1. Top of page
  2. Abstract

Two studies of working families are combined to demonstrate a strategy for producing reliable estimates from the combination of self-reported (large N) and observational (small N) data. Both studies examine where and how dual-career families spend time at home. The 500 Family Study is sociological and uses self-reported time diary data from a national sample; the CELF study is anthropological and uses observational scan sampling data from a regional sample of 32 families. The data are combined as if they constitute one sample, and an analytic solution for establishing the reliability of the resulting composite estimates of time use is provided. Merging the data sets provides validation for each study, neither of which is without potential methodological weaknesses. The advantages of combining data from the independent data collection methods are discussed, and selected substantive findings on families' activities are highlighted, illustrating similarities and differences between findings in the independent and combined data sets. Results show that working families spend significant time in a small spectrum of home spaces, particularly kitchens and living rooms, with leisure activities prevailing, but mothers, fathers, and children differ in where and how they spend their time. Overall, a template for merging data from different disciplines and methods is provided.

Social scientists have been increasingly interested in combining observations from large-scale data sets with smaller data sets that address similar questions but do not share identical methodological approaches. Despite this interest, the methodological literature has provided few models for combining data sets derived from different methods and varying population sizes. This study addresses a gap in the literature by joining two social science data sets, both of which examine the lives of working families. The first study is sociological and uses self-reported time diary data from a national sample; the second study is anthropological and uses observational scan sampling data from a regional sample. Data from both studies examine where working families spend time at home and what activities they are engaged in. In this paper, the data are combined, and an analytic solution for establishing the reliability of the resulting composite estimates of time use is provided.

Joining independent studies of different population sizes can complement the results derived from each study individually. Studies with small populations are often based on purposive samples that provide in-depth descriptions of the lives of the selected respondents. One limitation of these types of studies is that their results have limited applicability to other populations. Studies with large populations are often based on representative samples; however, they tend not to contain the fine-grained information that is characteristic of smaller population studies. Combining small data sets with larger ones can reveal whether patterns from smaller data sets are borne out with larger samples.

In this paper, we compare self-reported data from a national population study with observations recorded in the field from a separate smaller population study. Self-reported data generally result in some degree of bias as participants hesitate to report socially undesirable activities. On the other hand, self-reported data have an advantage in that such reports are likely to capture how a participant perceives or experiences certain activities. What looks like a chore to a researcher could be a form of relaxation or leisure to an individual. Observational studies provide actual recordings of individuals' activities, thereby minimizing biased reporting, but they also raise the prospect of biased behavior (as some individuals may engage in activities in the presence of an observer that they would not if an observer were not present).

We first describe what we know about working families' time use and how these two studies can inform issues about the time families spend together. Next, we present the strengths and limitations of the two studies and what the benefits are for considering the two data sets as a single sample. Researchers conducting small population studies cannot simply add their data to large-scale studies that are nationally representative. A systematic process needs to be followed to address issues that may arise when separate data sets that use different methods and sample sizes are joined. We describe the procedures we use to address not only the reliability of the composite measures in the combined data sets but also the reliability of the measures within each data set. By reliability, we mean the consistency of measurements or a set of measurements (Zimmerman, Zumbo, and Lalonde 1993). Although various scholarly studies have examined how working families manage their time at home, few attempts have been made to join multiple studies in such a manner that the results provide a richer picture than each can independently.


  1. Top of page
  2. Abstract

Where and how families spend their time together is not as straightforward a question as it may seem. Today's working families manage their lives quite differently than families in past decades. Fathers are taking more responsibility for housework and child care, and many dual-earner couples who feel the pressure to keep up with their jobs are spending considerable amounts of time working at home in work spaces around the house (Bianchi, Robinson, and Milkie 2006; Ochs et al. 2006). Often, parents are multitasking, working at home while also engaged in child care, housework, or other family activities (Schneider and Waite 2005). Results from the 500 Family Study indicate that parents and their children engage in hundreds of types of activities at home, and most of the time family members are doing more than one activity at the same time in different places in the home (Schneider and Waite 2005). Examining how family members, particularly parents, locate their activities in space and time may provide insights into the strategies they use for managing household obligations and achieving family togetherness.

Although it is rarely considered in analyses of working families, the interior architectural design of family homes undoubtedly plays an important role in the social dimensions of everyday family life (Handlin 1979; Rapoport 1969). The proper places to prepare and eat meals, receive visitors, congregate, and sleep are just a few of the many activities insinuated by the interior arrangement of walls, lighting, plumbing, and other architectural features (Graesch 2004; Low and Lawrence-Zuniga 2003). That said, few middle-class homebuyers have the luxury of considering how interior arrangements of house spaces will facilitate or constrain their particular family lifestyles (contra Riemer 1947). Many houses are sold, purchased, and reused over long periods, resulting in a stock of housing designed by architects decades ago and reflecting historically particularistic norms of family interaction. House costs, proximity to desirable schools, and attributes of the surrounding neighborhood are often primary concerns of homebuyers, leaving considerable potential for mismatch between the spatial-organizational needs of modern-day families and the interior layouts of houses.

Family activities, on the other hand, have received more scholarly attention. Since the 1950s, researchers have studied the types of activities that families choose to engage in when together (Riemer 1947; Thorpe 1956). Many early time-use studies depended on retrospective data provided by just one adult, often the homemaker (typically the wife).1 Eating dinner together was the activity shared most by families during the 1950s (Snow 1950; Thorpe 1956). By the 1980s, “social activities” (i.e., recreation, play, and visits with family or friends) accounted for 60 percent of the time families spent together (Robinson and Godbey 1997). Today, when families are together, it is largely reported to be at home (Schneider and Waite 2005).

More recently, researchers have used various methods to explore the time use and activity patterns of working families in spaces at home. Common tools used to measure time use include self-reports of work activities, retrospective daily time diaries, beeper studies, participant observation studies, and survey questions regarding general time-use patterns. The best method(s) for capturing how families spend their time have been the subject of some debate (e.g., see Jacobs and Gerson [2004] and Robinson and Godbey [1997]). Questions regarding the reliability of these methods have been raised by several researchers, who suggest that multitasking, household planning and management, and estimated time use for household labor may be overestimated or underestimated, primarily because of respondent bias (Jeong 2005; Lee 2005; Mulligan, Schneider, and Wolfe 2005). For example, Lee (2005) finds that mothers and fathers provide inconsistent estimates of the time they spend on household tasks and the time that their spouse spends on housework. One way to discern how much bias is being introduced in estimates of time use is to compare observations of what working families are doing at home with self-reports. This is the issue we pursue by comparing estimates from individual and combined data sets.


  1. Top of page
  2. Abstract

This study uses data obtained from two major research initiatives: one at the University of California, Los Angeles, Center on Everyday Lives of Families (CELF) and the other at the University of Chicago Center on Parents, Children, and Work.2 Researchers at both centers are examining how working families spend their time at home and use home spaces.3 The CELF Center is primarily anthropological and incorporates high-resolution videography data, interviews, and observational scan sampling data; the Chicago Center is primarily sociological and uses self-reported time diaries, surveys, and in-depth interviews. For this study, the CELF data were obtained through scan sampling, a data collection technique whereby researchers record activities of all family members in their homes at close intervals; the Chicago data were obtained from the Experience Sampling Method (ESM), a unique time diary method. The CELF scan sampling data are based on several days of observations of 32 families whereas the Chicago data rely on ESM responses from 500 families (the 500 Family Study).4

2.1. CELF Study Sample and Methods

The CELF study is based on 32 working families recruited into the project by newspaper advertisements, information bulletins distributed in third- and fourth-grade Los Angeles County public school classrooms, and word of mouth. From this list researchers phoned potential families for an initial telephone screening interview to determine eligibility. Eligible middle-class families were selected, and the center provided a monetary stipend when the fieldwork with each family was complete. All 32 families own and pay a mortgage on a home in the Los Angeles area. Each family is composed of two parents who both work more than 30 hours per week outside the home and two or three children, at least one of whom is 7 to 10 years old. The families are racially and ethnically diverse, and the sample includes same-sex as well as heterosexual parents.5

The CELF study used an integrated set of data collection methods with each family, including audiotaped interviews, digital mapping and photography of home spaces, videotaped self-narrated home tours, repeated salivary sampling of cortisol, extensive videotaping of family activities, and standardized questionnaires. For one workweek and weekend, family activities and interactions were documented by three ethnographers, two of whom operated semiprofessional video cameras and one of whom systematically observed (tracked) and recorded all family members' locations and activities at 10-minute intervals with a handheld computer (Arnold and Graesch 2002; Ochs et al. 2006). It is these systematically coded observations made at 10-minute intervals that are analyzed in this paper.

This method of documenting family home life is a variant of scan sampling and an effective technique for documenting family members' movement and activities in the home environment while simultaneously structuring observation periods, the frequency of observation, and behavioral coding (Ochs et al. 2006).6 Similar goals have driven the development and application of various forms of standardized observation in ethnographic settings, including focal-person follows, spot observations, and point sampling (e.g., Draper 1975; Erasmus 1955; Hawkes et al. 1987; Johnson 1975; Konner 1976; Munroe and Munroe 1971; Nerlove et al. 1974; O'Connell, Hawkes, and Blurton Jones 1991).

Scan sampling is a method developed by behavioral psychologists for obtaining ethological data on multiple subjects by observing individuals in turn and recording the behavioral state at the instant of observation (Altmann 1974; Dunbar 1976). Scan sampling procedures differ from other methods of standardized observation in that behavior is repeatedly coded at predetermined points in time (Hinde 1985). To this end, data that derive from scan sampling methods are well suited to research in which comparative analyses of activities, uses of objects, and uses of space draw upon observations made both within and among groups (Ochs et al. 2006).

The application of CELF scan sampling procedures allows for the sequential observation and coding of family member location, focal activities, and objects. For the purpose of minimizing certain sampling biases, such as overrepresentation of group activities (see discussions in Borgerhoff, Mulder, and Caro 1985; Cromley 1999; and Hawkes et al. 1987), on each observational round, CELF ethnographers scanned all family member locations and activities as quickly as possible before taking a moment to code individual behavior in a handheld database. This rapid scanning technique was facilitated by family tendencies to use only a few spaces in their homes and served to provide a systematic series of nearly simultaneous observations of individual behavior in the home environment (Ochs et al. 2006).

The resulting data set has the advantage of closely spaced and sequential observations of families' behavior, permitting the fine-scale analysis of activities and spatial proximity over a week in the home life of working families. Because scan sampling data capture family activities and uses of space every 10 minutes, they can record more fine-grained patterns of time allocations than those that emerge from time diary data, which do not prompt respondents as frequently. Even so, individuals often change location and engage in new activities in the short spans between observations (as captured with CELF video ethnography). Thus, unlike time-allocation studies that employ more inclusive sampling methods (such as focal-person follows), an absolute amount of time devoted to particular activities cannot be calculated. Instead, a reliable time-allocation estimate can be provided, but this number may slightly overestimate total time devoted to particular activities, and this number derives from data that reflect the frequency of observations in which person X is engaged in activity Y.

A potential source of bias in data that derive from ethnographic research stems from the effects of ethnographers' presence on participant behavior. Although CELF families were observed in naturalistic settings (their homes), family participants may have occasionally altered their interactions and activities so as to better approximate perceived behavioral norms. That is, individuals may adjust activities in the presence of an observer or avoid certain interactions or activities due to perceived social undesirability.

2.2. The 500 Family Study and Methods

The 500 Family Study obtained in-depth information on dual-earner middle-class families residing in eight U.S. cities. Families were recruited through phone solicitations as well as mail and newspaper advertisements. Additional families were referred by parents already participating in the study. The focal child in more than 300 families was a teenager, and in more than 150 families, the focal child was a kindergartener.7 In 28 families, data were collected from both teenagers and kindergarteners. Among the majority of parents in the sample (60 to 70 percent), at least one parent is college-educated, and the average household income is higher than the national median.8

The 500 Family Study used several methods to examine the home and work experiences of parents and children in the study, including surveys, in-depth interviews, and the Experience Sampling Method (ESM), a form of time diary that provides data on where families spend time, what they do, and how they feel about it. The 500 Family Study also included a cortisol study with a subsample of the population.

The ESM, developed by Csikszentmihalyi and colleagues (Csikszentmihalyi 1997; Csikszentmihalyi and Larson 1984), is a weeklong data collection process that asks participants to wear wristwatches that emit randomly preprogrammed beeps eight times each day. These beeps occur no less than 30 minutes apart and no more than 2 hours apart. When respondents are beeped, they are asked to record in a booklet where they are, what they are doing, who they are with, and how they are feeling, producing scaled responses regarding mood and psychological state at the moment a beep occurs. Emotions are experienced naturally and recorded in the moment, overcoming the problem of time-delay in recall. This time frame obviously produces much more unrecorded time in terms of family activities than the CELF scan sampling data collection method, which provides information about subjects every 10 minutes.

The ESM data have been analyzed with various statistical techniques, and results have been shown to be generalizable to diverse populations (Jeong 2005; Mulligan et al. 2005).9 However, the ESM has been criticized as being too burdensome in that the time and cognitive demands made on the participants are greater than demands of more standard survey techniques. If so, individual participants may underreport what they are doing so as to avoid interruption (Mulligan et al. 2005). Additionally, ESM data on family member location, companionship, and activity are subject to the bias of self report (including failure to report socially undesirable activity); individual perceptions of primary activity; varying definitions of rooms/spaces; the extent to which activities are localized or distributed across space; and variability in reporting the proximity (copresence) of another family member. While self-report bias may be problematic, self-reported data also have the unique advantage of allowing researchers to study how respondents experience and identify their time and activities in the moment they are occurring.


  1. Top of page
  2. Abstract

3.1. Combining the Data Files

Combining the two data sets was possible because of certain parallel study design elements, including the collection of detailed records of family members' locations and activities in the home environment. The decision to combine data sets occurred during discussions between the Chicago and UCLA research teams, when, upon examination of the respective data sets, it became clear that the different data collection techniques provided similar estimates regarding the spatial locations and activity structures when working families are at home. Initial descriptive statistics on a small set of variables revealed that, in both studies, the majority of time (over 60 percent) families spent together (defined as both parents and at least one child in the same space) was in the kitchen. Preliminary analyses also revealed several problems that would have to be overcome in order to establish reliability of combining two separate samples.

The data sets were combined by the Chicago researchers by adding CELF data to the Chicago data set as if it was constituted by one sample of respondents. A cleaned text file containing the scan sampling data from the CELF study was sent to researchers at the University of Chicago for importing into SPSS. Once the file was imported, revised coding schemes for location and activity were devised by both sets of researchers in order to facilitate the data merge. The two SPSS files were combined to form one comprehensive file sorted on family identification that included participants' family ID, person ID (father, mother, focal child, etc.), date, time, location, primary activity, secondary activity, and data set source (CELF or 500 Family Study). This study focused the analysis on weekday afternoons and evenings, as family members returned from work and school and transitioned into home life. The combined data set includes 12,060 ESM data points (beeps; about 25 per family) and 5004 CELF scan sampling observations (about 155 per family), which reflect all family member responses and scan sampling observations that occurred after family members returned home (typically after 4 pm) on weekday afternoons and evenings.10

For comparative purposes, family member location and activity codes in both abridged data sets required standardization. There were 30 location codes in the combined data set in the home and the immediate area surrounding the home, including garages and yards. Interior home spaces in the CELF data set were defined by ethnographers as areas bounded by fixed architectural features (e.g., walls and doors), large furnishings used to divide floor space (e.g., bookshelves and couches), and major fixtures (e.g., sinks, countertops, fixed appliances). Exterior spaces were demarcated by fences and differences in surface composition (e.g., grass, concrete, wood). These spaces were assigned numerical labels and common functional labels such as “kitchen,”“living room,”“bedroom,”“front yard,” and “patio” (Arnold and Graesch 2002). ESM data on family member location reflect spaces that were self-classified by respondents. Functional labels assigned by these respondents are very similar to labels assigned by CELF ethnographers, and preliminary analyses of CELF family-narrated video home tours suggest that how ethnographers defined these spaces is also similar to how CELF family members define their home spaces.

Over 800 activity codes generated by self-report and scan sampling observations are represented in the combined data set. In the interest of conducting analyses of space and activities with statistically significant results, these activity codes were collapsed into 13 categories. The following are most relevant to this discussion: (1) leisure, (2) household management, (3) chores, (4) communication, (5) child care, (6) schoolwork at home, (7) work at home, (8) eating, (9) personal time, and (10) personal care. Whereas a few of these categories are obvious in terms of the activities they include (e.g., child care or schoolwork at home), we clarify a few others here. Examples of “leisure” include, among other things, reading, watching TV, playing games, and web surfing. Examples of “chores” include cooking, cleaning, repairs, and laundry. “Communication” includes talking in person or on the phone, e-mailing, listening, and arguing. “Personal time” includes thinking, meditating, procrastinating, daydreaming, and praying. “Personal care” includes grooming, dressing, napping, and taking medications. Reliability tests were conducted on the collapse of the more than 800 activity codes to ensure that the process was consistent across coders, resulting in a Cronbach's alpha of .92.

3.2. Weighting the Samples

Descriptive analyses of the patterns of family members' locations and activities in the initial combined data set indicated that the results primarily reflected the Chicago data. The combined data set was biased because ESM respondents were more likely to be represented than CELF participants, given that there were nearly three times as many ESM data points as scan sampling data points (12,060 ESM data points from 465 families versus 5004 data points from 32 families for scan sampling).To ensure that the large number of ESM data points did not introduce a bias into the composite estimates—but at the same time not discarding any available data—a proportional weight was constructed to correct for the overrepresentation of ESM data.

We constructed this proportional weight by determining what the ratio of the two independent data samples should be in the combined data set in order to ensure that it would appropriately represent the two independent samples without bias. Based on the distribution of data points, we applied a proportional weight of .7070 to the ESM data points, which reduced the ratio of data points to 1:1.85. This ratio was considered a satisfactory combination of data points after a series of means tests on all ESM data revealed that the combined results were no longer correlated with only the ESM results.The weighted combined data set represents 5004 data points from the 32 families in the UCLA scan sampling data set, and 8526 data points from the 465 families in the ESM. All analyses of the combined data sets incorporate the proportional weight, as noted in the tables. In the following analyses, when statistics from individual data sets are reported, they reflect the unweighted individual data sets.

3.3. Establishing Reliability

To ensure reliability of results, a series of tests was conducted on the combined data set to make certain that the integrity of the separate data sets was not compromised during the process and that the conjoined results were consistent with the patterns in the independent data sets. First, we computed reliability within the respective data using Cronbach's alpha coefficients. These alpha coefficients reflect the internal consistency of the measures of time spent in each activity and location. The computations are based on the repeated observations of each variable. In the case of the scan sampling data, there are repeated observations of locations and activities for each individual within each family over the course of several days of data collection, and in the case of the ESM, there are repeated self-reports of locations and activities from each individual within each family over the course of the weeklong ESM collection. These were used to estimate internal consistency of the three measures (location, primary activities, and secondary activities). The three alpha coefficients (for time spent in location, time spent in primary activities, and time spent in secondary activities) were averaged within each individual data set, resulting in one alpha statistic for each data set. (See Feldt and Charter [2006] for more information on internal consistency reliability coefficients.)

The alpha statistics from each individual data set were then entered into an equation aimed at measuring their individual reliability statistics against that of the combined data set. This equation takes the separate alpha statistics for the UCLA and Chicago data sets and sums the product of each based on their proportion of the total data and divides this sum by the overall alpha statistic from the combined data set. The formula used is as follows:

  • image(1)

where N indicates the number of components and r indicates the Pearson's correlation coefficient. The overall reliability coefficient based on this computation was .78.11


  1. Top of page
  2. Abstract

Descriptive statistics and cross-tabulations from both the independent and combined data sets reveal where family members spend their time and what they are doing when in the home, both when together and apart. While we have ensured that the combined data set is statistically reliable, the comparison of time- and space-use estimates between the independent and combined data sets also permits further evaluation of the methodological utility of our approach. Family members' locations and activities on weekday afternoons and evenings are presented as percentages of total ESM responses, percentages of total scan sampling observations, and/or percentages of total combined weighted data points for weekday afternoons/evenings. These data sets include all ESM responses and all scan sampling observations during which individual family members were recorded at home (usually after 4 pm), but these data sets are not restricted to those observations in which all family members were simultaneously at home.

4.1. Location of Family Members in Home Spaces

To ensure accuracy of results concerning where participants spend time in the home, split-half coefficients were calculated for location variables. Responses for location from the combined data set were split based on data source (CELF or 500 Family Study) and compared with each other. Using scale reliability analyses, a split-half coefficient of .81 was calculated, indicating that we can be confident in our results concerning where working families locate themselves in the home.

Examination of home spaces shows that bedrooms, kitchens, and living rooms are the three most frequently occupied spaces among the families who participated in the CELF and 500 Family studies (Table 1). Both individual and combined data sets indicate that over 60 percent of ESM responses and scan sampling observations place individual family members—alone or together—in one of these three spaces. In the combined data set, living rooms are the most frequently inhabited space in family homes, representing 22.6 percent of total data points. It is important to note, however, that because we collapse multiple individual “living” room spaces (i.e., den, family room, formal living room) into a single category for this analysis, these data in some cases reflect an aggregate use of two or more distinct spaces in the home.12

Table 1.  Location of Participant—Whether Alone or with Others—When Signaled or Observed
SpaceCombineda (n= 13,530)Scan Samplingb (n= 5,004)ESMb (n= 12,060)
  1. Note: Statistics indicate proportions of time spent in locations on weekdays (usually after 4 pm). The n statistic represents the raw number of beeps or scan sampling points that occurred in the specified location for all family members in the study. The percentage statistics represent the proportion of the total that occurred in the specified location.

  2. aData in the combined column reflects proportionally weighted data used in the combined data set rather than raw Ns from the independent data sets.

  3. bData in the Scan Sampling and ESM columns reflect the individual unweighted data sets.

  4. cAll bedrooms combined.

  5. dLiving room spaces include studies, dens, family rooms, TV rooms, and formal living rooms.

  6. eUnidentified/other home space.

Living roomd3,05722.61,20324.02,62321.7
Dining room991 7.350410.16895.7
U/O home spacee1,261 9.300.01,78314.8
Home office644 4.82114.26135.1
Bathroom567 4.22555.14413.7
Yard or porch408 3.01873.73132.6
Breakfast nook168 1.21683.480.1
Other679 5.03106.25134.3

Thus, frequencies for living room and bedroom spaces may be slightly inflated in the samples due to the method of grouping observations and responses for functionally similar rooms. Therefore, we argue that kitchens emerge as the most intensively used spaces in U.S. middle-class family homes (20.9 percent in the combined data set). Homes typically feature only one kitchen area, and these spaces tend to be used by all members of the family. Observations show that kitchens routinely serve as settings for daily mealtime, school-related, organizational, and interactive family activities. In the independent ESM data set, kitchens and living rooms are the most frequently occupied spaces (21.7 percent each), and in the scan sampling data set kitchens rank third (19.6 percent).

An analysis of where individual family members spend time in the home on weekday afternoons and evenings reveals notable differences among mothers, fathers, and children.13 Combined data presented in Table 2 reveal that these working mothers spend the highest proportion of their time in the kitchen (28.7 percent) while fathers spend less time in the kitchen (21.7 percent). Fathers spend more time than mothers in the home office (6.7 percent compared to 4.2 percent).

Table 2.  Proportions of Total Data Points Spent in Home Spaces by Family Member (Combined Data)a
SpaceFather (n= 3,200)Mother (n= 5,117)Focal Child (n= 2,921)Second Child (n= 1,792)
  1. Note: Statistics indicate proportions of time spent in locations on weekdays (usually after 4 pm). The n statistic represents the raw number of beeps or scan sampling points that occurred in the specified location for each family member in the study. The percentage statistics represent the proportion of the total that occurred in the specified location for each family member.

  2. aData in this table reflect proportionally weighted data used in the combined data set rather than raw Ns from the independent data sets.

  3. bUnidentified/other home space.

Living room73422.91,12922.161020.947626.6  
U/O home spaceb48315.14819.41916.5814.5  
Home office2136.72174.21174.0673.7  
Dining room2146.73787.41936.61789.9  
Yard or porch1053.31442.8722.5744.1  
Breakfast nook230.7440.9602.1311.7  

Differences between parents' and children's use of home space also emerge. For example, children are spending most of their time in their bedrooms away from their families (33.9 percent for the focal child; 27.2 percent for the second child in the family, combined data set). This is consistent with past research indicating that adolescents spend much of their time alone (Csikszentmihalyi and Schneider 2000).

4.2. Family Primary and Secondary Activities

To ensure accuracy of the results for the activities participants engage in when at home, split-half coefficients were calculated for both the primary and secondary activity variables. The responses for the activity variables in the combined data set were split based on data source and compared with each other. Using scale reliability analyses, a split-half coefficient of .82 was calculated for the primary activity variable and .6714 for secondary activity variable. Both statistics indicate that we can be confident in our results concerning how working families spend their time at home.

Complete primary activity data were available for 92.3 percent of CELF family members and 100 percent of ESM respondents. Table 3 presents primary (major) and secondary activities, as self-reported in the 500 Family Study, observed by ethnographers in the CELF study, and estimated with the combined data set. Leisure activities are the activities families engage in most frequently during weekday afternoons and evenings at home, constituting 30.4 percent of primary activities and 26.7 percent of secondary activities in the combined data set. Although conventional wisdom has it that dual-career families are very busy and overscheduled, both at work and at home, these data suggest that leisure (typically TV-watching) represents a higher proportion of family activity than household chores (14.5 percent, combined data set). This prevalence of leisure activities is also apparent in the individual data sets, both of which also reflect proportionally greater allocations of family time to chores and communication than to other activities, such as household management and work at home.

Table 3.  Proportions of Total Data Points Spent in Primary and Secondary Activities
Primary ActivityCombineda (n= 12,988)Scan Samplingb (n= 4,464)ESMb (n= 12,057)
  1. Note: Statistics indicate proportions of time spent in primary or secondary activities on weekdays (usually after 4 pm). The n statistic represents the raw number of beeps or scan sampling points that occurred for the specified activity for all family members in the study. The percentage statistics represent the proportion of the total that occurred for the specified activity for each family member.

  2. aData in the combined column reflect proportionally weighted data used in the combined dataset rather than raw Ns from the independent datasets.

  3. bData in the Scan Sampling and ESM columns reflect the full, unweighted, set of data.

Personal care1,0458.02405.41,1389.4
Schoolwork at home1,1608.962814.17536.2
Work at home3812.9611.44523.7
Household management2772.1651.53002.5
Personal time2712.1681.52872.4
Secondary ActivityCombineda (n= 6,581)Scan Samplingb (n= 903)ESMb (n= 8,031)
Personal care4466.860.76237.8
Personal time4777.2131.46568.2
Schoolwork at home2063.1839.21742.2
Household management1091.710.11531.9
Work at home1121.7293.21171.5

Similarities in the activity budgets of families that participated in the CELF and 500 Family studies suggest that ESM and scan sampling methods provide reasonably similar estimates of time use. The effects of data biases commonly attributed to either method (e.g., overreporting by ESM respondents or the behavioral effects of CELF ethnographers' presence on family participants) cannot be entirely ruled out, but similarities in families' primary activities as estimated with the 500 Family Study and CELF data sets suggest that they are not of great consequence. Leisure activities, for example, constituted 30.1 percent of activities reported by ESM respondents and 30.9 percent of all scan sampling observations of primary activities. The comparison serves as a check for each method and indicates that researchers should have confidence that these methods provide some degree of certainty regarding estimates of time use.

The few differences in estimates of time use may highlight the different characteristics of the study samples as well as the sensitivity of either method to recording various family activities. For example, schoolwork activities were more frequently observed among CELF project families (14.1 percent) than reported by participants of the 500 Family Study (6.2 percent). This may be attributable to differences in the ages of focal children in the projects: target children in the CELF study were typically 7 to 10 years of age, while those in the 500 Family Study tended to be adolescents or kindergartners. Third- through fifth-grade children are more likely to solicit homework assistance from parents than much younger or older children, thus resulting in more instances of homework activities recorded for both parents and children. Parents' involvement in homework activities among CELF families was, in fact, notably higher than that reported by ESM respondents.

This higher level of parental involvement in homework activities among CELF families could also have been motivated by the presence of ethnographers in the home and a desire to behave in accordance with culturally constructed notions of ideal family life. However, behaving in ways that were not consistent with everyday activities and interactions in the home would have required a considerable expense of energy by parents, and notable breaks in home routines would have generated pointed questions from other family members, especially younger children. Given that questions or comments regarding parents' involvement in homework were not observed, we suspect the presence of CELF researchers had little effect on this behavior.

Another notable difference in family activity patterns is the extent to which families were engaged in activities classified as “personal care”—for example, dressing, bathing, grooming, and taking medications. Such activities account for 9.4 percent of family time in the ESM data set but represent only 5.4 percent of family time in the CELF scan sampling data set (see again Table 3). This difference across data sets is likely attributable to key differences in how data were collected in the studies, rather than particular characteristics of the study samples. Specifically, CELF ethnographers were unable to record activities that transpired behind closed doors, and many of the personal care activities likely occur in bathrooms or bedrooms.15 In contrast, ESM respondents reported their activities, when beeped, regardless of whether or not they were performing activities in private.

“Secondary activities” were recorded for 63.5 percent of ESM responses and 18.9 percent of scan sampling observations. This disparity highlights methodological differences inherent in gathering self-reported and observational (essentially, emic and etic) data. For example, ESM respondents who were engaged in one activity (e.g., cooking) but simultaneously were thinking about another (e.g., scheduling child transportation from school to home) may have recorded “cooking” and “thinking about schedule” as their primary and secondary activities. In contrast, CELF ethnographers could only document visibly apparent activities, such as verbal interactions, physical movement, and actions.

Secondary activity data are presented in the lower panel of Table 3. Combined ESM and scan sampling data indicate that communication (29.3 percent) and leisure (26.7 percent) are the most frequently performed secondary activities. However, the patterns in secondary activities reveal bigger differences between the independent data sets than the patterns in the primary activities. First, ESM respondents report engaging in leisure as secondary activities more often than observed for CELF family participants (27.7 percent compared to 20.5 percent). One explanation for this difference could be that scan sampling depends on researchers' observations and interpretations of activities. For example, an ESM respondent could have the television on while doing work at home and thus might report both activities. In contrast, an ethnographer may not record “watching television” as a secondary activity if the family member appears completely disengaged from the program. Indeed, the fact that there are over twice as many secondary activities documented by the ESM suggests that ethnographers' perceptions of activity engagement differ from those of family member participants.

Second, there is a notable difference in the frequency of schoolwork at home as a secondary activity for ESM and scan sampling respondents. CELF family participants were observed pursuing schoolwork as a secondary activity 9.2 percent of the time compared to only 2.2 percent of the time for ESM respondents. Again, this disparity in estimates of families' secondary activities likely emerges from key differences in the two study samples—namely, the age of the children. While the children in the CELF study needed more homework supervision because of their age, it is also the case that it may be easier for parents to prepare dinner while simultaneously helping a grade school child with spelling homework than it is for parents to multitask while helping an adolescent with advanced math. In other words, not only is it more likely for parents to assist 7 to 10 year olds with their homework compared with adolescents, but it is also more likely that when parents do assist adolescents, it demands more focus and is a primary rather than a secondary activity.

When the activities of individual family members are examined, age and gender differences in their primary and secondary activities underpin the broader trends in the combined data set. Fathers, for instance, were most frequently engaged in leisure (29.5 percent), chores (17.5 percent), and communication (11.7 percent) as primary activities (Table 4). Mothers, on the other hand, were far more likely to be engaged in chores (22.2 percent) than were fathers or children, and they spent notably less time in leisure activities (22.7 percent) than did their children (39.5 percent, averaged between focal and second children) or male spouses. Schoolwork (about 19 percent), personal care (about 9 percent), and communication (about 12 percent) are other primary activities aside from leisure frequently pursued by focal children and second children (see Table 4).

Table 4.  Proportions of Total Observations Spent in Primary or Secondary Activities by Family Member (Combined Data)a
Primary ActivityFather (n= 3,117)Mother (n= 5,062)Focal Child (n= 2,832)2nd Child (n= 1,675)
  1. Note: Statistics indicate proportions of time spent in primary or secondary activities on weekdays (usually after 4 PM). The n statistic represents the raw number of beeps or scan sampling points that occurred for the specified activity for each family member in the study. The percentage statistics represent the proportion of the total that occurred for the specified activity for each family member.

  2. aResults from the combined data set reflect proportionally weighted data used in the combined data set rather than raw Ns from the independent data sets.

Personal care2297.43907.72679.41428.5
Work at home1424.62074.1190.790.5
Schoolwork at home1043.31422.853518.931118.6
Household management973.11372.7260.990.5
Personal time712.31142.3511.8301.8
Secondary ActivityFather (n= 1,663)Mother (n= 2,931)Focal Child (n= 1,340)2nd Child (n= 581)
Personal care985.92057.01078.0295.0
Work at home402.4652.230.240.7
Schoolwork at home181.1220.81118.3539.1
Household management301.8682.3100.710.2
Personal time935.62157.31037.76110.5

Secondary activity data for individual family members are presented in the bottom panel of Table 4. Combined communication (close to 30 percent for both mothers and fathers), leisure (20.8 to 25.8 percent), and chores (12.8 to 8.6 percent) are the top-ranking activities for parents who reported or were observed being engaged in a secondary activity, although the rank order of top secondary activities is different than that observed for primary activities. It is not unexpected that parents frequently are communicating with children and spouses while attending to household chores or relaxing. The higher number of secondary activity data points collected for mothers indicates that they multitask more frequently than fathers. Furthermore, they are more often busy with chores (12.8 percent) and childcare (9.7 percent) than are fathers (8.6 and 7.6 percent, respectively).

4.3. Families Spending Time Together

Turning now to group or collective use of spaces, the ESM and scan sampling data sets were parsed to include only those observations in which two parents and at least one child were together in the same home space (Table 5). Overall, the frequencies of this type of congregation were very low in the ESM and scan sampling data sets, ranging from 7.7 percent (ESM data) to 16.6 percent (scan sampling data) of total weekday observations. Varying estimates of family togetherness may highlight fundamental differences in the sensitivity of data collection techniques to family members' perceptions of togetherness. For example, spatial proximity criteria applied by CELF ethnographers when documenting family togetherness may not be equivalent to family members' notions of togetherness when responding to the ESM query “Who are you with?” The combined data indicate that when family members were together, they were most frequently in the kitchen (see again Table 5). Living rooms and dining rooms were the next most frequent loci for group activities.16

Table 5.  Proportions of Total Data Points in Which Both Parents and at Least One Child Occupy the Same Space or Participate in the Same Primary Activity
SpaceCombineda (n= 1,085)Scan Sampling (n= 802)ESM (n= 400)
  1. Note: Statistics indicate proportions of time spent in locations and primary activities on weekdays (usually after 4 pm). The n statistic represents the raw number of beeps or scan sampling points that occurred in the specified location or for the specified activity for all family members in the study. The percentage statistics represent the proportion of the total that occurred in the specified location or for the specified activity for all family members.

  2. aData in the combined column reflect proportionally weighted data used in the combined datasets rather than raw Ns from the independent data sets.

Living room26924.821626.98220.5
Dining room18417.014818.55614.0
Primary ActivityCombineda (n= 967)Scan Sampling (n= 679)ESM (n= 400)
Personal care747.700.08320.8
Schoolwork at home171.8182.700.0

The combined data suggest that these types of interactions are mostly situated in the context of leisure activities (41.4 percent) and eating meals or snacks (34.1 percent), although communication (11.3 percent) is another frequently reported and observed activity. In both the ESM and the CELF data sets, there are few instances of two parents simultaneously engaging in schoolwork activities with children. These data suggest that much of the parent involvement in children's schoolwork activities entails the contributions of one parent at a time.

It should be noted that, among CELF families, there are no instances of two parents engaging in personal care activities with children. Most such activities, if they occurred, likely took place behind closed bathroom doors, and thus could not be recorded by CELF ethnographers. In addition, researchers would have had to observe both parents simultaneously helping their children with personal care. In contrast, ESM respondents reported a high frequency of personal care activities when the family is together. This is likely attributable to parents reporting instances of helping their children prepare for bed when in fact they were not directly involved in this somewhat abstract activity. That is, parents are more likely to assist young children with personal care (e.g., bathing and dressing) than adolescents, although they might report doing so for both.


  1. Top of page
  2. Abstract

Overall, the results reveal that the ESM and scan sampling data sets suggest similar and complex stories, both independently and when combined, with respect to how working families are managing their time. The two data sets provide general agreement about where people spend the majority of their time at home and what activities they engage in. The data reveal that when at home together, family members spend more of their weekday afternoon and evening time in the kitchen than in any other single space. Considering that opportunities for time together for dual-income working families are largely limited to the span between when parents and children return home on weekday afternoons or evenings and when children are put to bed (sometimes no more than 3 to 4 hours later), it is perhaps not surprising that family members choose to frequent the kitchen, where opportunities for interaction may be maximized as family members use the kitchen as a central location for eating, relaxing, doing work, communicating, and doing chores.

To date, few time-use studies use the family as the unit of analysis; however, the CELF and 500 Family studies are exceptions. Data collection methods make it possible not only to measure when families are together, but also to document how the mother, father, and children spend their time independently. Considering times when families are together, it appears that the home still remains a refuge of sorts for the American family. These busy families are more likely to engage in leisure activities when they return home from work and school than any other single activity. If our methods only recorded the experiences of one family member's activities and uses of space and projected those as representative of the family, results would clearly be skewed. Clear patterns in gender and age (parent versus child) are evident in the combined and independent data sets. Children are more likely to engage in leisure activities at home than parents, but fathers are more likely to engage in leisure activities than mothers.

While results indicate something of a gender story, the differences today are not as stark as has been suggested for the 1980s and 1990s (i.e., Hochschild 1989). Results from both studies indicate that mothers engage in chores more frequently than fathers and their children. However, the proportion of time mothers spend on chores at home during these time periods is not substantially higher than that spent by fathers, consistent with recent research suggesting that fathers are spending more time on housework than in previous decades (see Lee 2005; Lee and Waite 2005). Still, working fathers enjoy considerably more leisure than time spent on chores, and mothers divide their time more evenly among chores, leisure, communication, and child care, indicating a persistence of a modest gender distinction in the household division of labor. What is perhaps the most interesting finding about chores is that the time spent on them in both the CELF and 500 Family data sets is so similar. It would appear that both methods are able to detect with some degree of certainty how the division of labor in households is allocated, dispelling doubts that fathers (and mothers) may significantly overreport the time they spend on this socially desirable activity.

The greatest differences between the two data sets appear in the time spent on homework and personal care. In the instances of homework, it would seem that the higher percentage of time spent on homework by adults and children in the CELF data set is most likely driven by the younger age of the focal children in this sample compared to the 500 Family Study. Although the difference in the average age of the children in the samples is only about five years, there are considerable differences in the ways that parents interact and spend time on homework with their high schoolers, middle schoolers, and kindergarteners. This is a function of developmental distinctions and the differing demands that schools put on children in these age groups. These differences underline the importance of considering potential sources of bias when joining or comparing studies with seemingly similar populations.

Differences in time spent on personal care appear to be mainly attributable to methodological differences stemming from data collection procedures. In the 500 Family Study, respondents can report how they spend their time in places where the ethnographers are unable to record, such as in private spaces (e.g., bathrooms) and behind closed doors. However, it is important to underscore that even though the estimates for time use in personal care in these two samples are different, personal care is not a major family activity compared to leisure, chores, and communication, where the data are more consistent. Nonetheless, limitations in the abilities of independent methods to collect certain types of data must be noted when interpreting results from combined data sets.

In summary, results show that the analysis and interpretation of combined ESM and scan sampling data requires a careful consideration of the effects of differences in participants and data collection procedures. Only by examining patterns in individual project data sets (as an additional confirmation of the success of the combination of the data) were we able to discover significant differences in frequencies of observed/reported schoolwork activities, personal care activities, and personal time activities. Sharing data in the social sciences has often been approached with caution because of the risk of breaching confidentiality rights of study subjects. However, when proper precautions are taken (for example, compliance with Human Subjects/IRB protocols), collaborative efforts can enrich the conclusions that can be derived from a single data set.

The CELF study is an intensive study of a small number of families, with a large number of observations per family. The scan sampling data set used here provides a rich, fluid, and sequential picture of family life since actions are recorded at narrow intervals for hours at a time. The results from the CELF study are consistent with the ESM, despite the latter's much lengthier spans of unrecorded time. Further family time use research should take advantage of the potential for sequential data to allow for an understanding of how families transition from interaction to interaction. This in-depth detail of the scan sampling method permits the construction and analysis of detailed activity budgets that, while subject to the risks of researcher interpretation, are consistent with the self-reported data of the 500 Family Study. Combining the CELF data set with the 500 Family Study allows UCLA researchers to confirm findings in the data with those from a larger population.17

Similarly, the combined analyses permit University of Chicago researchers to confirm patterns found in self-reported survey data, subject to many biases such as underreports or overreports of certain types of activities, with findings drawn from intensive ethnographic research. Thus, the process detailed in this paper both verifies the results from both studies and enriches the methodological literature. In doing so we provide a template for future projects with some parallel study design elements that seek to combine data that use different methodological approaches and vary with respect to sample sizes. This type of work seems especially promising in fields like medical sociology where researchers studying the HIV/AIDS epidemic could combine their small data sets (based on local data collection in specific regions of Africa) with larger national or international data sets. Similarly, sociologists focusing on globalization could combine ethnographic data with larger economic and development studies if similar measurement elements can be identified.

Forging a collaboration based on data from two different types of social science methods is not a straightforward task. Often, the combination of data from multiple studies may compromise the constituent studies. This collaboration is unique in that it merges new ethnoarchaeological applications of scan sampling methods with commonly used social science time-use methods. The combining of data entailed a rigorous process during which both data sets had to be carefully scrutinized and checked to ensure accurate results. It has been the careful process of negotiation between the two centers that has kept the integrity of each study, and the combined results, as the focus of the collaboration.

  • 1

    Early studies showed most family time together consisted of only the mother and children together, and it was rare that one child would spend time with parents without siblings present (Davey and Paolucci 1980).

  • 2

    The Alfred P. Sloan Foundation funds both studies. The methodological approaches and sample selection occurred within three years of each other. However, at the time each study was initiated, the investigators had not considered merging the data sets. The idea to merge the data occurred after both teams had completed their data collection independently of one another.

  • 3

    The studies focused on families in which the parents were both employed. Most mothers and fathers were employed full-time (worked for pay at least 37.5 hours per week in the Chicago sample and 30 hours per week in the UCLA sample), including several workers who were self-employed. There were also some couples where one partner worked part-time or had recently lost a job.

  • 4

    The actual sample sizes used in these analyses vary from the full sample; these variations are noted in the discussion of merging the data sets.

  • 5

    For more information on the design and findings of the CELF study, refer to Providing more specific description of the families would violate confidentiality rules because of the small sample size.

  • 6

    Other CELF publications and research papers have referred to these scan sampling methods as “tracking” procedures, an in-house term that derives from conversations on how to systematically “track” family members in their homes.

  • 7

    A focal child in each family was selected to participate in data collection procedures. Adolescents completed a survey, the ESM, and an in-depth interview. Kindergarteners completed standardized child assessments and participated in an informal interview.

  • 8

    For a complete description of the sampling procedures, see Hoogstra (2005).

  • 9

    Results from the ESM have been confirmed by other time-use surveys, such as those used by Bianchi and Robinson (1997), and standard survey responses. The ESM has not been previously validated with observational data.

  • 10

    Early mornings were omitted from the current analysis because the typical individual preparations for the day amounted to just a few minutes of time spent together. Families with parents who worked nontraditional hours were also excluded because they account for only 3 percent of the sample. Observations of third children in families in the CELF sample were excluded as there were very few observations and the ESM does not collect this information.

  • 11

    A Cronbach's alpha value above .70 is generally accepted as indicating reliability.

  • 12

    A similar method of collapsing location categories was conducted for bedroom spaces, which are often used exclusively by particular family members (i.e., parent's room, focal child's room, etc.).

  • 13

    In families with same-sex parents (family n= 6), both parents were classified as either fathers (men) or mothers (women).

  • 14

    Although this alpha statistic is just below .70, we continued with the analysis because of our confidence in the data set as a whole (alpha = .78; see page 20). The low alpha is likely due to the difference in sample size for secondary activities in the two studies.

  • 15

    CELF videographers and ethnographers shared an understanding with family members that the act of closing a door signaled a request for privacy from data recording procedures and devices.

  • 16

    In contrast, outdoor (yard and porch) spaces are relatively rarely used, particularly for leisure activities (Arnold and Lang 2007).

  • 17

    The 500 Family data set is more nationally representative in terms of both racial breakdown and income distribution.


  1. Top of page
  2. Abstract
  • Altmann, Jeanne. 1974. “Observational Study of Behavior: Sampling Methods. Behaviour 49:22767.
  • Arnold, Jeanne E., and Anthony P. Graesch. 2002. “Space, Time, and Activities in the Everyday Lives of Working Families: An Ethnoarchaeological Approach.” Working Paper No. 2. UCLA Center on Everyday Lives of Families, Los Angeles , CA .
  • Arnold, J. E., and U. A. Lang. 2007. “Changing American Home Life: Trends in Domestic Leisure and Storage among Middle-Class Families. Journal of Family and Economic Issues 28(1):2348.
  • Bianchi, S. M., and J. Robinson. 1997. “What Did You Do Today? Children's Use of Time, Family Composition, and the Acquisition of Social Capital. Journal of Marriage and the Family 59:33244.
  • Bianchi, Suzanne M., John P. Robinson, and Melissa A. Milkie. 2006. Changing Rhythms of American Family Life. New York : Russell Sage.
  • Borgerhoff Mulder, Monique, and T. M. Caro 1985. “The Use of Quantitative Observational Techniques in Anthropology. Current Anthropology 26:32335.
  • Cromley, Ellen K. 1999. “Mapping Spatial Data.” Pp. 51124 in Mapping Social Networks, Spatial Data, and Hidden Populations, edited by J. J.Schensul, M. D.LeCompte, R. T.TrotterII, E. K.Cromley, and M.Singer. Walnut Creek , CA : AltaMira Press.
  • Csikszentmihalyi, Mihaly. 1997. Finding Flow. New York : Basic Books.
  • Csikszentmihalyi, Mihaly, and Reed Larson. 1984. Being Adolescent: Conflict and Growth in the Teenage Years. New York : Basic Books.
  • Csikszentmihalyi, Mihaly, and Barbara Schneider. 2000. Becoming Adult: How Teenagers Prepare for the World of Work. New York : Basic Books.
  • Davey, Alice J., and Beatrice Paolucci. 1980. “Family Interaction: A Study of Shared Time and Activities. Family Relations 29:4349.
  • Draper, Patricia. 1975. “Cultural Pressure on Sex Difference. American Ethnologist 4:60016.
  • Dunbar, Ronald I. M. 1976. “Some Aspects of Research Design and their Implications in the Observational Study of Behaviour. Behaviour 58:7898.
  • Erasmus, Charles J. 1955. “Work Patterns in a Mayo Village. American Anthropologist 57:32233.
  • Feldt, Leonard S., and Richard A. Charter. 2006. “Average Internal Consistency Reliability Coefficients. Educational and Psychological Measurement 66:21527.
  • Graesch, Anthony P. 2004. “Notions of Family Embedded in the House. Anthropology News 45(5):20.
  • Handlin, David P. 1979. The American Home: Architecture and Society 1815–1915. Boston : Little, Brown.
  • Hawkes, Kristen, Hillard Kaplan, Kim Hill, and Magdalena Hurtado. 1987. “A Problem of Bias in Scan Sampling. Journal of Anthropological Research 43:23946.
  • Hinde, Robert A. 1985. “Comment to M. Borgerhoff Mulder and T.M. Caro's “The Use of Quantitative Observational Techniques in Anthropology.” Current Anthropology 26:331.
  • Hochschild, Arlie R. 1989. The Second Shift: Inside the Two-Job Marriage. New York : Avon Books.
  • Hoogstra, Lisa. 2005. “The Design of the 500 Family Study”. Pp. 1838 in Being Together, Working Apart: Dual-Career Families and the Work-Life Balance, edited by B.Schneider and L.Waite. New York : Cambridge University Press.
  • Jacobs, Jerry A., and Kathleen Gerson. 2004. The Time Divide: Work, Family, and Gender Inequality. Cambridge , MA : Harvard University Press.
  • Jeong, Jae-Gea. 2005. “Obtaining Accurate Measures of Time Use from the ESM.” Pp. 46182 in Being Together, Working Apart: Dual-Career Families and the Work-Life Balance, edited by B.Schneider and L.Waite. New York : Cambridge University Press.
  • Johnson, Allen. 1975. “Time Allocation in a Machiguenga Community. Ethnology 14:30110.
  • Konner, Melvin J. 1976. “Maternal Care, Infant Behavior, and Development Among the !Kung.” Pp. 21845 in Kalahari Hunter-Gatherers: Studies of the !Kung San and Their Neighbors, edited by R. B.Lee and I.DeVore. Cambridge , MA : Harvard University Press.
  • Lee, Yun-Suk. 2005. “Measuring the Gender Gap in Household Labor: Accurately Estimating Wives' and Husbands' Contributions.” Pp. 22947 in Being Together, Working Apart: Dual-Career Families and the Work-Life Balance, edited by B.Schneider and L.Waite. New York : Cambridge University Press.
  • Lee, Yun-Suk, and Linda J. Waite. 2005. “Husbands' and Wives' Time Spent on Housework: A Comparison of Measures. Journal of Marriage and the Family 67:32836.
  • Low, Setha M., and Denise Lawrence-Zuniga. 2003. “Locating Culture.” Pp. 147 in The Anthropology of Space and Place: Locating Culture, edited by S. M.Low and D.Lawrence-Zuniga. Malden , MA : Blackwell Publishing.
  • Mulligan, Casey B., Barbara Schneider, and Rustin Wolfe. 2005. “Nonresponse and Population Representation in Studies of Adolescent Time Use. Electronic Journal of International Time Use 2:3354.
  • Munroe, Ruth H., and Robert L. Munroe. 1971. “Household Density and Infant Care in an East African Society. Journal of Social Psychology 83:313.
  • Nerlove, Sara B., John M. Roberts, Robert E. Klein, Charles Yarbrough, and Jean-Pierre Habicht. 1974. “Natural Indicators of Cognitive Development: An Observational Study of Rural Guatemalan Children. Ethos 2:26595.
  • Ochs, Elinor, Anthony P. Graesch, Angela Mittmann, Thomas Bradbury, and Rena Repetti. 2006. “Video Ethnography and Ethnoarchaeological Tracking.” Pp. 387409 in The Work-Family Handbook: Multi-Disciplinary Perspectives and Approaches to Research, edited by M.Pitt-Catsouphes, E. E.Kossek, and S.Sweet. Mahwah , NJ : Lawrence Erlbaum.
  • O'Connell, James F., Kristen Hawkes, and Nicholas Blurton Jones. 1991. “Distribution of Refuse Producing Activities at Hadza Residence Base Camps: Implications from Analyses of Archaeological Site Structure.” Pp. 6176 in The Interpretation of Archaeological Spatial Patterning, edited by E. M.Kroll and T. D.Price. New York : Plenum Press.
  • Rapoport, Amos. 1969. House Form and Culture. Upper Saddle River , NJ : Prentice Hall.
  • Riemer, Svend. 1947. “Sociological Perspective in Home Planning. American Sociological Review 12:15559.
  • Robinson, John P., and Geoffrey Godbey. 1997. Time for Life: The Surprising Ways Americans Use Their Time. University Park , PA : Penn State University Press.
  • Schneider, Barbara, and Linda J. Waite. 2005. Being Together, Working Apart: Dual-Career Families and the Work-Life Balance. New York : Cambridge University Press.
  • Snow, Charles P. 1950. “A Study in the Development of a Technique for Determining the Amount of Time and Type of Activities which Family Members Share.” Master's thesis, Department of Sociology, University of Georgia , Athens , GA .
  • Thorpe, Adam. 1956. “Patterns of Family Interaction within the Home.” Ph.D. dissertation, Department of Sociology, Michigan State University , Lansing , MI .
  • Zimmerman, Donald W., Bruno D. Zumbo, and Coralle Lalonde. 1993. “Coefficient Alpha as an Estimate of Test Reliability Under Violation of Two Assumptions. Educational and Psychological Measurement 53:3349.