echoes of disadvantage across the generations? The influence of long-term joblessness and separation of grandparents on grandchildren

Kirsten Hancock, Telethon Institute for Child Health Research, Centre for Child Health Research, The University of Western Australia

Ben Edwards, Australian Institute of Family Studies

Stephen R. Zubrick, Telethon Institute for Child Health Research, Centre for Child Health Research, The University of Western Australia

Download printable version: 4. Echoes of disadvantage across the generations? The influence of long-term joblessness and separation of grandparents on grandchildren (PDF 791 KB)

4.1 Introduction

Intergenerational disadvantage refers to the situation in which multiple generations of the same family experience high and persisting levels of social exclusion, material and human capital impoverishment, and restrictions on the opportunities and expectations that would otherwise widen their capability to make choices (d'Addio, 2007; Frazer & Marlier, 2007). Levels of earnings, education, occupational status, wealth, decisions about family formation and receipt of welfare support have been found to persist across generations, suggesting that there is low intergenerational social and economic mobility (d'Addio, 2007). While such low mobility is beneficial for families from high socio-economic backgrounds, it is a real issue for disadvantaged families. d'Addio neatly summarised the issue as follows: "when intergenerational mobility is low, poverty during childhood will not only undermine the health, nutrition and education prospects of children, but will also increase the chances that the children of the next generation will grow up in low-income households" (p. 11).

It is important to appreciate that intergenerational disadvantage extends beyond the transmission of economic and material impoverishment to encompass the contextual circumstances that contribute to its perpetuation. For example, there is robust evidence to suggest that the likelihood of relationship separation also persists across generations in many different countries (e.g., Wolfinger, 2005, 2011). International studies of intergenerational income mobility have found that when the father is absent, the correlation between a child's earnings in later life and their father's earnings decreases, because children living in lone-parent families tend to move downwards in the income distribution compared with children from intact families (Biblarz & Rafferty, 1993; Bjorklund & Chadwick, 2003; Bratberg, Rieck, Marshall & Vaage, 2011; Fertig, 2007).1 These findings are also consistent with work by the Organisation for Economic Co-operation and Development (OECD) that suggested that intergenerational disadvantage is more likely in children who live only with their mothers (d'Addio, 2007).

Most of these studies have focused on just two generations. One of the few studies of the effects of divorce on multiple generations, by Amato and Cheadle (2005), reported that divorce in the grandparent generation is associated with lower educational attainment and more marital discord among grandchildren, as well as poorer relationships between the grandchildren and their own parents.2

While Australia has a relatively low rate of joblessness overall compared to other OECD countries, it nevertheless has one of the highest rates of family joblessness among single-parent households (Whiteford, 2009). Policy interest in family joblessness therefore remains (Department of Prime Minister & Cabinet, 2009), as it is still a marker of entrenched disadvantage - especially intergenerational disadvantage - that affects children's life chances. Several studies have shown that children living in jobless households have poorer social-emotional wellbeing and learning outcomes (Gray & Baxter, 2012; Gray, Taylor, & Edwards. 2011). However, it remains uncertain whether joblessness in succeeding generations is associated with poorer child development outcomes than joblessness in one generation alone.

The purpose of this chapter is to document parents' childhood experiences of growing up within disadvantaged families and assess the extent to which these experiences are reflected in adulthood. We also examine the effects of persistent disadvantage - in this context, disadvantage that is experienced over two generations - on children's wellbeing and development, and how these effects compare to children with just one generation of disadvantage, or no history of disadvantage at all. This is a rare examination of joblessness over three succeeding generations and its onward effects upon children, and is one of the few in the research literature.3 In this chapter, we use data from both the B and K cohorts of Growing Up in Australia: The Longitudinal Study of Australian Children (LSAC), to address the following questions:

  • What proportion of maternal and paternal grandparents of LSAC children experienced family joblessness or separation when LSAC parents were growing up?4
  • To what extent is parents' experiences of family joblessness and separation associated with grandparents' experiences of joblessness and separation?
  • What effect does this family history have on children's development?
  • How do children fare when their family has two generations of disadvantage, compared to children with disadvantage only in their parents' generation, only in their grandparents' generation, or no history of disadvantage at all?

4.2 Data and definitions

Employment in the grandparent generation

Data on maternal and paternal grandparents' employment were collected at Wave 4 via a face-to-face interview with the parent who knew best about the child's health, development and care at Wave 1 (Parent 1). Parent 1 responded on behalf of themselves and Parent 2 (who lived in the same household as Parent 1). Parent 1 was asked a range of questions pertaining to the main breadwinner in their family when they (and Parent 2) were 14 years old: "Thinking back to when you were 14 years old, was your mother or father the main breadwinner?".5 The respondent could nominate their mother, father, someone else or a combination of these options for their response; for example, both mothers and fathers could be nominated as the main breadwinners.

For each person nominated as a main breadwinner, the respondent was then asked: "At that time, did your mother/father/someone else work in a job, business or farm?". If "yes", the respondent was then asked: "Was your mother/father/someone else unemployed for a total of 6 months or more while you were growing up?".

Similar questions were asked regarding Parent 2's family of origin. This information was also collected from parents living elsewhere from Parent 1 (PLEs) in a telephone interview at Wave 4. As family structure could change across the waves, family-of-origin data for fathers could be collected from either a male parent who resided in the same household as Parent 1 or from a biological father who was living in a different household to Parent 1 (male PLE) or, for a small minority of children, from both. Where family-of-origin data were collected from either source, whichever data were available were used to represent fathers' experiences growing up. Where family-of-origin data were collected from both a father with whom Parent 1 lives (i.e., step-father) and a male PLE, precedence was given to data collected from male PLEs, as in the majority of cases this person was the biological father.6

Based on answers to the employment questions, a "family joblessness" variable was derived separately for the maternal and paternal grandparents of the LSAC study child. For the purpose of this chapter, a jobless family in the grandparent generation refers to families where the main breadwinner either did not work in a job, business or farm, or was unemployed for a period of 6 months or more. Where both the grandmother and grandfather of the study child were nominated as being the main breadwinners of their families, these families were only classed as being jobless households if both grandparents did not work or were unemployed for a period of 6 months or more. Where "someone else" was named as a breadwinner, they were the sole breadwinner for the majority of cases.7 For simplicity, we excluded from analyses the small number of cases (n = 18) where both a grandmother and "someone else" or a grandfather and "someone else" were nominated by parents as being the main breadwinners.

It is important to note that while the breadwinner questions related to the time when the study child's parents were 14 years old, the questions pertaining to the unemployment of the main breadwinner were framed as whether it occurred while they were "growing up". The measure of unemployment in the grandparent generation therefore does not capture when it occurred. Additionally, for households where both grandparents were nominated as breadwinners, unemployment may have occurred at different times, and one breadwinner may have been employed while the other was not. Therefore, dual breadwinner households may not have been "jobless" per se. However, as few parents indicated that both grandparents were main breadwinners (see Table 4.3), and for ease of discussion when assessing intergenerational disadvantage, the "jobless" terminology has been adopted for this chapter.

Separation in the grandparent generation

Data on separation in the grandparent generation were collected from parents at multiple points across Waves 2, 3 and 4. At Wave 2, data on whether grandparents had separated or divorced (and if so, the age of parents when this occurred)8 were collected via the Parent 1 and Parent 2 leave-behind questionnaires. At Waves 3 and 4, the same questions were asked during the face-to-face interview with Parent 1. As with the employment questions, Parent 1 responded to the questions on behalf of Parent 2. Data on separation in the grandparent generation were also collected from PLEs at Waves 3 and 4 via the telephone interview. The data on separation were maximised, where possible, so that if data were missing from one wave they were supplemented from another.9

Joblessness and separation in study child's family

To classify joblessness in the parent generation, we adopted the approach used by Gray and Baxter (2011). At each wave, LSAC parents were coded according to their employment status, with parents working full-time given a score of 1, parents working part-time a score of 0.5 and parents who were either not working or unemployed given a score of 0. Households scoring 0 (i.e., did not have one parent working at least part-time) were coded as being jobless for that wave. Joblessness across all four waves was then collated to determine the number of waves that the household was jobless. Note that this classification only refers to employment status at the time of interview, and does not cover either having a job or being jobless between waves.

For family separation, families that were lone-parent households at any wave were classified as being an "ever lone-parent family". Under this definition, a lone-parent household would include those families where a parent had passed away, or had never lived with the study child. The vast majority of lone-parent households, however, resulted from family separation; therefore in this chapter the term "separated families" - with respect to the study child's parents - is used interchangeably with "lone-parent households".10

Two-generation family history of joblessness and separation

In order to assess the effect of multiple generations of disadvantage on study children's social and academic outcomes, we derived family history variables for each study child relating to maternal and paternal experiences of joblessness and family separation. Overall, four separate variables were created to document family background: maternal history of joblessness, maternal history of separation, paternal history of joblessness, and paternal history of separation. For each of these variables, study children fell into one of four categories:

  • no history;
  • G1 only: generation 1 only (the study child's grandparent[s]);
  • G2 only: generation 2 only (the study child's parent[s]); and
  • G1 + G2: generations 1 and 2.

For maternal history of joblessness, for example:

  • G1 only: refers to study children whose maternal grandparent(s) were jobless; in other words, the study child's mother grew up experiencing joblessness in her family;
  • G2 only: refers to study children whose parent(s) were jobless at least once over the four waves, but whose maternal grandparents were not;
  • G1 + G2: refers to study children whose maternal grandparent(s) were jobless and whose parent(s) were jobless.

A paternal history of joblessness variable was also created in the same manner. As "G2 only" refers to the combined experience of the study child's mother and father, the maternal and paternal family history variables were not mutually exclusive. Therefore, a child coded as "G2 only" for maternal history may also have paternal grandparents who were jobless; in this case, the child would be coded as "G1 + G2" on a paternal history of joblessness.

4.3 Joblessness and separation in the grandparent generation

The percentages of maternal and paternal grandparents who were ever separated or were jobless are shown in Table 4.1, while the combined experiences of separation and joblessness are shown in Table 4.2. For both cohorts, 26-28% of LSAC grandparents had ever separated, and 18-21% of maternal grandparents and 12-14% of paternal grandparents were jobless when parents were aged around 14 years. When the experiences of both separation and family joblessness were combined (Table 4.2), 17-19% of maternal grandparents had separated only, 10-11% were jobless only, 8-11% were both separated and jobless, and 64-62% had experienced neither separation nor joblessness. Percentages for paternal grandparents were similar: 17-18% separated only, 8% jobless only, 3-6% both separated and jobless, and 68-72% neither.

Table 4.1: Maternal and paternal grandparents of the study child who were ever separated or were jobless, B and K cohorts, Waves 1-4
B cohort grandparents K cohort grandparents
Maternal Paternal Maternal Paternal
% n % n % n % n
Ever separated 27.7 4,140 25.1 3,981 25.8 4,000 21.7 3,864
Jobless household 21.2 3,902 14.0 3,471 18.0 3,756 12.1 3,389

Note: The residual proportions of non-separated and non-jobless households are omitted. For example, 28% of maternal grandparents had separated, while 72% of maternal grandparents had never experienced separation. The latter category has been omitted from the table.

Table 4.2: Maternal and paternal grandparents' combined experience of separation and joblessness, B and K cohorts, Waves 1-4
B cohort grandparents K cohort grandparents
Maternal (%) Paternal (%) Maternal (%) Paternal (%)
Neither separated nor jobless 61.6 67.8 63.8 71.5
Separated only 17.4 18.4 18.5 16.7
Jobless only 10.5 8.1 10.0 7.9
Both separated and jobless 10.5 5.7 7.6 3.4
Total 100.0 100.0 100.0 100.0
No. of households 3,825 3,371 3,642 3,308

Note: Percentages may not total exactly 100.0% due to rounding.

When considering the influence of grandparents' separation and disadvantage on parents, it is important to understand that before the introduction of the Child Support Scheme in 1988 (the period when most grandparents were raising LSAC parents), unless mothers re-partnered following separation, they and their children were usually at considerable financial disadvantage (Funder, Harrison, & Weston, 1993; McDonald, 1986).11

In addition to the lack of child support, lone mothers also had much lower rates of employment than they do now. For instance, Hayes, Weston, Qu, and Gray (2011) used data from the Australian Bureau of Statistics Labour Force Status and Other Characteristics of Families series to demonstrate that, in 1983, close to 70% of lone mothers did not have any job and just over 10% had part-time work, whereas by 2009, just under 50% of lone mothers had no job and about 25% had part-time work.

To better understand the experience of separation and unemployment in the grandparent generation, understanding who the main breadwinners were provides a useful context around the experience of disadvantage in these families. Table 4.3 provides an overview of the main breadwinners in the grandparent generation, according to separation experience. For maternal grandparents who separated, 48% of B cohort maternal grandmothers and 41% of K cohort maternal grandmothers were the main breadwinners, compared to just 7% of maternal grandmothers (both cohorts) who had never separated. Similarly, 32% of paternal grandmothers were the main breadwinner in families where paternal grandparents had separated, compared to just 6% of paternal grandmothers who had not separated. In contrast, 83% of maternal grandfathers and 86-87% of paternal grandfathers were the main breadwinners for intact families, compared with 36-43% and 56-57% respectively for those who had ever separated. Finally, "someone else" was more likely to be a main breadwinner if grandparents had separated than if they had not separated (e.g., for the B cohort, 11% for maternal grandparents who had separated compared with 1% who had not, and 7% for paternal grandparents who had separated compared with 1% who had not).

Table 4.3: Main breadwinners in grandparent generation, by whether ever separated, B and K cohorts, Waves 1-4
Main breadwinner B cohort grandparents K cohort grandparents
Ever separated (%) Never separated (%) Total (%) Ever separated (%) Never separated (%) Total (%)
Maternal grandparents
Grandfather only 36.0 82.8 69.8 42.9 83.1 72.7
Grandmother only 48.3 7.3 18.7 41.3 6.6 15.6
Both grandfather and grandmother 4.3 9.2 7.9 5.4 9.4 8.4
Someone else only 11.4 0.7 3.6 10.5 0.9 3.4
Total 100.0 100.0 100.0 100.0 100.0 100.0
No. of households 1,056 3,073 4,129 953 3,025 3,978
Paternal grandparents
Grandfather only 56.2 85.7 78.4 57.4 87.0 80.7
Grandmother only 32.0 5.5 12.0 32.4 5.6 11.3
Both grandfather and grandmother 5.0 7.7 7.1 6.2 6.2 6.2
Someone else only 6.8 1.1 2.5 4.0 1.2 1.8
Total 100.0 100.0 100.0 100.0 100.0 100.0
No. of households 2,875 923 3,798 2,914 755 3,669

Note: Percentages may not total exactly 100.0% due to rounding.

The time when grandparents were likely to have separated (the 1980s), 34% of the grandmothers who separated were living with their own parents or friends in the months following divorce (McDonald, 1986). On average, these women had separated five years after marriage.

Table 4.4 shows the percentage of grandparents who were jobless according to who the main breadwinner was.

Table 4.4: Maternal and paternal grandparents who were jobless, by main breadwinner, B and K cohorts, Waves 1-4
Main breadwinner Maternal grandparents jobless Paternal grandparents jobless
% n % n
B cohort
Grandfather only 12.0 2,994 8.5 2,942
Grandmother only 52.8 707 44.6 415
Both grandfather and grandmother 26.2 83 14.2 47
Someone else 32.7 118 47.7 67
Totals 21.2 3,902 14.0 3,471
K cohort
Grandfather only 11.0 2,988 7.3 2,923
Grandmother only 44.9 591 41.5 382
Both grandfather and grandmother 26.8 67 28.4 36
Someone else 40.2 110 39.0 48
Totals 18.0 3,756 12.1 3,389

Note: The residual proportions of non-jobless households by main breadwinner are omitted. For example, for the B cohort there were 2,994 households where the maternal grandfather only was the main breadwinner, 12% of these households were jobless, while 88% were not jobless. The latter group has been omitted from the table.

Joblessness was clearly more common where only grandmothers were the main breadwinners. For the K cohort, for example, in families where the grandmother was the main breadwinner, 45% of maternal grandmothers and 42% of paternal grandmothers experienced joblessness, compared with only 11% and 7% of families where the grandfather was the main breadwinner. Joblessness was also common in families where "someone else" was the main breadwinner (e.g., 39-40% in K cohort families) and where both grandmothers and grandfathers were breadwinners (e.g., 27-28% for K cohort families).

4.4 The intergenerational continuity of family joblessness

In Table 4.5 we examine the persistence of disadvantage across generations by showing the percentage of LSAC parents who have been jobless in at least one wave, according to grandparents' experience of family joblessness. Because Australia has one of the lowest rates of employment for single parents among OECD countries (50%, compared to an OECD average of 71%; OECD, 2007), and, since 1980, single-parent families have accounted for at least half of all jobless families with children (Whiteford, 2009), the results in Table 4.5 were separated by whether the LSAC family was ever a lone-parent family across Waves 1-4.12

Table 4.5: Joblessness in parent and grandparent generations, by whether child's family was ever a lone-parent family, B and K cohorts, Waves 1-4
Joblessness in grandparent generation Joblessness in never lone-parent family Joblessness in ever lone-parent family
% n % n
B cohort
Neither maternal or paternal grandparents jobless 6.2 2,045 53.1 258
Either maternal or paternal grandparents jobless 8.6 583 70.1 a 117
Both maternal and paternal grandparents jobless 20.0 a, b 100 79.8 a 24
Totals 7.3 2,728 60.5 399
K cohort
Neither maternal or paternal grandparents jobless 5.3 2,002 41.9 314
Either maternal or paternal grandparents jobless 8.1 457 54.6 a 127
Both maternal and paternal grandparents jobless 15.6 a 84 65.4 a 20
Totals 6.2 2,543 46.7 461

Notes: Due to the small number of households where the study child's family was ever a lone-parent family and where both maternal and paternal grandparents were jobless, those results should be interpreted with caution. a Percentage is significantly different (at p < .05) to the families where neither maternal nor paternal grandparents were jobless (using pair-wise chi-square tests). b Percentage is significantly different (at p < .05) to the families where either maternal or paternal grandparents were jobless (using pair-wise chi-square tests). The residual proportions of non-jobless households by grandparent joblessness are omitted from the table. For example, out of 2,045 never lone-parent B cohort families where neither maternal nor paternal grandparents were jobless, 6.2% experienced joblessness, 93.8% did not. The latter group has been omitted from the table.

LSAC parents who were lone-parent families at any point in time were also overwhelmingly more likely to have been jobless at any point in time, with 61% of B cohort and 47% of K cohort families (who were ever lone-parent families) having been jobless, compared to 7% of B cohort and 6% of K cohort families who had never been lone-parent families.

Within both LSAC family structures, being in a jobless family was also associated with experiences of grandparent joblessness. For example, within two-parent B cohort families who had been jobless for one wave or more, 20% were those where both the mother and father had experienced family joblessness in their family of origin, compared to only 6% where neither parent had had such an experience.13 The percentages were much higher for ever lone-parent B cohort families who had been jobless for one wave or more, with 80% of these families having had both parents experiencing family joblessness growing up, compared to 53% having had neither parent experiencing family joblessness - a significant difference. Similar patterns were found for the K cohort.

4.5 The intergenerational continuity of separation

Previous research has consistently shown that the likelihood of separation in adulthood is greater for adults whose own parents have separated (D'Onofrio et al., 2008; Wolfinger, 2005, 2011). In line with previous research, Table 4.6 shows that a larger percentage of LSAC parents separated if either the maternal or paternal grandparents had separated. This percentage was even greater if both maternal and paternal grandparents had separated. For K cohort families where both maternal and paternal grandparents had separated, the risk of parental separation was more than double that of families with no grandparent history of separation (45% compared to 17%), and likewise for B cohort families (31% compared to 14%). For both cohorts, the percentage of families who were lone-parent families at any wave was also significantly higher for families where either the maternal or paternal grandparents had separated compared with families where there was no history of grandparent separation. Notably, a higher percentage of K cohort than B cohort families were lone-parent families by Wave 4. This is due to the fact that parents would have had more time to separate.14

Table 4.6: Families who had ever been a lone-parent family, by whether grandparent generation separated, B and K cohorts, Waves 1-4
Separation in grandparent generation Ever lone-parent families (%) Ever lone-parent families ( n)
B cohort
Neither maternal or paternal grandparents had separated 13.8 2,180
Maternal grandparents only had separated 21.4 a, b 656
Paternal grandparents only had separated 21.3 a, b 660
Both maternal and paternal grandparents had separated 31.2 a 278
Totals 17.9 3,774
K cohort
Neither maternal or paternal grandparents had separated 16.5 2,239
Maternal grandparents only had separated 27.6 a, b 603
Paternal grandparents only had separated 21.0 a, b 535
Both maternal and paternal grandparents had separated 44.8 a 230
Totals 21.1 3,607

Notes: a Percentage is significantly different (at p < .05) to the families where neither set of grandparents had separated (using pair-wise chi-square tests). b Percentage is significantly different (at p < .05) to the families where both set of grandparents had separated (using pair-wise chi-square tests).

4.6 Intergenerational disadvantage and children's development

Earlier research using LSAC data has demonstrated that family joblessness is associated with poorer developmental outcomes for children across the learning, social-emotional and physical health domains, particularly for children living in families where joblessness endures over time (Gray & Baxter, 2011). Longitudinal analyses have also been conducted on US data showing that parental divorce is associated with a decline in children's psychosocial wellbeing and academic achievement (Potter, 2010). In this chapter, we examine how the experience of separation and joblessness across two generations affects two key domains of children's development: social-emotional problems and academic performance.

Child social-emotional problems

Social-emotional problems were measured using the Strengths and Difficulties Questionnaire (SDQ; Goodman, 1997). The SDQ is a 25-item scale that assesses peer problems, conduct problems, hyperactivity, emotional problems and prosocial behaviour. Total problem scores, which are the sum of scores across the four problem subscales and could range from 0 to 40, were used in this chapter. Using Goodman's recommended cut-off points (Goodman, 1997), children scoring 14 or above (borderline or abnormal range) were classified as likely having social-emotional problems.15 Wave 4 data were used for both cohorts.

Intergenerational joblessness and children's social-emotional problems

As the majority of data for LSAC mothers were based on self-reports and are therefore more reliable than data for fathers, which were largely collected from LSAC mothers, in this section we focus on maternal grandparents' experience of family joblessness, though the experiences of paternal grandparents are also presented for information.

Figure 4.1 shows the percentage of study children with social-emotional problems according to patterns of joblessness in the grandparent and parent generations. For both cohorts, as the extent of intergenerational disadvantage increased in terms of joblessness, the percentage of children with social-emotional problems increased in a step-wise pattern. For example, 17% of B cohort children had social-emotional problems where maternal grandparents had been jobless (G1 only), 24% where parents (but not grandparents) had been jobless (G2 only), increasing to 33% for children whose grandparent(s) and parent(s) had both been jobless (G1 + G2). For those children without joblessness in either generation, only 13% had child social-emotional problems. A similar pattern was shown for K cohort children, though the percentages were higher in each category of having a history of joblessness relative to the B cohort.

Figure 4.1: Children with social-emotional problems, by history of family joblessness over two generations, mothers' and fathers' experiences, B and K cohorts, Wave 4

Children with social-emotional problems, by history of family joblessness over two generations, mothers' and fathers' experiences, B and K cohorts, Wave 4 - as described in acccompanying text.

Notes: B cohort: mother (n = 3,876), father (n = 3,447); K cohort: mother (n = 3,713), father (n = 3,348). Maternal history, G1 + G2 vs No history: B cohort: χ2(1, n = 558) = 5.5, p = .045; K cohort: χ2(1, n = 525) = 11.3, p < .001.

Notably, the percentage of children with social-emotional problems was significantly higher for both cohorts if they had two generations of joblessness in their family compared to joblessness at G2 only and to no joblessness history (33%, 24% and 13% respectively for the B cohort, 43%, 27% and 12% for the K cohort). These results suggest that irrespective of whether LSAC parents were jobless, the employment experiences of maternal grandparents are associated with children's social-emotional problems. Furthermore, there appears to be an additive effect, where two generations of disadvantage (here, joblessness) has a larger association with children's wellbeing than just one generation of disadvantage.

Intergenerational separation and children's social-emotional problems

Figure 4.2 shows the percentages of children with social-emotional problems according to the history of separation in their family. Again, we focus on maternal history in our discussion, but also include paternal history in the figure for information. In contrast to joblessness, whether or not maternal grandparents had separated had little bearing on the percentages of study children with social-emotional problems. For both cohorts, there was no significant difference in the percentage of children with social-emotional problems for children with no history of separation (13-14% across cohorts), compared to children whose maternal grandparents had ever separated or divorced (15-16% across cohorts). The percentage of children with social-emotional problems was higher if their parents had ever been a lone parent (25-27% across cohorts) compared to children whose parents were never separated (24% for the B cohort and 31% for the K cohort), but there appeared to be no additional effects of separation in the grandparent generation beyond this. In sum, there is no evidence that family separation in the grandparent generation had any influence on children's social-emotional problems.

Figure 4.2: Children with social-emotional problems, by history of family separation over two generations, mothers' and fathers' experiences, B cohort and K cohort, Waves 4

Children with social-emotional problems, by history of family separation over two generations, mothers' and fathers' experiences, B cohort and K cohort, Waves 4 - as described in acccompanying text.

Notes: B cohort: mother (n = 3,958), father (n = 3,836); K cohort: mother (n = 3,823), father (n = 3,721). Maternal history, G2 only vs No history: B cohort: χ2(1, n = 2,936) = 72.4, p < .001; K cohort: χ2(1, n = 2,895) = 40.5, p < .001. Maternal history, G1 + G2 vs G2 only: B cohort: χ2(1, n = 734) = 0.7, p = .489; K cohort: χ2(1, n = 846) = 4.3, p = .062.

Literacy and numeracy

The Academic Rating Scale (ARS) provides a measure of school performance in literacy and mathematical ability (numeracy), based on the reports of the child's teacher. The scales used in LSAC were adapted from the versions developed for the Early Childhood Longitudinal Study (National Center for Education Statistics, n.d.). Scores are based on teachers' assessments of students relative to other children of the same age level. Unique items were used at each wave to assess age-relevant literacy and numeracy competencies. At Wave 4, the scales consisted of 10 literacy and 8 numeracy items for the B cohort, and 9 literacy and 10 numeracy items for the K cohort. Teachers were asked about items such as: "The study child understands and interprets a story or other text read aloud (e.g., identifies an author's purpose, identifies persuasive techniques through information presented and language choices)", and could choose from one of 5 responses where 1 = "not yet"; 2 = "beginning"; 3 = "in progress"; 4 = "intermediate" or 5 = "proficient". Scores were summed and scaled to range from 1 to 5, with higher scores representing greater proficiency. To facilitate comparisons across domains, the scores were standardised for each cohort to have a mean of 0 and a standard deviation of 1. The standardisation of scores in this manner means that differences can be interpreted as effect sizes, which describes the magnitude of differences between groups. To facilitate interpretation, a difference of .20 of a standard deviation unit is considered a small difference, .50 medium, and .80 a large difference (Cohen, 1988). To put these "rules of thumb" into perspective, in the social sciences, the majority of differences found are small or medium.

Intergenerational joblessness and children's academic performance

The relationship between family history of joblessness and language and literacy scores is shown in Figure 4.3, and for numeracy in Figure 4.4, with very similar results shown for both cohorts and academic domains. As in the previous sections, we focus on maternal history for discussion, though paternal history is also provided. Academic performance was highest for children with no family history of joblessness, and lowest for children with two generations of family joblessness, with differences between these two groups being statistically significant. Notably, average scores were also lower for children with two generations of joblessness compared to children with joblessness only in the parent generation.

Figure 4.3: Children's literacy scores, by history of joblessness over two generations, mothers' and fathers' experiences, B and K cohorts, Wave 4

Children's literacy scores, by history of joblessness over two generations, mothers' and fathers' experiences, B and K cohorts, Wave 4 - as described in acccompanying text.

Notes: B cohort: mother (n = 3,146), father (n = 2,631); K cohort: mother (n = 3,010), father (n = 2,572). Maternal history, G1 + G2 vs No history: B cohort: t(2,382) = 8.9, p < .001; K cohort: t(2,326) = 9.4, p < .001. Maternal history, G1 + G2 vs G2 only: B cohort: t(433) = 2.4, p = .019; K cohort: t(409) = 1.9, p = .056

Figure 4.4: Children's numeracy scores, by history of joblessness over two generations, mothers' and fathers' experiences, B and K cohorts, Wave 4

Children's numeracy scores, by history of joblessness over two generations, mothers' and fathers' experiences, B and K cohorts, Wave 4 - as described in acccompanying text.

Notes: B cohort: mother (n = 3,103), father (n = 2,594); K cohort: mother (n = 2,930), father (n = 2,460). Maternal history, G1 + G2 vs No history: B cohort: t(2,351) = 8.7, p < .001; K cohort: t(2,261) = 9.3, p < .001. Maternal history, G1 + G2 vs G2 only: B cohort: t(426) = 2.4, p = .017; K cohort: t(404) = 2.1, p = .038.

Again, this may be suggestive of an additive effect of generational joblessness, where the experience of two generations of joblessness is associated with poorer academic performance than having just one. Also noteworthy is the size of the difference - with an average score of -0.6, children in families with two generations of joblessness were performing around half a standard deviation lower than children with no family history of joblessness, which is a medium effect.

Intergenerational separation and children's academic performance

The relationship between family separation and children's academic performance is shown in Figure 4.5 for literacy and Figure 4.6 for numeracy, with very similar results shown for both academic domains. For both cohorts, there were no discernible differences in literacy and numeracy scores between children with no family history of separation and those whose maternal grandparents (only) had separated. Children whose parents (only) had separated, however, performed significantly worse, with scores around -0.30 of a standard deviation lower than children who had no family history of separation (e.g., no history = 0.1; G2 only = -0.2). The average scores for children with two generations of family separation were significantly worse compared to children whose parents only had separated, with the exception of the K cohort on the numeracy measure.

Figure 4.5: Children's literacy scores, by history of family separation over two generations, mothers' and fathers' experiences, B and K cohorts, Wave 4

Children's literacy scores, by history of family separation over two generations, mothers' and fathers' experiences, B and K cohorts, Wave 4 - as described in acccompanying text.

Notes: B cohort: mother (n = 3,218), father (n = 3,122); K cohort: mother (n = 3,100), father (n = 3,049). Maternal history, G2 only vs No history: B cohort: t(2,389) = 6.1, p < .001; K cohort: t(2,355) = 6.2, p < .001. Maternal history, G1 + G2 vs G2 only: B cohort: t(582) = 2.1, p = .035; K cohort: t(660) = 2.0, p = .046.

Figure 4.6: Children's numeracy scores, by history of family separation over two generations, mothers' and fathers' experiences, B and K cohorts, Wave 4

Children's numeracy scores, by history of family separation over two generations, mothers' and fathers' experiences, B and K cohorts, Wave 4 - as described in acccompanying text.

Notes: B cohort: mother (n = 3,173), father (n = 3,078); K cohort: mother (n = 3,017), father (n = 2,960). Maternal history, G2 only vs No history: B cohort: t(2,355) = 5.8, p < .001; K cohort: t(2,288) = 6.7, p < .001. Maternal history, G1 + G2 vs G2 only: B cohort: t(573) = 2.0, p = .043; K cohort: t(646) = 1.0, p = .301.

Again, these results suggest that the experience of grandparents matters for the development of children, though the effects of separation are somewhat smaller than those shown for joblessness in the grandparent generation.

4.7 Summary and discussion

The aim of this chapter was to examine the experience of intergenerational separation and joblessness, and to document how these experiences relate to the key child development indicators of social-emotional wellbeing and academic performance. Findings from this chapter suggest that the echoes of early disadvantage in the grandparent generation can be heard in the continuity of family joblessness in the study child's family. These disadvantages are related to children's developmental outcomes at 6-7 and 10-11 years of age.

Consistent with previous research, the percentage of LSAC families that were ever lone-parent families was substantially higher where either set of the study child's grandparents had separated, and higher still if both the maternal and paternal grandparents had separated (e.g., Amato & Cheadle, 2005; Amato & DeBoer, 2005; Wolfinger, 2005, 2011). Similarly, the proportion of parents who were jobless for at least one wave was much higher where either or both the maternal and paternal grandparents had also experienced family joblessness, compared to families where neither set of grandparents had this experience. There were clear continuities in the experiences of joblessness and separation across the grandparent and parent generations.

Broadly, the continuity in these intergenerational relationships suggests that the LSAC children whose grandparents or parents have experienced separation or joblessness may themselves face a greater risk of separation and joblessness as adults. Indeed, already by age 6-7 and 10-11 years, the children in families who have experienced persistent intergenerational disadvantages have already fallen substantially behind their peers with respect to their academic performance and social-emotional development. These findings show that intergenerational disadvantage is pervasive and their effects upon the youngest generation of a family begin early.

We have also shown that the legacy of grandparents extends beyond their own children and into the next generation. The social-emotional and academic outcomes for study children whose grandparents experienced family joblessness were worse compared to children whose grandparents had not experienced joblessness. This pattern held irrespective of whether the study child's own parents had experienced joblessness. A similar pattern was found for separation, though this held only for families where the parents had separated.

The analyses in this chapter have some limitations. Firstly, retrospective reports are fallible, so there is likely to be greater measurement error in the reporting of the history of family joblessness and separation, particularly for fathers' experiences, which were often reported by the mother as a proxy. However, the history of significant life events such as parental divorce and an extended period of family joblessness are likely to be recalled with fewer errors than other types of events that are more transitory in nature (e.g., Mitchell, 2010), and therefore many studies employ the life history method of recall of significant life events (e.g., the Household Income and Labour Dynamics in Australia survey; see Wooden & Watson, 2007). Secondly, the associations between family joblessness and separation in the grandparent generation and children's social-emotional problems, numeracy and literacy have not taken into account any other "third" variables that may underlie this association, and therefore may overestimate the strength of the relationships reported in this chapter. Other factors from the grandparent generation - such as grandparents' mental health problems or alcohol or drug use - could be confounding variables. However, it would be a mistake to use contemporaneous socio-demographic variables such as socio-economic status in the child's family as covariates, as these types of variables are likely to be "outcomes" of intergenerational disadvantage in the parents' family of origin. (For a discussion of the issues in estimating the causal effect of intergenerational disadvantage, see Sharkey & Elwert, 2011).

While these analyses provide an important first step towards understanding the persistence of multigenerational inequality, very little is still known about this, either in Australia or internationally. Studies examining the influences of two or more generations of family separation and joblessness on children's outcomes are rare. The propensity for children to be living in a jobless lone-parent family at some point in time has been, and still is high in Australia (Whiteford, 2009). Rates of maternal employment and education have increased over time (Hayes et al., 2010), which has co-occurred with some reduction in jobless lone-parent families, but the rate of joblessness is still two in five of lone-parent families.

Further analyses of LSAC data in the coming waves could also examine the combined influence of joblessness and separation of grandparents on their grandchildren. Mare (2011), in a recent review on multigenerational inequality, also suggested that when the occupational status of both parents and the returns to education that women are enjoying are accounted for, the likelihood of an intensification of intergenerational inequality will increase. "Matthew effect" aside,16 one of the important lessons from this chapter is that history is not destiny, and though there are echoes of disadvantage, there are also many children who do not follow the pattern of intergenerational disadvantage. Parents' own efforts to overcome the disadvantages they encountered while growing up may be one possible explanation for this resilience, but an examination of the role that social institutions and social policies have played over the years may also provide clues about how to improve children's life chances.

4.8 References

Amato, P. R., & Cheadle, J. (2005). The long reach of divorce: Divorce and child well-being across three generations. Journal of Marriage and Family, 67, 191-206.

Amato, P. R., & DeBoer, D. (2001). The intergenerational transmission of marital instability across generations: Relationship skills of commitment to marriage? Journal of Marriage and Family, 63, 1038-1051.

Belsky, J., Conger, R., & Capaldi, D. M. (2009). The intergenerational transmission of parenting: Introduction to the special section. Developmental Psychology, 45,1201-1204.

Biblarz, T., & Raftery, A. (1993). The effects of family disruption on social mobility. American Sociological Review, 58, 97-109.

Bjorklund, A., & Chadwick, L. (2003). Intergenerational income mobility in permanent and separated families. Economics Letters, 80, 239-246.

Bratberg, E., Rieck, K., Marshall, E., & Vaage, K. (2011). Intergenerational earnings mobility and divorce (Working Papers in Economics No. 09/11). Bergen, Norway: University of Bergen, Department of Economics.

Cohen, J. (1988), Statistical power analysis for the behavioral sciences (2nd Ed.). Hillsdale, NJ: Lawrence Earlbaum Associates.

d'Addio, A. (2007). Intergenerational transmission of disadvantage: Mobility or immobility across generations? A review of the evidence for OECD countries (OECD Social, Employment and Migration Working Papers No. 52). Paris: OECD.

Department of Prime Minister and Cabinet. (2009). A stronger, fairer Australia: National statement on social inclusion. Canberra: Department of Prime Minister and Cabinet.

D'Onofrio, B. M., Turkheimer, E., Emery, R. E., Harden, K. P., Slutske, W. S., Heath, A. C. et al. (2007). A genetically informed study of the intergenerational transmission of marital instability. Journal of Marriage and the Family, 69(3), 793-809.

Fertig, A. R. (2007). Is intergenerational earnings mobility affected by divorce? (Center for Research on Child Wellbeing Working Paper 2002-04). Princeton, NJ: Princeton University.

Frazer, H., & Marlier, E. (2007). Tackling child poverty and promoting the social inclusion of children in the EU: Key lessons synthesis. Vienna: Peer Review and Assessment in Social Inclusion, European Commission.

Funder, K., Harrison, M., & Weston, R. (1993). Settling down: Pathways of parents after divorce. Melbourne: Australian Institute of Family Studies.

Goodman, R. (1997). The Strengths and Difficulties Questionnaire: A research note. Journal of Child Psychology and Psychiatry, 38, 581-586.

Gray, M., & Baxter, J. (2011). Parents and the labour market. In Australian Institute of Family Studies, The Longitudinal Study of Australian Children Annual statistical report 2010 (pp. 29-41). Melbourne: AIFS.

Gray, M., & Baxter, J. (2012). Family joblessness and child well-being in Australia. In Investing in children: Work, education, and social policy in two rich countries. Washington, DC: Brookings Institution.

Gray, M., Taylor, M., & Edwards, B. (2011). Unemployment and the wellbeing of children aged 5-10 years. Australian Journal of Labour Economics, 14, 153-172.

Hayes, A., Weston, R., Qu, L., & Gray, M. (2010). Families then and now: 1980-2010 (Facts Sheet). Melbourne: Australian Institute of Family Studies.

Mare, R. D. (2011). A multigenerational view of inequality. Demography, 48, 1-23.

McDonald, P. (Ed.). (1986). Settling up: Property and income distribution on divorce in Australia. Melbourne: Prentice-Hall of Australia.

Mellor, D. (2005). Normative data for the Strengths and Difficulties Questionnaire in Australia. Australian Psychologist, 40(3), 215-222.

Merton, R. K. (1968). The Matthew effect in science. Science, 159(3810), 56-63.

Mitchell, C. (2010). Are divorce studies trustworthy? The effects of survey nonresponse and response errors. Journal of Marriage and the Family, 72, 893-905.

National Center for Education Statistics (n. d.). Early Childhood Longitudinal Study: Kindergarten (ECLS-K). Washington, DC: Department of Education.

Organisation for Economic Co-operation and Development. (2007). Babies and bosses: Reconciling work and family life. Vol. 5: A synthesis of findings for OECD countries. Paris: OECD.

Potter, D. (2010). Psychosocial well-being and the relationship between divorce and children's academic achievement. Journal of Marriage and Family, 72, 933-946.

Sharkey, P., & Elwert, F. (2011). The legacy of disadvantage: Multigenerational neighborhood effects on cognitive ability. American Journal of Sociology, 116, 1934-1981.

Whiteford, P. (2009). Family joblessness in Australia. Canberra: Social Inclusion Unit, Department of the Prime Minister and Cabinet.

Wolfinger, N. H. (2005). Understanding the divorce cycle: The children of divorce in their own marriages. New York: Cambridge University Press.

Wolfinger, N. H. (2011). More evidence for trends in the intergenerational transmission of divorce: A completed cohort approach using data from the General Social Survey. Demography, 48(2), 581-592.

Wooden, M. & Watson, N. (2007). The HILDA survey and its contribution to economic and social research (so far). The Economic Record, 83, 208-231.

Footnotes

1 Fertig (2007) reported that this is not due to the absence of the father from the household. Their results from sibling fixed effects analyses suggested that when fixed unobserved factors were taken into account, there was no correlation between fathers' and children's earnings based on whether the children grew up in separated or intact families. Factors that predispose parents to divorce and affect earnings are the likely explanation for this; for example, it is possible that mental illness predisposes parents to divorce and also can affect their earnings.

2 The association between grandparents' divorces and grandchildren's outcomes was explained by family characteristics of the parents, including lower education, more marital discord, more divorce and greater tension in parent-child relationships.

3 Moreover, many three-generation studies have been of relatively small groups of children living in particular areas of the US (Belsky, Conger, & Capaldi, 2009).

4 In this chapter, the LSAC study child is always referred to as the child, the parent is always the study child's parent and the grandparent is always the grandparent of the study child.

5 It is useful to get a sense of the years when parents were 14 years of age. Based on the median age when mothers of the B cohort were 14 years of age, it was 1987. As might be expected, fathers in the B cohort and mothers and fathers in the K cohort were older on average; the median years when they were 14 years of age were 1986, 1987 and 1984 respectively.

6 At Wave 4, almost all PLEs were the biological parent of the study child (99% for the K cohort, and nearly 100% for the B cohort).

7 When asked to specify who the "someone else" was, 21% were step-fathers; 12% were grandparents; 20% were other relatives, including aunts, uncles, siblings or a family friend; 20% said they lived in a state home or a foster home; 10% said themselves; 5% relied on government assistance, such as pensions; and 12% were coded as "other".

8 Age when separation occurred is not addressed in this chapter. Within the group in which the grandparents had separated, the majority of separations occurred before the parent was 14 years old (72% for mothers and 61% for fathers), and by age 18 this had increased to 87% of mothers and 79% of fathers.

9 As the data from Wave 2 were based on self-report by Parent 2 (rather than Parent 1 responding on behalf of Parent 2), Wave 2 data were given preference and then supplemented with data from Waves 3 and 4 where necessary. As with the grandparent employment data, data on family separation from fathers and PLEs were combined. If data were collected from both a father figure (e.g., step-father) and a male PLE, the experiences of the male PLE were used to represent the experience of fathers.

10 For example, in the B cohort analytical sample for this chapter, 1,004 of 4,274 children were ever in a lone-parent household. Over 99% of these cases resulted from a separation.

11 Prior to 1988, the Spousal Maintenance Scheme was in effect. Based on a survey conducted in 1987, previous research published by AIFS suggests that the median maintenance per child paid to the mother at that time was $20 per week, as reported by separated mothers (Funder, Harrison & Weston, 1993).

12 As noted previously in the analytic sample for the B cohort, for example, over 99% of cases of ever lone-parent families resulted from a separation.

13 Of two-parent families who were in jobless households for 3-4 waves, 9% were those where both parents' families had been jobless in their family of origin (the grandparents' families), compared to 3% where neither parent had experienced joblessness in their family of origin.

14 For example, at Wave 1, 68% of B cohort parents who had not separated had lived together for 5 years or more, compared to 98% of K cohort parents who had not separated.

15 According to the Goodman's (1997) cut-off standardisation, approximately 20% of the population sample is expected to be in the borderline and abnormal bands. With LSAC data, however, only 11% (B cohort) to 13% (K cohort) of children fell into these categories (at the 14-point cut-off). According to Mellor (2005), on the other hand, the Australian banding for "borderline" (top 20%), is 12 points for Parent 1 reports (14 for self-reports), which corresponds with the LSAC data (the top 20% of LSAC is around 12 points for both cohorts).

16 The "Matthew effect" is a reference to the rich getting richer through the use of their greater economic and social capital, and the poor consequently getting poorer, as first noted by Robert Merton (1968). The reference is derived from the Gospel of Matthew: "For to all those who have, more will be given, and they will have in abundance; but from those who have nothing, even what they have will be taken away" (Matthew 25:29).

Top