NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Institute of Medicine (US) and National Research Council (US) Committee on the Science of Research on Families; Olson S, editor. Toward an Integrated Science of Research on Families: Workshop Report. Washington (DC): National Academies Press (US); 2011.

Cover of Toward an Integrated Science of Research on Families

Toward an Integrated Science of Research on Families: Workshop Report.

Show details

3Studying How Families Cope with Poverty and Economic Stress: The Role of Quantitative and Qualitative Methods

Demographic research is essential in understanding rapidly changing family forms and dynamics, but demographic research alone cannot capture the full and rich complexity of the family. Other kinds of research are needed to understand such issues as the relationships in families or family influences on child health and well-being.

Poverty and economic stress remain realities of daily life for a substantial proportion of American families and children. Recent increases in the number and proportions of families in poverty make imperative the need to understand how these families adapt to adversity. All four presentations described in this chapter examined families that are under economic stress. Also, individual families and families in certain populations react in different ways, generating considerable variation within broader trends.

Studies of families under stress are a particularly good example of the ways in which qualitative and quantitative approaches can be combined to provide a better understanding of developmental processes than can either approach on its own (Yoshikawa et al., 2008). Quantitative research involves the collection or analysis of numeric representations of the world. Survey and questionnaire data as well as biological or physiological data are often analyzed in quantitative units that serve as proxies for phenomena that are often quite complex. Qualitative methods rely on nonnumeric representations of the world—words, texts, narratives, pictures, and observations. As a holistic enterprise that includes the social, neurological, and biological sciences, family research relies on both kinds of data, although particular disciplines may emphasize one form of data collection and analysis over another.

Quantitative and qualitative approaches do not simply offer alternative ways of measuring and understanding reality. Rather, their combination provides a more complete picture of family structures, processes, and relationships. Furthermore, each approach can inform and complement the other through the examination of basic assumptions, theoretical models, and new constructs.


The New Hope Program was a three-year antipoverty demonstration program implemented in Milwaukee, Wisconsin, in the mid-1990s (Duncan et al., 2007; Mistry et al., 2008; Yoshikawa et al., 2006). New Hope offered an alternative approach to the issue of welfare reform, focusing on work-based supports designed to “make work pay” (Duncan et al., 2007). The program's premise was that, if people were working, they should not be poor. It provided income supplements for people working 30 hours or more a week, subsidies for purchase of private health insurance if benefits were not available through employment, child care assistance and subsidies if required, community service job placement, and individualized assistance from program representatives to help find jobs or deal with specific issues. In this way, the program sought to ensure that take-home income was above the poverty line.

The context of low-wage work and its impacts on family functioning and child outcomes are particularly amenable to an approach that mixes quantitative and qualitative methods, said Rashmita Mistry, associate professor of education at the University of California, Los Angeles. She described the Child and Family Study component of the evaluation of the New Hope Program. Funding for the evaluation was provided by several funding agencies, including the MacArthur Foundation's Research Network on Successful Pathways Through Middle Childhood and the National Institute of Child Health and Human Development (NICHD). The interdisciplinary team included an economist, two developmental psychologists, and a cultural anthropologist, and their evaluation drew on three sources of data. The first was administrative records data, such as earnings, earning supplements, welfare assistance (Aid to Families with Dependent Children and Temporary Assistance for Needy Families), food stamps, and Earned Income Tax Credit assistance. The second was survey data from parents of children ages 6 and older and teachers, encompassing 550 families and approximately 900 children, ages 1 to 10 at baseline. The third was an embedded longitudinal qualitative study, covering three years, of 40 randomly selected families with the participants interviewed multiple times per year. “The approach to conducting the interviews was, no tape recorders, no note taking,” said Mistry. “It was engaging the participants in a conversation about a variety of topics.”

The studies presented by Mistry were based on secondary (nonex-perimental) analysis of data collected as part of the New Hope evaluation and were informed by a family economic stress perspective (Conger and Elder, 1994; McLoyd, 1990). The essential idea of this perspective is that economic hardship is an important pathway through which poverty harms children's development. The subjective experience of dealing with financial adversity on a continual basis or the sudden loss of income due to unemployment influences a parent's mental health, provoking stress and depression. This can affect parenting practices, such as nurturance, warmth, and discipline strategies, which in turn influence children's well-being and learning.

The first study Mistry described looked at patterns of income change and related those patterns to indicators of material and psychological well-being assessed at baseline and five years later; quarterly income data were available from 1995 to 2000, along with survey data. One important finding from the quantitative data analysis was that average total income changed little over those five years. However, this overall trend masked important differences in trajectories by sources of income. For example, as welfare assistance dropped, income and other forms of assistance grew. Impacts on material and psychological well-being were measured through such indicators as disruptions of heat or electricity, difficulty paying the rent, having a checking account or a credit card, and the amount of time spent worrying about how to make ends meet.

Parents who reported higher incomes at the start of the study or whose income increased significantly across the five-year period showed lower levels of material hardship and financial worry at the end of the five-year period. However, measures of psychological well-being showed little or no improvement. “You find that there is very little evidence of direct effects of changes in income over the five-year period on these indicators of psychological well-being,” Mistry said. Many of these families were still poor by objective definitions. For many of them, the money they brought home was never enough to meet all their needs.

The aim of the second study was to explore relationships among low-income mothers' management of finances and expenditure demands, family processes, and child well-being. The embedded qualitative analysis involved two primary research questions. First, how do low-income mothers meaningfully distinguish among categories of expenditures? Second, what are the consequences of managing expenditure demands for mothers' psychological well-being?

“My natural tendency in looking at these data, which are all narrative and interview based, was to find some kind of coding scheme, apply that coding scheme to the data, and then quantify the data,” said Mistry. “It is about epistemology. It is the way I was trained to work with data, even when it is open-ended data. But at the end of the day, I looked at our results and realized that this doesn't tell a story. What got lost were these women's experiences. What got lost was humanizing this experience and being able to tell the story as they saw it.”

Mistry and her colleagues started again. They received training in how to work with qualitative data. The reanalysis of the data led to a stronger telling of the story, from the women themselves, of the impact of economic pressure. Mistry also expressed her preference to work with collaborators for whom qualitative research was their principal form of training (her chief collaborator on this project was a cultural anthropologist).

One theme that became quickly salient was the distinction that women were making between the pressure to meet basic needs, things like paying rent and putting food on the table, and having a little money to engage in “extras” that had important psychological consequence for them. These might seem small and insignificant to others but were important to these families, such as taking their kids to McDonald's for a meal or, for mothers, getting their nails done occasionally. This was tied explicitly to these women's concept of what it means to be a mother. “It wasn't just about the income they brought in through government supports and earnings. It was really about being able to utilize their resource pools.” Falling behind was a constant source of worry and anxiety. Affording modest extras provided a sense of fulfillment and happiness. Women also talked about taking on additional jobs, cutting back on expenditures, and relying heavily on kinship and friendship networks to meet their children's needs.

In closing, Mistry asserted that the value of a mixed-methods approach in the studies she described is the ability to cross-validate findings and reduce bias. A mixed-methods approach also can help to identify processes of transmission and link processes to outcomes. A final advantage of mixed-methods approaches is that they can help to replicate and generalize findings, helping to make research more amenable to policy makers. However, such approaches are also resource-intensive, requiring extra training, finances, and time.


Rebekah Levine Coley, associate professor of applied and developmental psychology at Boston College, discussed the Three City Study, begun in the late 1990s to assess the well-being of children and families following welfare reform in Boston, Chicago, and San Antonio. This multidisciplinary, multimethods study was led by a team of scholars from multiple disciplines, including developmental psychologists, urban health and family sociologists, and labor economists. The Three City Study had multiple components, including surveys, direct assessments, structured observations, participant observation ethnographies, and collection of administrative data. The team selected one focal child per family and studied two cohorts of children, with three waves of survey data spread over six years (Angel et al., 2009).

Coley described several lessons from the study that relate to the use of multimethod studies in family research. First, in choosing which ages to study, the priorities for the team were to focus on the developmental stages that are most responsive to environmental influences or insults and stages in which developmental contexts can be measured. They decided that if they tried to select children from birth to age 18, both the sample and the methods would be diluted. Also, if trying to assess all children from birth to 18, it would be difficult to train interviewers and to make a coherent survey instrument that is developmentally appropriate over all those ages (Angel et al., 2009).

Ultimately, the team decided to focus on two separate cohorts: zero to 4-year-olds and 10- to 14-year-olds. These cohorts were followed for six years, so ultimately the entire span of childhood from birth to age 20 was covered. But by focusing on a narrower age range, developmentally appropriate measures could be targeted toward particular age groups. Also, focusing on adolescents allowed the team to rely on considerable self-reporting.

An important component of the Three City Study was the Embedded Developmental Study (EDS). It was conducted with all the 2- to 4-year-old children and their families from the main survey sample. The project team decided that this age group was particularly important in considering welfare reform, since it is an age at which parents can have particular difficulties combining parenting and employment.

Very young children cannot report on their own well-being, so the EDS included four separate components: additional interviews with mothers, videotaped child-mother activities, interviews with biological fathers, and observations of child care practices and interviews with child care providers. The study generated “a huge amount of information,” according to Coley.

The first wave of the EDS was conducted when the children in this sample were 2 to 4 years old, and the second wave when they were about 3 and a half to 6. In the third wave, when the children were in elementary school, some of these components were less appropriate, so the study conducted interviews with teachers and collected school administrative data.

Another major component of the Three City Study was a participant ethnography with 256 families over a three-year period. In contrast to the New Hope Study, the ethnography was not conducted with a subset of the survey families. The ethnography families were selected from a different sample of families in the same neighborhoods and from the same racial and ethnic groups as the survey families. One reason for this strategy was to reduce the burden on respondents. Another reason was to avoid cross-contamination between the ethnography and the survey.

The ethnography was connected to the survey in numerous ways. Like the EDS, it focused specifically on families with preschool-age children. It also developed modules that mimicked or paralleled the modules or the topics covered in the survey, creating opportunities for coordination between the ethnography and the survey.

The separation of the ethnography from the survey had both pluses and minuses. It successfully lowered respondent burden for the ethnography families and for the survey families, but it also reduced the potential for coordination between the ethnographies and the surveys. However, it led to different types of mutual influence between the research teams (Angel et al., 2009).

The sources of information about family members varied by developmental period and by role in the family. For young children, information was obtained from such sources as direct assessments, structured observations, and parent and child care provider interviews. Adolescents were able to provide considerable information themselves, and the project interviewed adolescents directly as well as their caregivers. Also, for such topics as adolescent sexual risk behaviors, it is reasonable to presume that adolescents themselves would be better reporters than their parents. Similarly, there is agreement in the field that fathers' reports of their own parenting behavior are preferable to relying solely on mothers' reports. But this approach has strengths and weaknesses. Response rates for fathers are lower, which means that relying on fathers' reports will produce a smaller sample and introduce selection bias into the sample (Angel et al., 2009).

Conceptual considerations also can dictate whom to ask about different concepts. For example, for parental monitoring and knowledge about adolescents' activities, parents and youth are likely to have different perspectives. The Three City Study and other studies have found correlations in the range of .2 or .3 between parent and youth reports on measures of adolescent externalizing behaviors or parenting behaviors, which is not very high. Some people interpret that to mean that the validity of the measures is poor. Another perspective is that taking different perspectives into account is important. With behaviors like parental monitoring and knowledge, it might be more important to look at what youth think their parents know about their actions and behaviors than to consider what parents think they know, since perceptions have a stronger influence on a young person's behavior.

A third way to think about choices among sources of family research data is to test their validity. Which measure has better predictive validity to behaviors or outcomes of interest? For example, when mothers' and fathers' reports on fathers' parenting behaviors were compared, the fathers' reports were slightly more strongly related to children's cognitive skills (Hernandez and Coley, 2007). Moreover, a composite of fathers' and mothers' reports had the strongest predictive validity to children's outcomes.

Triangulating information across different sources of reports in a family has other strengths. Obtaining different perspectives from different people provides increased reliability of measurement and potentially increased predictive validity. However, there are clear weaknesses. Information may conflict among reporters and over time. “If you ask multiple people in a household what the family structure is, you will get multiple answers,” Coley said. For example, when mothers and fathers were asked whether the father lived in the household with the child, there was an 11 percent discrepancy rate in the responses.

As another example, when adolescents and mothers were asked to report on the father's involvement with the adolescent, discrepancies arose about whether the father was alive or dead. Coley observed: “Some fathers who in the first wave were reported to be deceased in the second wave had come back to life.”

Composites may increase reliability and validity but may also mask real relationships. For example, if adolescents' and parents' reports of monitoring are combined, a relationship between adolescent reports and their own outcomes can be missed.

A third issue is choosing measures. Developmental appropriateness involved choosing measures that were appropriate for the age of the child at the time and could also be used over time. To look at growth over time, measures need to be consistent over time. They also should be applicable over social status. For example, measures should be appropriate for both resident and nonresident fathers. With low-income families, there is a lot of fluidity in family structures. Parents move in and out of the household and move in and out of relationships. If the questions about resident and nonresident fathers are different, change cannot be studied over time.

Measures also need to be culturally appropriate. The Three City Study looked primarily at black and Hispanic families, but most measures in developmental psychology and related fields were developed for middle-class white families. The team spent more than a year piloting survey measures and structured observational measures to make sure that the measures were culturally appropriate and would work in the settings in which they were used.

The time between waves also proved to be important. One of the goals of the study was to make adjustments and improvements with each wave. With only about a year and a half between the first and second wave, there was not enough time to make adjustments in the interview protocols. “You really need a big chunk of time between waves to make adjustments,” said Coley.

The iterative process between the ethnography and the survey— allowing each method to inform the other—did not occur as much as hoped until the third wave. However, making adjustments in instruments can raise issues, since changing measures makes it difficult to assess change over time.

Several innovative methods incorporated into the study had great benefits. All of the survey instruments were preloaded into computers to support validity checks. Information from previous interviews and basic information about the people in a household were preloaded so that conflicts could be detected and inform cross-checks. In addition, respondents were trained for sensitive topics. They worked with headsets and laptop computers, heard the questions through the headsets, and answered the questions on the laptops. This approach has been shown to increase the validity of reporting on sensitive topics, such as sexual activity and substance use.

Areas that still need work include the issue of child elicitation (drawing out responses from children in research interviews or surveys) and bidirectionality. Children influence their families, not just the other way around, and investigations of families need to take these interactions into account. More attention needs to be given to the complexity and instability of family systems and how to access fathers, especially in low-income families. Better measures of positive child functioning and family processes are needed. Coley observed, “We are pretty good these days at measuring problems and risks, but our measures are much less valid for measuring positive productive behaviors. The measures we had in our survey of positive youth behaviors and positive parenting had such limited range that they are really not useful.” Finally, there is a need for opportunities for more mutual influence among the components of a study. Coley suggested that program directors need to “try to increase as much as we can the mutual influence and communication between these components in a timely fashion.”


Questions of meaning infuse research on families, said Paul Spicer, professor of anthropology at the University of Oklahoma. Instability in families does not just produce new family forms. It also produces cognitive instability as people struggle to make sense of their experiences.

The methods of anthropological research differ from those used by demographers, who work with national data sets. Yet some of the issues raised by demographers who study family structure echo those studied by anthropologists seeking to examine psychiatric distress in American Indian communities.

Stress and trauma are endemic in American Indian families. In recent epidemiological work by Spicer and his colleagues, rates of poverty in tribes were about 50 percent in the southwestern United States and about 60 percent among the Northern Plains tribes (Beals et al., 2005). This compares with a poverty rate of about 10 percent found in the National Comorbidity Survey (NCS), a nationally representative household survey of the prevalence of mental disorders in the United States. Social and health services for these tribes—and particularly mental health services— are severely underdeveloped or nonexistent. Health literacy levels are also likely to be quite low. Spicer said, “Our attempts to develop messaging campaigns and to think about home visiting and educational interventions suggest that there are fundamental difficulties in translating some of the most basic constructs we take for granted into terms that families can understand.”

Rates of alcohol dependence can be high, although these rates are not uniform and do not necessarily conform to the stereotypes common in the broader society. Spicer's epidemiological research suggests that they are about 50 percent higher for men and twice as high for women in the samples he studies than the rates found in the NCS.

Rates of posttraumatic stress disorder are also about 50 to 60 percent higher than in the NCS. But the measured rates of depression are lower than would be expected—about half the U.S. rate. In part, this was due to the instrument used to measure depression, which elicited evidence of both depressive episodes and other symptoms of depression separately rather than in the context of a discrete episode of major depression (Beals et al., 2005). “The way we ask these questions matters intensely for the findings we have,” observed Spicer.

Patterns that exist in stories and discourse may reveal as much—or more—than they do in survey data, Spicer said. People cannot always make sense of their experiences. “Indeed, one of the hallmarks of the experience of trauma and dislocation and loss is that it can be so disorienting that you can't find coherence in your experiences.” But the investigation of how people construct meaning in the midst of chaos is critically important. Open-ended, narrative approaches provide a way of understanding these processes in ways that responses to survey questions do not.

Finally, even open-ended narrative approaches cannot describe the reality of people's existence. Much that is of interest may not be explicitly understood. Explicit observational research is therefore important to help understand what is not fully comprehended.

With support from the Administration for Children and Families and the National Institute of Child Health and Human Development, Spicer has worked on embedding ethnography in psychological research. He and his colleagues have developed new tools to understand neighborhoods in reservations, to explore areas for which good measurement may be lacking, and to move beyond self-reports.

Part of this work involved coming to a new understanding of “neighborhoods.” Native communities are very different, both within and among reservations. In looking at these differences among communities, the researchers uncovered new factors of interest, including the risks posed by chemical contamination, primarily agricultural contamination, and problems with animals, most notably with dogs. Where dogs are treated poorly, children also seem to have great difficulties.

Community engagement has become a core requirement of doing research with American Indians. Research in tribal contexts has always required explicit tribal approval. The process of gaining this approval can lead researchers in different directions than they anticipated. “We have come to see community consultation as the centerpiece of doing the kind of research that can inform work on health disparities,” Spicer said. For example, in research on stress and young children's development, discussion of the social dynamics in reservation communities led to the addition of chemical contamination and problems with animals as factors to be considered.

A particular challenge in this research is the critical significance of historical trauma. People in tribal communities talk about the impact of history on the way they interact with children or about the predicaments they see in the current and future generations. Parents relate their desire to do differently for their children than what was done for them and about their concerns that they were producing the same circumstances for their children that they had explicitly hoped to avoid. Furthermore, much of what is hypothesized to be significant about historical trauma cannot be spoken, at least initially, so it is not amenable to survey approaches. “Existing measurement approaches are inadequate to get at this,” Spicer said.

Spicer's research has looked at cognitive development, school readiness, and differences in language development. Two visitors, an ethnographer and a clinician, were sent to 40 of 120 homes in one sample, one as an observer and one as the lead interviewer. Following the visit, the observer dictated reflections of parent-child interactions and the environment, “so we have very detailed records of our impressions of what was going on, both in terms of the interpersonal environment as well as the physical environment.”

Open-ended interviews make it possible to elicit accounts in ways impossible to do with survey tools. Especially when ethnographers were paired with clinicians in interviews, “we were struck by how much more powerfully emotive the discourse was in this open-ended context than it was in the survey, where people are saying yes-no or rating things one to seven.”

One notable observation Spicer and his colleagues have made involves an observed lag in language development early in life in the Northern Plains tribe with which they have conducted their research. Clinicians were struck by a lack of engagement and interaction between parents and children. Yet the ethnographers were struck as well by the potentially inappropriate cultural lens that clinicians were using in evaluating the lack of verbal interaction, since children's development can be supported nonverbally. These paired observations suggest disengagement in the context of stress and parents' experiences with poverty, substance abuse, mental health issues, and trauma but also emphasize the importance of developing messages that are consistent with cultural norms and expectations of infant care.

Spicer and his colleagues regularly capture their visits on videotape and audiotape, even in the homes of families with very complex needs. “It is quite possible if you take the time to build the rapport and if you staff your project locally as much as possible,” he said. Surveys and coding schemes may change, but the interaction archived in a video is permanent and can be coded in many different ways.

Observations in real-life contexts provide important information for understanding the impact of trauma and loss in the context of persistent cultural values. This information is essential in constructing public health messages to engage parents more fully in their children's development. Explicit measurement certainly is needed, but it is not often available. And even if a measure is available, it may not always be easily interpreted or understood. “A large national survey is going to be relatively silent on a lot of the processes that the ethnographer might want to hear about,” Spicer commented.

The tribes are interested in Spicer's research only insofar as it offers the prospect of improving the lives of their members. At the front end of their research, the researchers make a commitment to translate their findings into interventions. That means understanding parents' experiences in ways that are respectful of where they have come from, what they want, and how they can be sustained. Approaches to research based on discourse analysis allow scientists to understand how meaning can emerge from trauma and loss. Approaches to meaning also are central in thinking about what messages may resonate in particular communities. “There is a lot of room for sensitivity to meaning in the kinds of work we do in public health campaigns.”

A final issue Spicer raised is how questions of spirituality or religion enter into daily life in native communities. The parents in his research study uniformly feel estranged from their household traditions. Many turn to spiritual practices, and research has underscored the vital role that spiritual involvement can play. “When you look at what distinguishes people who are able to quit drinking and construct sober lives for themselves and the lives of people who are not, involvement in spiritual traditions of all sorts—be they Christian or tribal—appears to make a crucial difference. It is probably the one thing that does make a crucial difference. All people with alcohol dependence have tried to quit drinking—that is one of the hallmark symptoms. But what distinguishes those who are able to stay quit from those who go back to drinking appears, in both the quantitative and qualitative work we have done, to be largely driven by involvement in spiritual traditions.”


Economic, psychological, and cognitive studies of reading have all demonstrated that early skills are extremely important for later achievement. In particular, the skills with which children begin kindergarten or first grade are highly predictive of their rates of growth over time and their acquisition of more advanced and sophisticated skills.

This finding has special relevance for the study of achievement gaps among socioeconomic, racial, and ethnic groups, said Heather Bachman, assistant professor in education at the University of Pittsburgh. Studies of academic trajectories from kindergarten through fifth grade indicate a persistence of achievement gaps (Bachman and Mohan, 2007) or a widening of gaps (Votruba-Drzal et al., 2009; Bornstein and Bruner, 1989) over time. Even in the midst of education reforms, such as the No Child Left Behind Act of 2001, these gaps are relatively stable or in some cases may be widening.

Bachman and her colleagues study the processes behind these disparities and ways to promote competence and resilience for children. A variety of theoretical frameworks guides this work. For example, Bornstein and Bruner divide the differences in parenting for school readiness along two primary dimensions: socialization practices and more didactic practices (Bornstein and Bruner, 1989). Other theoretical frameworks for understanding parental teaching of early literacy skills emphasize the resources and investment of time and money (Becker, 1991; Foster, 2002) or the psychological distress associated with less stimulating and responsive parenting (Conger et al., 2002; McLoyd, 1990).

An important task is to identify emergent literacy skills. According to the National Early Literacy Panel (National Institute for Literacy, 2008), precursor skills include alphabet knowledge, phonological awareness, writing letters or one's own name, oral and receptive language, phonological memory, and the use of colors. Conventional reading skills are generally measured using standardized reading tests. However, it is not always clear why disparities in early reading occur, said Bachman. Do children recognize the letters of the alphabet? Do they know the phonological correspondences between letters and sounds?

Quantitative measures of parental teaching practices rely on a variety of assessments, such as the Home Literacy Environment (HLE) scales or the HOME-Cognitive Stimulation subscale (Bradley and Caldwell, 1984). These measures tend to have two major components. They measure resources, such as the number of books, magazines, newspapers, computers, and educational videos in a house. Or they monitor parents' behaviors, such as teaching letters, reading to a child, taking a child on educational outings, or limiting TV watching. One important item is how much parents read to themselves to model literate behaviors and limit TV watching, which is positively associated with reading acquisition.

These quantitative measures have several limitations. One is that it is sometimes hard to track change in these measures over time. Parents who provide books and read to younger children are also quite likely to do that after children go to school. To draw causal inferences, it would be informative to see if these processes change over time and measure the differences in functioning among children. Instead, these measures tend to sample one point in time or average longitudinal data. “They are very good at discriminating between child differences and achievement and less useful in predicting change over time,” said Bachman. Also, these measures tend to be global composites that are used to predict many kinds of academic skills. But parenting practices tend to be more specific to selected domains.

As an example of multidisciplinary research, Bachman cited a study of parental involvement in education (El Nokali et al., 2010). She and her colleagues used the NICHD study of early child care and youth development to examine within-child changes in parental communication with teachers and home-based involvement with child trajectories. They found few links between increases in parental involvement and improved academic outcomes. Instead, increased efforts at parent involvement led to declines in behavior problems and increases in prosocial skills among children. “It was a very unexpected finding that could lead to a number of exciting conclusions later on when applied to other kinds of parenting practices,” she said.

Reading researchers have found similar effects when working with parents on shared book reading activities. Whitehurst and colleagues trained parents to label the pictures in a book, to talk to children about some of the illustrations, and otherwise provide more engaged reading. Such reading has promoted receptive vocabularies among children, so they are learning more vocabulary words, but it has not transferred to early reading acquisition (Whitehurst et al., 1994). Again, the effects are domain specific, and the use of global measures of parenting could be masking more domain-specific causal mechanisms.

In general, said Bachman, the take-home message from the past several decades of research is that researchers commonly face restriction-of-range issues in observations and reports of HLE practices in low-income families, as well as for some minority groups. The parents of children from lower income backgrounds tend to score much lower on measures of home literacy promotion. But it is not clear if these lower scores are driven by fewer resources, by parenting behaviors, or by both. Additional literacy socialization practices may be operating among low-income families that are not adequately captured by traditional quantitative scales.

Bachman also highlighted some of the qualitative and mixed-methods research that has identified socialization practices associated with high achievement among children from low-income backgrounds. For example, the Baltimore Early Childhood Project, which started in the early 1990s, followed 80 children and families either from their prekindergarten year to third grade or from first grade to third grade and periodically collected parent diaries, ecological inventories, interview data, and standardized child assessments. Two major themes seemed to distinguish lower- and middle-income families' approaches to teaching literacy. When reading was viewed as entertainment, the themes that emerged were that reading is fun and enjoyable and that parents and children should choose books and topics in which they're interested. Middle-income parents more frequently endorse these kinds of practices, and there are few racial differences in income groups.

The other approach is that reading is meant to acquire skills. Children were encouraged to acquire the letter names and the letter-sound correspondences and practice these skills. There was less emphasis on reading for enjoyment or entertainment.

When parents in either the middle- or low-income groups endorsed reading as entertainment, children tended to benefit over time, with higher reading scores on different standardized assessments. The orientations for entertainment versus skill predicted differences in third grade reading over and above the effect of help with homework and other more common HLE perspectives. This work also raised interesting questions about how children not only acquire basic skills but also become engaged learners inside and outside school.

In another study, Reginald Clark followed black, low-income families and identified high-achieving and low-achieving high school students. He conducted interviews to ask about parenting practices and used participant observation in homes to look at routines and practices. Differences emerged not only in the involvement of parents and their support of achievement but also in their attributions of why their children were succeeding or failing in school (Clark, 1983). The parents of high-achieving students felt personal responsibility for their child's achievement when academic difficulties were encountered. They were proactive in talking to teachers or trying to find other kinds of assistance, and they created some routines and rituals at home that supported children's learning and achievement (Clark, 1983).

The parents of low-achieving students had a sense that their children were struggling but did not know how to improve the situation. They tended to blame the children for academic difficulties—for example, by saying that the students were not working hard enough. These parents also were less proactive in resolving issues with teachers.

Finally, recent research studies have examined the effects of “coparents” in low-income children's lives. These include coresiding grandparents of young mothers, custodial grandparents, social or cohabiting fathers, or older siblings who might be taking on some teaching responsibilities. Particularly in some low-income or immigrant families, the teaching role traditionally ascribed to parents could actually be delegated to other members of the family.

For example, one study looked at Indo-Chinese refugees with very low incomes and very low English fluency among parents or children when they moved to the United States (Caplan et al., 1992). Many of the children had missed several months or even years of formal schooling while in relocation camps. Yet many of these children adapted relatively quickly to school and performed well in school. In a subsample of 200 families for which in-depth interviews were conducted, the researchers found that although parents lacked education and facility with English, they took on more housework and other responsibilities to free up older siblings who could help the younger children with their homework after dinner. There were clear routines and rituals following the evening meal. The older siblings taught the younger siblings not only the content but also the skills, habits, and attitudes to become literate and engaged learners. Using conventional theoretical perspectives, it would be hard to measure this assistance in terms of time and money. The parents were not increasing their time in teaching; they were increasing their time in other household responsibilities to free up other family members' time to teach.

Multidisciplinary research has started to uncover the rituals and routines that support learning even in disadvantaged environments. These practices support not only the acquisition of skills but also the attributional and motivational characteristics of young learners. These qualitative and mixed methods could inform future survey research as well.


During the discussion session, Rebekah Levine Coley was asked what the Three City Project would do if a substantial addition of money was available. She responded that the funds should be used to do more analysis rather than expand data collection. “With huge studies like the Three City Study, we have a vast amount of data that hasn't been analyzed. Particularly if you look at the mixed-methods piece, the ethnography and survey and other components, we have done far too little real mixed-methods analysis with these data.”

Coley also pointed out that the biggest lost opportunity was the lack of enough time to do multidisciplinary communication and collaboration. “Partly it was the scale of the Three City study. It was so large, and each component was so large. Even though we had three years before we went into the field, for a study of this size and complexity there wasn't enough time and resources to have adequate communication and meetings.” The ethnographic team and the survey team started from very different places with very different assumptions and very different norms. There were also some inequalities. There was one senior ethnographer and five senior people who were more quantitative. Even for a 10-year study with $20 million in funding, there was not adequate time and resources to do the optimal level of collaborative planning. Very few people, prior to the most recent cohort of scholars, have the necessary multidisciplinary training for mixed-methods research and for coherently merging different theoretical models and perspectives.

Some of the groups also had somewhat different concerns. For example, the ethnographers had greater concerns about confidentiality and about fulfilling their obligations to respondents. Many of their sample participants were recruited through personal contacts and snowballing, where people involved in the study recommend others for recruitment into the study. There was also a concern about identifying people. For example, in the work on early childhood education, there was a concern about breaking down children's child care arrangements by Head Start centers versus other centers, because in many of the cities only a few Head Start centers participated in the ethnography. The ethnographers were concerned that if children were identified as being in Head Start versus other centers, the Head Start centers that participated in ethnography would feel singled out or that their confidentiality had been broken.

In response to the same question about how she would use additional funds for evaluation of the New Hope Program, Rashmita Mistry said that she would investigate how children spend their time outside the home and school. “It would have been nice to have some money to spend some time in those other settings that children were spending time in.” She was also interested in children's conceptions of their economic status. “One of the things that I don't think we did as well is to get a lot of this information from the child's perspective. … I would love to be able to go in there and in the qualitative piece do some very in-depth interviews with kids, … being able to see how this unfolds for kids parallel to the information that we have from parents.”

When asked what additional research she would do with additional funds, Bachman said that she was particularly intrigued by variations in socialization procedures across ethnic and racial groups. Some groups have gained more education, and attempts to disaggregate class and race could be fruitful, especially given the number of immigrant families in the United States. She was also interested in adding a qualitative dimension to her research, given that longitudinal surveys have generated lots of quantitative data. “Issues in parenting for early achievement have undergone so much research over the last 20 or 30 years,” she said. “Now is the time to get back into more qualitative work.”

Copyright © 2011, National Academy of Sciences.
Bookshelf ID: NBK56257
PubReader format: click here to try


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (1.4M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...