Maternal Employment and Parent-Child Relationships in Single-Parent Families of Low-Birth-Weight Preschoolers

Abstract

Background

The influence of premature birth of an infant in female-headed, single-parent families together or in conjunction with family environment factors, such as employment of the mother, on the mother-premature child relationship has not been considered in past studies.

Objectives

To explore differences in parent-child and family relationships for employed and nonemployed single mothers of low–birth-weight (LBW) and full-term preschool children and to describe the relationships of the mothers employment status, employment history, and employment attitude-behavior consistency to parent-child and family relationships.

Methods

Single mothers with LBW (n = 60) and full-term (n = 61) preschool children provided data on their employment situation, the Parenting Stress Index, the Feetham Family Functioning Survey, and the Home Observation for Measurement of the Environment.

Results

Employed mothers had more positive perceptions and provided more enriching home environments for their children. Greater attitude-behavior consistency was associated with more positive perceptions of the parental role.

Conclusion

Thus, in single-parent families, employment and consistency are positive influences on the mother-child relationship.

Concerns about the effects of premature birth of an infant on the mother-child relationship and the family system have been voiced by researchers and clinicians in nursing and other disciplines, although no differences in quality of attachment based solely on birth status have been found (Easterbrooks, 1989). Concerns have also been raised about the mother-child relationship in female-headed, single-parent families. The influence of both factors together or in conjunction with family environment factors, such as employment of the mother, on the mother-premature child relationship has not been considered. Thus, the purpose of this study was to explore differences in parent-child and family relationships among employed and nonemployed single mothers of LBW and full-term preschool children and to describe associations between employment-related variables and parent-child and family relationships.

Effects of maternal employment on mother-child and family relationships have been investigated in two studies of families with preterm infants. Cohen (1978) found that nonemployed mothers showed more positive attentiveness, toddlers vocalized more to nonemployed mothers, and nonemployed-mother-child dyads displayed more positive reciprocal interactions. However, when single-parent families were omitted from the analyses, differences between employed- and nonemployed-mother groups were no longer significant. In a longitudinal study of two-parent families with a premature infant, Youngblut, Loveland-Cherry, and Horan (1991, 1993, 1994) found no differences in family cohesion or adaptability for employed- and nonemployed-mother families at 3, 9, 12, or 18 months postpartum. Employed mothers were less satisfied with their families at 12 and 18 months (Youngblut et al., 1993, 1994), but not at 3 or 9 months (Youngblut et al., 1991, 1993). Number of hours employed was not related to family functioning at 3, 9, 12, or 18 months.

The effects of maternal employment status in studies that include single-parent families vary, depending on the degree of statistical or design control for number of parents in the family and other relevant variables used, such as maternal education and family income. When simple bivariate relationships are examined, the effects of maternal employment for children are sometimes positive and sometimes negative. In a small sample of single-parent and two-parent families Weinraub and Wolf (1983), using no statistical control, found that greater number of hours employed was related to greater maternal control of preschool children, but not to quality of mother-child communication or maternal nurturance. In their study of low-income families (48% single-parent families) with second graders, Vandell and Ramanan (1992) found that quality of home environments was positively related to number of hours employed in the child’s first 3 years and at the time of study; again no statistical control was used. MacKinnon, Brody, and Stoneman (1982) found that divorced employed-mother families scored lower on all Home Observation for Measurement of the Environment (HOME) subscales than either married employed- or married nonemployed-mother families. Because this study did not include a group of divorced nonemployed-mother families, it is not clear whether the differences were related to employment status or marital status.

When the sample was limited to single-parent families, consistent findings about the positive effects of maternal employment were obtained. Guidubaldi, Cleminshaw, Perry, Nastasi, and Lightel (1986) found that maternal employment in divorced families was related to better mother-son relationships but was not associated with quality of the mother-daughter relationship. These researchers did not identify the statistics used to analyze their data, so it is not clear if possible covariates were statistically controlled. Kurtz and Derevensky (1994) found that divorced families with school-age children scored higher on the Active Recreational subscale of the Family Environment Scale (FES) when mothers were employed than when mothers were not employed. Scores on the other FES subscales were not significantly different between employed- and nonemployed-mother families. Although Kurtz and Derevensky attribute this difference to the employed mother’s perception of herself as the psychological parent, it is likely that this difference reflects the greater financial resources that the single mother’s employment provides, since mothers’ education and family income were not controlled.

In the one study that controlled family form and family income by design and mother’s education statistically, effects of maternal employment disappeared. Jackson (1993) found that low-income, Black, single, employed mothers’ perceptions of their 3- or 4-year-old child’s behavior were not related to number of hours employed or their preference for employment, when the child’s sex and mother’s education were controlled.

Negative effects of divorce but positive effects of maternal employment on mother-child relationships were found in several studies. Poehlmann and Fiese (1994) found less enriching home environments for toddlers in divorced homes than intact homes and in nonemployed-mother families than employed-mother families, controlling for socioeconomic status. In a study by Pett, Vaughan-Cole, and Wampold (1994), mothers’ employment status was not related to quality of mother-child interaction; but divorced mothers were more controlling and less supportive in their interactions with their preschool children than married mothers. In that study, employed mothers were more accepting of their child’s behavior than nonemployed mothers, regardless of whether socioeconomic status was controlled.

Effects of consistency between mothers’ employment attitudes and employment behaviors on mother-child and family relationships have rarely been considered. In a longitudinal study of two-parent families with preterm infants, inconsistent mothers were less satisfied with family than consistent mothers at 18 months (Youngblut et al., 1994), but not at 3, 9, or 12 months postpartum (Youngblut et al., 1991, 1993). Greater choice about their employment status was related to mothers’ perceptions of greater family cohesion and satisfaction with family at 18 months. When investigated separately for employed and nonemployed mothers, perceived choice was positively related to family cohesion at 3 months and to satisfaction with family at 9 and 12 months for nonemployed mothers only (Youngblut et al., 1993). In a sample of primarily White, single, employed-mother families, Goldberg, Greenberger, Hamill, and O’Neil (1992) found that more negative beliefs about the costs of employment for children (representing more employment attitude/behavior inconsistency) was associated with more problem behavior in the children, controlling for a variety of work and family demands and demographic and social support variables.

The aims of this study were to (a) explore differences in home environments, the degree of stress in the mother-child relationship, and the mother’s satisfaction with family functioning for employed and nonemployed single mothers of LBW and full-term preschool children; and (b) describe the effects of the mother’s employment status, employment history, and employment attitude-behavior consistency on the home environment, stress in the mother-child relationship, and satisfaction with family functioning. In the larger study, data were collected from a sample of female-headed single-parent families and two-parent families with LBW and full-term preschool children twice, about 1 year apart. Data from the first data collection time point with the single-parent families only are reported here.

Method

Sample

The sample of 121 female-headed, single-parent families included 61 families with full-term preschoolers and 60 families with LBW preschoolers. These two gestational age (LBW vs. full-term) groups were balanced by the child’s chronological age (3, 4, or 5 years old). All families with appropriately aged LBW preschoolers from three Level III Neonatal Intensive Care Units (NICUs) and a systematic random sample of families with full-term preschoolers from two normal newborn nurseries were identified. Families were contacted without knowing family structure (single- or two-parent family), and those who refused to participate usually did not provide this information. Thus, response rate was calculated for the sample as a whole at the time of completion of the single-parent sample. The “willing to participate” rate was 71%. A full description of the families is reported elsewhere (Youngblut, Singer, Madigan, Swegart, & Rodgers, 1997).

The LBW child in the family met the following criteria: born before 36 weeks gestation weighing less than 2,500 grams; appropriate weight for gestational age; and hospitalized for at least 1 week in a Level III NICU. The full-term child in the family met the following criteria: birth between 38 and 42 weeks gestation; discharged home with the mother after birth; and without preterm siblings that were born within 10 years of the study child’s birth. Mothers who were told their LBW or full-term child had more than a 2-year developmental delay were excluded; none of the single-parent families was excluded based on this criterion. Single-parent families were eligible to participate if the mother was single (not living with a man in the father role) at the time of recruitment and had not lived with a man in the father role for at least 6 months before recruitment (Youngblut et al., in 1997).

Mothers’ mean age was 29.5 (SD = 6.5) at the time of the study. Most mothers were African American (66.1%), had completed high school (76.9%), had never been married (70.2%), and had sole custody of the study child (92.6%). Almost half (46.3%) had been single since the study child’s birth. The average number of children in the family was 2.5 (SD = 1.44).

Thirty-four women were employed at the time of the study for an average of 34.3 (SD = 10.23) hours per week, but half (n = 61) had a history of employment during the index child’s life. Women reported their “usual” occupation as homemakers (n = 51), unskilled (n = 9), skilled or semi-skilled (n = 24), clerical or sales (n = 17), and professionals (n = 20), based on Hollingshead’s classifications (Hollingshead, 1975). Most women reported total family incomes of less than $10,000 (68.6%) for the previous year. Other sources of financial support included Aid to Families With Dependent Children, Social Security Income, or “Welfare” (n = 74); child support (n = 21); and relatives (n = 2). Thus, the sample consisted of primarily low-income families. Mothers in the preterm and full-term groups were not significantly different on maternal age, education, income (mother’s portion or total), socioeconomic status, and proportion of the child’s life with either a single or employed mother based on t test comparisons and on maternal race and current employment status with χ² analysis.

Half of the study children were male (52.9%); 54 (44.6%) were first-borns. The LBW group had a mean birth-weight of 1,444 g (SD = 527.2) and a mean gestational age of 30.5 weeks (SD = 3.17), whereas the full-term group had a mean birth-weight of 3,331 g (SD = 514.2) and a mean gestational age of 39.6 weeks (SD = 1.60). Average NICU stay for the LBW children was 46.1 days (SD = 33.34). Few preschoolers had experienced complications of prematurity; 13 had an intraventricular hemorrhage (10 -Grade I, and one each with Grades II, III, and IV), 2 had cerebral palsy, and 2 had bronchopulmonary dysplasia. Number of hospital stays and total days hospitalized, excluding hospitalization at birth, were significantly higher for preterms (M = 1.0, SD = 2.45, and M = 16.64, SD = 52.59, respectively) than full-terms (M = .15, SD = .36, and M = .93, SD = 2.61), t (60.5) = 2.63, p = .01, and t (57.28) = 2.27, p =.03, respectively (unequal variance estimates).

Instruments

Mothers indicated their employment status as either employed or not employed and provided the number of hours per week that they currently were employed. In addition, mothers described their patterns of employment since the study child’s birth, including when they began employment after the child’s birth, the jobs they had held, the duration of each job, and the number of hours per week employed in each job, on a Life History Calendar (Freedman, Thornton, Camburn, Alwin, & Young-DeMarco, 1988). The Life History Calendar constructed for the study contained 5 segments, each with 12 blocks representing the 12 months in each of 5 years beginning with the study child’s birth. Major life events, such as residential moves, births, deaths, and hospitalizations of the study child, were also recorded to aid the mothers’ memory.

Maternal employment attitude-behavior consistency was measured with a computed discrepancy variable (consistency). To calculate the discrepancy variable, mothers were asked to indicate the number of hours per week they would prefer to work outside the home. Consistency was then computed by subtracting the actual work hours from the preferred work hours and taking the absolute value. Lower scores on this computed variable indicated greater attitude-behavior consistency. Employed mothers in this sample reported greater consistency than nonemployed mothers, M = 9.5, SD = 6.48; and M = 25.9, SD = 13.13, respectively (Youngblut et al., 1997).

HOME (Caldwell & Bradley, 1984), a measure of the quality of the home environment, contains eight subscales: (a) learning materials, (b) language stimulation, (c) physical environment, (d) responsivity, (e) learning stimulation, (f) modeling and encouraging of social maturity, (g) variety of stimulation, and (h) acceptance. Based on the interview with the mother and direct observation of the mother-child dyad, interviewers rate each item as present, scored “1,” or absent, scored “0.” Higher scores indicate a more stimulating, affirming home environment. Construct validity of the HOME is supported by its moderately strong correlations with cognitive development scores (Caldwell & Bradley). Caldwell and Bradley reported internal consistency reliabilities of .93 for the total scale and from .53 to .88 for the subscales. Alphas in the current study were .93 for the total scale and from .60 to .85 for the subscales. Interrater reliability in the current study was assessed monthly for all possible pairs of interviewers and ranged from 78% to 100% (M = 94.2%). Interviewer retraining occurred after any month with an interrater reliability below 80%.

The Parenting Stress Index (PSI) (Abidin, 1990) measures the degree of stress (dysfunction) involved in the parent-child system. The child domain contains six subscales: child adaptability/plasticity, acceptability of child to parent, child demandingness, child mood, child distractibility/hyperactivity, and child reinforces parent. The parent domain contains seven subscales: parent attachment, restrictions imposed by parental role, parent’s sense of competence, parent depression/unhappiness/guilt, parental health, social isolation, and relationship with spouse. The relationship with spouse sub-scale was not used. Mothers rated each of the 99 items on a 5-point Likert scale from 1, “strongly agree” to 5, “strongly disagree.” Appropriate items for each domain are summed, so that higher scores indicate higher levels of stress (dysfunction). Abidin considers scores above the 80th percentile for his normative sample to be problematic, indicating significant dysfunction. Construct validity of the PSI is supported by significant correlations between PSI scores and parental anxiety and by group differences between parents of children with and without disabilities (McKinney & Peterson, 1984). Abidin reported internal consistencies of .70 to .84 for the subscales, .90 for the child domain, and .93 for the parent domain. Internal consistency in the current study was .54 to .79 for the subscales, .88 for the child domain, and .93 for the parent domain.

The Feetham Family Functioning Survey (FFFS) (Roberts & Feetham, 1982) measures the parent’s satisfaction with dyadic and higher order relationships within the family and between the family and others. Respondents rate two questions — (a) How much is there now? and (b) How much should there be? — for each of the 25 items on a 7-point scale ranging from 1 “little” to 7 “much.” A discrepancy score is created by subtracting responses to a from b and adding the absolute values. Items in the FFFS about relationships with spouse were problematic for single mothers, despite modifications to the instrument’s instructions recommended by Feetham (personal communication, July 1993). Thus, the total FFFS score was computed without the spouse questions, yielding a possible range of 0 to 90 for the discrepancy-scores for the 15-item modified scale. Lower scores indicate greater satisfaction with family functioning. Construct validity of the 25-item FFFS is supported by its strong correlation (r = −.68) with Pless and Satterwhite’s Family Functioning Index (Thomas & Barnard, 1986). Roberts and Feetham report internal consistency of .81 and test-retest reliability of .85 for the discrepancy score. Internal consistency for the current study was .77.

Procedure

After approval by the appropriate Human Subjects Review committees, families were sent a letter briefly describing the study and identifying their assigned interviewer. The interviewer made a follow-up telephone call to screen for inclusion criteria and determine the family’s interest in participating in the study. Parent self-complete and interview data and child developmental testing were collected in the family’s home. Means and standard deviations for variables central to this report are in Table 1.

TABLE 1.

Descriptive Statistics for Main Study Variables

Variable	Mean	SD
Hours employed per week	10.28	17.00
Employment attitude/behavior consistency	21.04	13.82
Proportion of child’s life in employed mother family	0.25	.34
Home Observation for Measurement of the Environment	39.94	9.76
Learning materials	6.71	2.91
Language stimulation	5.91	1.52
Physical environment	5.60	1.96
Responsivity	4.68	1.75
Learning stimulation	3.58	1.41
Modeling and encouraging social maturity	3.01	1.45
Variety of stimulation	6.11	1.89
Acceptance	3.40	1.03
Parenting Stress Index—child domain	106.37	21.70
Child adaptability/plasticity	27.13	6.42
Acceptability of child to parent	13.88	4.86
Child demandingness	20.39	5.14
Child mood	10.91	3.64
Child distractibility/hyperactivity	25.72	5.20
Child reinforces parent	10.70	3.61
Parenting Stress Index—parenting domain	127.15	28.45
Parent attachment	13.85	4.49
Restrictions imposed by parental role	19.27	5.43
Parent’s sense of competence	30.03	6.62
Parent depression/unhappiness/guilt	20.47	6.20
Parental health	12.53	3.73
Social isolation	14.05	4.90
Feetham Family Functioning Survey	15.14	9.06

Results

Analysis of variance (ANOVA) was used to test the main and interaction effects of maternal employment status (employed vs. not employed) and gestational status (LBW vs. full-term) on the child and parent domains of the PSI and the total scores for the HOME and FFFS (Table 2). There were significant main effects of employment status for total HOME scores and the Child Domain of the PSI. None of the other main or interaction effects was significant. Employed mothers provided more positive and affirming home environments and had more positive perceptions of their children than nonemployed mothers.

TABLE 2.

Main and Interaction Effects of Maternal Employment Status and Gestational Status on Parent-Child and Family Relationships

	Groups				Main Effects
	EM LBW	NEM LBW	EM Fullterm	NEM Fullterm	Employment Status	Gestational Status	Interaction Effects
Variable	M (SD) (n = 17)	M (SD) (n = 43)	M (SD) (n = 17)	M (SD) (n = 44)	F	F	F
HOME	41.5 (11.68)	38.7 (10.73)	43.7 (4.47)	36.5 (10.71)	5.67*	.001	1.10
PSI domains:
Child	101.0 (21.0)	112.3 (22.78)	101.3 (19.15)	111.0 (21.74)	5.58*	.01	.04
Parent	123.7 (17.79)	132.2 (31.56)	122.2 (26.56)	132.1 (28.84)	2.51	.02	.02
FFFS	13.3 (5.77)	17.6 (10.38)	14.9 (8.45)	14.7 (8.65)	1.20	.10	1.33

^*p< .05.

HOME = Home Observation for Measurement of the Environment; PSI = Parenting Stress Index; FFFS = Feetham Family Functioning Survey.

Associations between categorical PSI scores and employment status, gestational status, and employment-by-gestational status were explored using χ² analyses. Only the association between employment status and parent domain categorical scores was significant; more nonemployed mothers, but fewer employed mothers, than expected had scores considered problematic (> 80%), χ² (1, N = 121) = 4.15, phi = .19, p = .04.

Although employment had a positive effect for mother-child relationships, it is possible that the significant finding was because of confounding variables. Thus, hierarchical multiple regression was used to explore the relationships between employment-related variables and the child and parent domains of the PSI and the total scores for the HOME and FFFS, controlling for selected child, mother, and family characteristics. The control variables entered in the first stage were gestational status, mother’s education, the number of children in the family, proportion of the index child’s life spent in a single-parent family, the mother’s single category (never married vs. separated, widowed, or divorced), and total family income. In the second stage, the number of hours per week currently employed and the proportion of the child’s life spent in an employed-mother family were added. Employment attitude-behavior consistency was added in the third stage.

Correlations among the nine independent variables were examined for evidence of significant multicollinearity, generally considered to occur when correlations are above .80 (Berry & Feldman, 1985; Lewis-Beck, 1980). None of the correlations exceeded .80. The two highest correlations were between proportion of child’s life with an employed mother and hours per week employed, r = .64, and consistency, r = −.51. Regression results were the same with and without the proportion of child’s life with an employed mother variable. Thus, because of this variable’s theoretical importance, results are reported here with the proportion variable included.

Only the regression equations with the HOME and the PSI Child Domain as the dependent variables were significant (Table 3). Number of children in the family was negatively related to HOME scores; however, none of the employment-related variables were significant. Lower attitude-behavior consistency (higher scores) was related to greater perceived stress in the child domain (less positive attitudes about the child).

Table 3.

Significant Regression of Parent-Child and Family Variables on Employment-Related Variables

		Stage 1	Stage 2	Stage 3
Dependent Variable	Independent Variables	βa	βa	βa
HOME
	LBW vs. fullterm group	.03	.04	.05
	Mother’s education	.08	.07	.07
	No. of children in family	−.40**	−.39**	−.39**
	Proportion child’s life in single-parent family	−.15	−.15	−.15
	Never married vs. separated, widowed, divorced	.19	.18	.16
	Total family income	.03	.006	.02
	Hours employed per week currently		.08	−.005
	Proportion child’s life in employed-mother family		−.002	−.003
	Employment attitude-behavior consistency			−.16
Adjusted R2		.23**	.22**	.23**
PSI Child Domain
	LBW vs. fullterm group	.03	.004	−.005
	Mother’s education	−.12	−.12	−.12
	No. of children in family	.07	.08	.06
	Proportion child’s life in single-parent family	−.05	−.05	−.08
	Never married vs. separated, widowed, divorced	−.02	.007	.04
	Total family income	.01	.10	.07
	Hours employed per week currently		−.38**	−.20
	Proportion child’s life in employed-mother family		.25	.25
	Employment attitude-behavior consistency			.34**
Adjusted R2		−.04	.02	.10*

^aStandardized regression coefficients.

^*p<.05.

^**p<.01.

To investigate which aspects of the mother-child relationship were associated with consistency and whether they differed for employed- and nonemployed-mother families, correlations between the HOME and PSI subscales and consistency were evaluated as post hoc analyses. Correlations between consistency and HOME and PSI subscales were small (r < .23) and nonsignificant for employed mothers. In nonemployed-mother families, lower consistency (higher scores) was related to less stimulation of language and learning behavior (r = −.23, p = .05) and less responsivity (r = −.24, p = .03) on the HOME scale. Non-employed mothers with lower consistency (higher scores) perceived more stress on 8 of the 12 PSI subscales: child acceptability to parent (r = .28, p = .02), child demandingness (r = .30, p = .009), child mood (r = .38, p = .001), child distractibility (r = .33, p = .004), parent attachment (r = .32, p = .005), restrictions imposed by the parental role (r = .30, p = .009), parental sense of competence (r = .26, p = .02), and parental depression/unhappiness/guilt (r = .32, p = .005). Thus, consistency was significantly related to measures of the mother-child relationship in families with nonemployed single mothers, but not in families with employed single mothers.

Discussion

Employed single mothers provided more simulating home environments for their children and had more positive perceptions of their children. However, when sociodemographic factors were statistically controlled, mothers’ employment was not related to any of the mother-child or family measures. This finding is consistent with other studies of single mothers that used statistical control (Jackson, 1993; Pett et al., 1994). The absence of effects for maternal employment when maternal and family factors are controlled suggests that employment status represents the effects of other factors that may vary with employment status, such as number of children (Klerman & Leibowitz, 1994; Youngblut et al., 1997), mother’s education (Klerman & Leibowitz; Mauldin & Meeks, 1990; Shapiro & Mott, 1994), and family income. Thus, maternal employment may have no independent effect of its own but rather reflect mother-child relationship effects of a variety of sociodemographic variables that combine to allow or compel the mother to seek employment.

Employment variables are not related to the FFFS, the family functioning measure used in this study. This is in contrast to previous studies by Youngblut et al. (1993, 1994) where employed mothers were more satisfied with their families than nonemployed mothers measured with the FFFS. Two differences between those studies and this one may account for the difference. First, ages of the index preterm children were 12 and 18 months in the two previous studies, but 3 to 5 years in the current study. Family developmental tasks are different for families with infants and toddlers than for families with preschool children; perhaps effects of family environment factors on family dynamics also differ in amount and type across different family life cycle stages. Second, the previous study samples contained only two-parent families, whereas the current study’s sample was limited to single-parent families. The lack of statistically significant relationships in the current study may be owing to the FFFS measure. Despite the developer’s expectations (Feetham, personal communication, July 1993) and suggested modifications to the instrument’s instructions, single mothers in this study had trouble responding to the items about relationships with spouse. In addition, indicators of family functioning may be different for single-parent families than for two-parent families. Identification of these aspects and inclusion of appropriate items could increase the validity of the FFFS as a measure of family functioning for single-parent families.

Measures of mother-child relationships and family functioning were similar for families with preterms and fullterms. Youngblut and colleagues (1991, 1993, 1994) also found no effects of gestational age on family functioning. Despite recruitment from NICU admission records, only a few LBW preschoolers in this sample were severely affected by their early birth. A sample of families with moderately or severely affected preterms may differ from families with fullterms in the quality of the mother-child and family relationships. However, our finding is consistent with the theory by Kaplan and Mason (1960) of parental adjustment to the premature birth of an infant. According to Kaplan and Mason, the final psychological task imposed by prematurity is for the parent to be able to see that the differences between preterms and full-terms apparent in the neonatal period are only temporary. Thus, the lack of differences could be interpreted as an indication that the families in this study have accomplished this final task.

Consistency was related to more positive perceptions (less perceived stress) of the child. Goldberg et al. (1992) also found that greater employment attitude/behavior consistency was related to fewer child behavior problems reported by single mothers. Bivariate correlations between consistency and the HOME and PSI subscales suggest that consistency is most important in families with nonemployed mothers. This finding supports the results of previous studies (Youngblut et al., 1993, 1994) where nonemployed mothers reported greater family cohesion and satisfaction with family when they perceived greater choice in their employment decision. The mechanism for this effect is not known. However, since Hock and DeMeis (1990) found that nonemployed mothers who preferred employment had higher depression scores than other mothers, it is possible that the effect of inconsistency is transmitted through the mother’s emotional or psychological state to the mother-child relationship and perhaps to the child and other family members. Additional research is necessary to explore the path that this effect takes.

In summary, employment status and gestational status had little systematic effect on the mother-child relationship in female-headed, single-parent families. Although quality of the mother-child relationship was similar for families with LBWs and full-terms, it is not clear from this study whether the lack of differences is because premature birth, by itself, does not place the family at risk or because the vigilance and early intervention often provided by nurses and other health care professionals prevent dysfunction. Employment of the single mother was positively related to the quality of the relationship between mother and child; however, controlling for specific sociodemographic variables eliminated these effects. As in previous studies, consistency between employment attitudes and employment behavior was related to more positive mother-child relationships. Thus, consistency appears to be more important than employment status for optimal mother-child outcomes, especially for nonemployed mothers.