How reliable are fathers' reports of involvement with their children?: a methodological report.

Family researchers have long recognized the value of having multiple informants when studying families. However, the reliability of fathers' reports, especially on their level of involvement with their children, has sometimes been questioned. Using time diary data from a sample of White, well-educated, first-time parents (N = 52 couples), this study moves beyond a simple comparison of father and mother reports of father involvement by using a method that corrects for the times mothers are not able to observe fathers' level of involvement. Results indicate father and mother reports of father involvement are highly similar. This finding suggests that fathers are reliable reporters of their own involvement with their young children via time diaries. Previously reported differences between father and mother reports of father involvement may be due to methodological issues more than the gender of the reporter.

Keywords: father involvement, reliability, time diaries, mother and father reliability

**********

For many years research data on families came almost exclusively from mothers. One assumption of this "wives' family sociology" was that "wives' reports of family life were more accurate (contain less measurement error) than those of husbands" (Thompson & Williams, 1982, p. 1000). Though there is now an appreciation for the value of multiple informants when studying the family, doubts remain about the reliability of data obtained from fathers, as evidenced by comments in the literature that call into question the accuracy of data obtained from fathers (e.g., Hofferth, Pleck, Stueve, Bianchi, & Sayer, 2002; Roggman, Fitzgerald, Bradley, & Raikes, 2002; Seltzer, 1991). However, the assumption that data from fathers are less reliable than data from mothers has rarely been subjected to direct empirical testing. The purpose of this study is to examine empirically the reliability of fathers' reports of involvement.

An example of scholarly doubts about fathers' ability to accurately describe their own behavior is found in a statement made by Roggman et al. (2002): "Fathers do not always provide accurate or complete information about their own behavior, but that information can be usefully supplemented by others who observe them" (p. 23). This statement is made in a chapter discussing issues in studying fathers and as rationale for the importance of using multiple informants (including mothers and children) when researching families. Though a valid point, the absence of a qualifying statement to point out that there are no informants (including mothers) who can always be assumed to provide fully accurate or complete information calls into question the credibility of data gathered from fathers.

In a similar fashion, though independent sources of evidence for the reliability of mothers' reports of their own involvement is consistently absent, only the reliability of fathers' reports of involvement is questioned. For example, Hofferth et al. (2002), in their discussion of what fathers do in families, use data from the National Survey of Adolescent Males (NSAM) with a sample of young fathers (ages 21-27) with young children averaging 2.5 years old. The NSAM looks at a number of issues relating to the father-child relationship. To address the time spent interacting with their children, the survey asks fathers to estimate how much time they spend a day reading to, playing with, and caring for their children. In their discussion of the data, Hofferth et al. state, "Fathers appear to over-estimate the time they spend playing with and reading to children in the NSAM compared to time use studies" (p. 85). What is not said is that general estimates of involvement in domestic activities, by both women and men, are generally higher than reports from time use studies (see Marini & Shelton, 1993; Press & Townsley, 1998). "Overestimation" appears to be a function of the methodology of "estimation" rather than a problem specific to fathers' reports.

In the same vein, Seltzer (1991) used the National Survey of Families and Households (NSFH) to look at the relationships between fathers and children who live in separate households. She states that she used the mothers' reports of father involvement because the sample of mothers was "more representative Of families in which the children live apart from their father than the sample of nonresident fathers" (p. 83) and because of the low response rate from nonresident fathers. An additional reason, however, was that "nonresident fathers' reports about their involvement with children may be less accurate, overestimating their contributions to childrearing, compared to resident mothers' reports about how fathers behave" (p. 83). Seltzer supported this assertion with aggregate comparisons of mothers' and fathers' reports of child-support payments. She thus generalized from the accuracy of fathers' reports on child-support payments to the accuracy of their reports of their social involvement (frequency of visits, mail and phone contact, and extended visits) with their children.

These types of statements in the academic literature fuel the untested assumption, which amounts to a bias, about the unique untrustworthiness of fathers' reports of involvement with their children. This study aims to test the reliability of fathers' reports of involvement by comparing them with mothers' reports of father involvement. If the assumption that mothers are reliable reporters is correct, then congruence between their reports of fathers' involvement and reports by fathers themselves provides evidence for the reliability of fathers' reports. Ideally, of course, we would have a third-party report as well, in order to determine the reliability of both mothers and fathers, but that is beyond the capacity of the current study. At a minimum, this study provides an independent test of the reliability of fathers' reports from the person best positioned to observe father-child interactions--the mother.

Two core issues are relevant to understanding the reliability of father data: the reliability of the method of measurement and the source of data used to establish reliability. The following review of literature is intended to demonstrate the impact of both of these issues on evidence of the reliability of father data.

Regarding the measurement of father involvement in domestic activities, there are three common methods: direct question estimates, frequency scales, and time diaries, with each having varying degrees of reliability. Of the three, direct question estimates appear to be the most common method used. In this method parents are asked to estimate how much time they each spend doing certain household and childcare activities in a week. This method has been found to produce higher estimates of time spent in activities, however, than the time-diary method, as noted earlier (Juster & Stafford, 1985; Marini & Shelton, 1993; Press & Townsley, 1998; Robinson, 1977, 1996).

The second method of measuring involvement, the use of frequency scales, taps the frequency of contact in different areas. For example, the National Survey of Children (NSC) used the following scale as a measure of father involvement as reported by a child: "[How often] do you and (he) do things together that you enjoy: a. often, b. sometimes, c. hardly ever?" Ahrons (1983), in her study of predictors of father involvement post divorce, used a similar scale to assess father involvement and found a significant discrepancy between fathers' and mothers' reports of father involvement "with a striking tendency for fathers to overestimate or mothers to underestimate the involvement" (p. 62). In discussing this discrepancy, she urges caution about collecting data from only one spouse. It is important to recognize that the scale method taps estimates that are quite subjective and is therefore likely to be affected by the over-estimation/under-estimation dynamic.

A third method of measuring involvement in domestic activities is the time diary. Participants are contacted and asked to give an account of their activities throughout a particular day (a 24-hour period, usually midnight to midnight). Participants may be interviewed regarding how their time was spent on a particular day (usually the day prior to the interview). Alternatively, participants may fill out a time diary regarding what they are doing throughout the day or fill out the diary at the end of the day. Participants generally fill out the diary without knowing the agenda of the researcher. Cross-validation of this method through "beeper" or the "experience sampling method" (where participants are asked to write down what they are doing when signaled by a beeper at random times during the day--e.g., Larson, 1989, or Robinson, 1985) and by comparing "random hour" aggregates to time-diary data have shown the time-diary method to be a valid measure of time use and less subject to participant biases (Larson; Robinson).

The source of the data used as evidence for the reliability of fathers' reports of involvement can also influence how the reliability of father data is understood. A common practice for establishing the reliability of father data is to compare fathers' reports of their behavior with mothers' reports of the fathers' behavior. However, this practice is not entirely without weakness. Berk & Shih (1980) compared husbands' and wives' reports of contributions to household labor and found that "gender" affected reporting levels. That is to say, gender affected both the likelihood of a self-reported contribution to a task and the agreement between reporters about the level of contribution. In addition, the authors specifically identified the tendency for a parent to "underestimate" a partner's involvement in childcare. Ahrons (1983), in discussing the discrepancy between father and mother reports of father involvement, noted how fathers and mothers perceive father involvement differently and how this contributes to the discrepancy between parental reports. Without considering these dynamics, and as long as mothers are considered to be the more reliable reporters, then fathers will be seen as overreporting their involvement, and the reliability Of fathers' reports will be in question.

This study examines the reliability of fathers' reports of involvement with their children in a way that combines two methodologies previously not used together in the research literature: the time-diary method and both fathers' and mothers' reports of father involvement. In addition, a new methodological approach will address a potential limitation of using mothers to assess reliability of father scores: the problem arising from times when the mother is not available to observe the father's activities with the child.

METHOD

SAMPLE

Data for this study were taken from the Parenting Together Project (PTP), an intervention study designed for first-time parents in the transition to parenthood. As part of the study, time diaries were mailed to the participants two weeks before their sixth-month postpartum assessment. The fathers and mothers were asked to choose together a typical workday and nonwork day, usually a weekend day, and then to each fill out a time diary for those two days. Of the 165 couples who participated in the study and completed time diaries, 111 (93%) completed diaries for the same workday, and 104 (91%) completed diaries for the same nonwork day. Having both the father's and mother's diaries from the same days allowed for a comparison between the father's report of involvement with his child and the mother's report of the father's involvement with his child. Data for this study were taken from nonwork day time diaries because these had a far greater number of entries by both fathers and mothers.

Due to the resource intensity required to code both the father and mother reports of father involvement on the time diaries, we decided to code a 50% randomly chosen subset of the total sample, or 52 father diaries and 52 mother diaries. The procedure of analyzing a randomly selected subset of the larger sample is commonly used to assess inter-rater reliability or validate measures in a cost-effective manner (e.g., McBride & Mills, 1993, or Robinson, 1985).

In the subsample for this study, 90% of the couples were married and 10% cohabiting. The fathers' ages ranged from 20 to 43 years (mean = 31) and mothers' from 18 to 41 years (mean = 30). Ninety percent were White; 5% were Asian American; 2.5% were African American; and 2.5% were Hispanic, Native American, or other non-White. The sample was well educated: 50% had at least a college degree, 29% had a graduate/professional degree; 5% had only a high school degree. Income levels were also relatively high: 25% had a combined income greater than $100,000 per year, 70% of the sample had an income greater than $50,000 per year. Ninety-six percent (all but two) of the fathers and 77% (all but 12) of the mothers reported being employed. Employed fathers worked an average of 42.7 hours per week, and employed mothers worked an average of 27.9 hours per week. Analyses revealed no significant differences between the subsample and the larger sample on any of these demographic variables. There were also no significant differences on the two father-involvement variables of interest in this study. Finally, although data for this study were taken from both the intervention and control groups in the original Parenting Together Project, the two groups were collapsed here because there were no significant differences on the variables of interest in this study.

PROCEDURES

As noted earlier, the time diaries were mailed to the participants two weeks before their sixth-month postpartum assessment. Participants were instructed to each fill out a time diary for two days--a typical workday and nonwork day (midnight to midnight--a 24-hour period) describing the activities of the child, themselves, and their partner, in relation to the child. Once the participants agreed upon a particular workday and nonwork day, they were given the option of filling out the time diary throughout the day (ongoing) or at the end of the day (retrospective). In the instructions, participants were asked to have the diaries completed by the previously scheduled assessment date. For the assessment, a research assistant would meet with the couple in their home, have them complete a battery of paper-and-pencil self-report questionnaires, and videotape the parents interacting with their child. If one or both of the parents had not completed the time diaries, the research assistant would ask the parent or each of the parents to fill out a time diary for the preceding day.

MEASURING FATHER INVOLVEMENT

The "total time diary," an instrument adapted from the time-diary interview used by McBride (1990, 1991), was used to study the amount of time fathers spent interacting with, or being accessible to, their children. The diary consisted of three pages divided into 15-minute blocks of time for a 24-hour period. Each page contained three columns, labeled "Baby's activity," "Your Activity," and "Partner's Activity." Therefore, on each diary, the second column contains a description of the recording parent's own activities, and the third column contains his or her the description of the activities of the partner.

Using a modified version of McBride's (2001) coding system, fathers' and mothers' diaries were first coded for four variables: engaged interaction (those times the parent was actively involved with the child and the child was the focus of the activity); accessibility (when the parent was available for interaction yet not actively interacting with the child); unavailable (time when the parent was not available to interact with the child--separate from sleep time); and sleep time. If a 15-minute block of time was left blank, then it was inferred from the activities described in the preceding and following blocks of time how to code that block of time according to the four variable codes. This primarily turned out to be time when the parent was accessible but not interacting with the baby. Though unavailable time and sleep time were coded, they are not the primary focus of this study and therefore will not be discussed further.

Coding reliability for the FFP study was established by having a second person code a subset (20%) of the 134 available time diaries. A level of agreement was established between the coders by counting the number of 15-minute sections for which there was agreement and then dividing that number by 96 (the number of 15-minute periods in 24 hours). This yielded a percentage of agreement between raters for a 24-hour period; raters averaged 93% agreement. When the coding reliability for engaged interaction and accessibility were examined separately, there was found to be 91% and 95% agreement, respectively. The primary coder's scores were used for the statistical analysis, and the 104 diaries for this study were taken from the overall sample in which the couple had completed diaries for the same nonwork day.

When comparing fathers' and mothers' reports of father involvement, there were times when the mothers clearly did not have access to direct knowledge of the type of father involvement--for example, when she was away from the home or sleeping while the father was with the baby. Often a mother would write what the father was doing even though she was not available to observe what was happening. Though she may have had an idea of what he was doing (by guessing or by his report), it appeared mothers could only reliably report from first-hand knowledge on his "accessibility" if the father was the responsible adult caring for the child. It was assumed that when the mother was away from the child, she could not always reliably report on the specific type of father involvement with the child because she was not available to directly observe whether, for example, the father was playing face-to-face with the child (engaged interaction) versus reading the newspaper while the child played nearby (accessibility). Thus, any differences between mothers' and fathers' reports would be difficult to interpret.

To deal with this problem, the diaries were coded in two ways to yield different kinds of scores for the mother and the father. First, the diaries were coded for engaged interaction and accessibility just as the fathers and mothers reported, yielding "uncorrected" scores for these variables. Next, to "correct" for those times the mother could not accurately report on the father's activity, the diaries (both the mother and father reports), were recoded with the engaged interaction times being coded as accessibility for those times when the mother reported that she was unavailable to observe what the father was doing (she was away from the father-child dyad or asleep). This yielded "corrected" scores for engaged interaction and accessibility. Twenty-seven of the couples' diaries (52% of the sample) needed this correction. The findings from the two methods will be reported separately.

ANALYSIS

Larsen and Olson (1990) identified a strategy for determining the similarity of dyadic reports in families. They suggest first using correlations to demonstrate the linear relationship between the scores of the two family members, in this case the reports of fathers and mothers on father involvement. However, high correlations alone cannot be understood as an indication of similarity between a pair of scores. There could be a strong linear relationship between the variables even in the presence of major differences in the actual magnitude of the scores between the partners. To address this issue, Larsen and Olson suggest using paired t-tests to "answer whether individuals (e.g., husbands) as a whole are different from other individuals (e.g., wives) as a whole" (p. 28). High correlations and nonsignificant t-tests together would indicate similarity in scores.

Two phases of analysis were conducted, with uncorrected data and corrected data, using correlations and t-tests to compare fathers' and mothers' scores for engaged interaction and accessibility.

RESULTS

Using uncorrected scores, the correlations between fathers' and mothers' scores on engaged interaction and accessibility were r = .818 (p < .001) and r = .821 (p < .001), respectively. Next, the means for engaged interaction and accessibility were calculated (see Table 1). Fathers reported a mean of 3.09 hours of engaged interaction during the day sampled, while mothers reported fathers engaging in 2.93 hours of interaction, a nonsignificant difference of 10 minutes. Means for father accessibility were equivalent: 11.48 hours.

The second phase of analysis addressed the effect of correcting the time-diary reports for those times when the mother was not able to report from first-hand knowledge on the father's engaged interaction. As reported earlier, 27 couple diaries needed this correction. Correlations between the father's and mother's reports for this subsample, before the correction was done, were r = .842 for engaged interaction and r = .701 for accessibility. After these time diaries were corrected, the correlation increased only slightly: for engaged interaction (r = .847) and accessibility (r = .719), a nonsignificant difference of correlations based on Fischer's r to z transformation procedure. Similarly, there were no significant differences between the father-mother report correlations for the uncorrected subsample (n = 25) and corrected subsample (n = 27) groups either before or after the corrections were made.

Fathers' and mothers' corrected reports of engaged interaction were also very similar: means of 2.67 hours and 2.79 hours, respectively, for engaged interaction, and 11.86 hours and 11.48 hours for accessibility (see Table 1). Interestingly, after the correction, the mean for the mothers' reports of the fathers' level of engagement was slightly higher (eight minutes) than the mean of the fathers' report of his level of engagement. Mothers reported fathers to be 23 minutes less accessible than fathers did for themselves (all differences are nonsignificant).

DISCUSSION

In contrast to earlier studies that reported differences between fathers' and mothers' reports on involvement in domestic labor and childcare (Ahrons, 1983; Berk & Shih, 1980; Press & Townsley, 1998), and in contradiction to the bias in the field about the trustworthiness of father data, this study found close similarity between fathers' and mothers' reports of father involvement. If the assumption that mothers are reliable reporters is accepted, then these results provide empirical support for the reliability of fathers' self-reports of involvement. This finding held after scores were corrected for times when the mother was not available to report on the specific form of father involvement.

Perhaps differences found in prior research are more a function of the method than an inherent tendency of fathers to overreport or provide unreliable data. In addition, the previous finding that mothers underestimate what fathers do in the home may be a product, in part, of her not knowing what the father was doing. It appears this error can affect time-diary data, though maybe to a lesser degree than other measurement approaches and in a way that can be corrected.

The attempt to "correct" mothers' and fathers' reports for those times the mother was not able to personally observe the father interactions appeared to increase the similarity between the reports of engaged interaction. Yet the fact that mother-father reports were already so similar may call into question the need for this procedure in future time-diary studies of father involvement with infants.

Earlier studies had identified a gender effect on estimates of participation in domestic work (Press & Townsley, 1998) and on reports of contributions to house hold labor (Berk & Shih, 1980). The current study did not find significant differences between fathers' and mothers' reports of father involvement. This suggests that gender did not influence time-diary reports of fathers' involvement.

Results need to be understood in the context of how involvement is defined in this study compared to definitions used in other studies. While this study focused strictly on those activities where the father was directly interacting with his infant or immediately available, the level of similarity between father and mother reports might be different with a broadened definition of involvement. For example, involvement might include providing a secure environment or giving children opportunities to learn to care for themselves and others (see Dollahite, Hawkins, & Brotherson, 1997).

Though the results of this study support the reliability of fathers' reports of involvement with their children, there are several limitations. The fact that this sample was primarily White, well educated, and from a high socioeconomic status (SES) limits the ability to generalize the results of this study to other populations. Persons of other ethnic backgrounds or those having lower education or SES may perceive involvement differently and/or conceptualize time differently. Therefore, they may use the time diary differently, and the reliability of fathers' reports in populations with these characteristics would need to be established separately. In addition, the ability to generalize these findings to reports of involvement with older children is limited by the age of the children (six months) in this sample. The life-course stage of infancy might produce more congruence between parents' reports than other stages when children are more independent.

It is also important to note that the finding of high similarity between fathers' and mothers' time-diary reports is only evidence of the reliability of fathers' reports if mothers themselves can be assumed to be reliable reporters. The present study had no independent source of evaluation of the accuracy of the mothers' reports of father involvement. Though a trained outside observer could be a third independent source, there are a number of limitations to this alternative, including cost, logistical challenges, and possible effects of the observer on how the father interacts with his child.

Results of this study cannot be generalized to other methods of measuring father involvement. There is sufficient evidence that asking for perceptions about the level of contribution, scaling questions, and direct questions asking for estimates of involvement often yield differences between mother and father reports (see Berk & Shih, 1980; Coley & Morris, 2002; Hofferth et al., 2002; Warner, 1986). Though not directly addressed in the studies reviewed, conceivably, the characteristics of time-diary research might enhance the congruence between father and mother reports over other methods of measurement. Further research needs to be conducted to establish the reliability of fathers' reports for other methods of measurement.

In summary, this study supports the reliability of fathers' reports of involvement with their children, particularly when using time-diary reports. In contrast to other studies, this study has demonstrated that fathers' reports of involvement with their children appear to be very similar to mothers' reports. The reliability of fathers' reports remained even when efforts were made to remove the effect of mothers reporting what fathers are doing when they were not present to actually know. Time diaries seem to be less susceptible to the influences of gender and social desirability and therefore appear to be a good method for measuring father involvement.

Although in recent years family research has included the father's perspective on family life, there are still remnants of distrust regarding the reliability of data from fathers. These conclusions appear to be based on methodologically suspect grounds. It also appears that the method of measurement, not gender alone, affects reliability. Different methods of measuring father involvement give qualitatively different data. Though various methods are used, there rarely is discussion about the effect of the method, as opposed to the gender of the respondent, on the accuracy of the data. Greater effort needs to be taken in the literature to address how method, in addition to gender, contributes to the reliability of data from fathers.

Table 1
Father Involvement with His Child as Reported by Fathers and Mothers

                                Fathers               Mothers

                          Mean (hrs.)    SD     Mean (hrs.)    SD

Uncorrected Scores (a)
  Engaged interaction         3.09      1.48        2.93      1.56
  Accessibility              11.48      1.87       11.48       2.0

Corrected Scores (b)
  Engaged interaction         2.67      1.49        2.80      1.46
  Accessibility              11.86      1.52       11.48      1.47

                              Confidence
                            interval (95%)

                             Lower      Upper

Uncorrected Scores (a)
  Engaged interaction        -0.10      0.42
  Accessibility              -0.03      0.32

Corrected Scores (b)
  Engaged interaction        -0.45      0.19
  Accessibility              -0.06      0.82

(a) n = 52 couples

(b) n = 27 couples

REFERENCES

Ahrons, C.R. (1983). Predictors of paternal involvement post-divorce: Mothers' and fathers' perceptions. Journal of Divorce, 6, 55-69.

Berk, S.F., & Shih, A. (1980). Contributions to household labor: Comparing wives' and husbands' reports. In S.F. Berk (Ed.), Women and household labor (pp. 191-227). Beverly Hills: Sage.

Coley, R.L., & Morris, J.E. (2002). Comparing father and mother reports of father involvement among low-income minority families. Journal of Marriage and Family, 64, 982-997.

Dollahite, D.C., Hawkins, A.J, & Brotherson, S.E. (1997). Fatherwork: A conceptual ethic of fathering as generative work. In A.J. Hawkins & D.C. Dollahite (Eds.), Generative fathering: Beyond deficit perspectives (pp. 17-35). Thousand Oaks, CA: Sage.

Hofferth, S.L., Pleck, J., Stueve, J.L., Bianchi, S., & Sayer, L. (2002). The demography of fathers: What fathers do. In C.S. Tamis-Lemonda & N. Cabrera (Eds.), Handbook of father involvement: Multidisciplinary perspectives (pp. 63-90). Mahwah, NJ: Erlbaum.

Juster, F.T., & Stafford, F.P. (1985). Introduction and overview. In F.T. Juster & F.P. Stafford (Eds.), Time, goods, and well-being (pp. 1-18). Ann Arbor: Survey Research Center, Institute for Social Research, University of Michigan.

Larson, A., & Olson, D.H. (1990). Capturing the complexity of family systems: Integrating family theory, family scores, and family analysis. In T.W. Draper & A.C. Marcos (Eds.), Family variables: Conceptualization, measurement, and use (pp. 19-47). Newbury Park, CA: Sage.

Larson, R. (1989). Beeping children and adolescents: A method for studying time use and daily experience. Journal of Youth and Adolescence, 18, 511-530.

Marini, M.M., & Shelton, B.A. (1993). Measuring household work: Recent experience in the United States. Social Science Research, 22, 361-382.

McBride, B.A. (1990). The effects of a parent education/play group program on father involvement in child rearing. Family Relations, 39, 250-256.

McBride, B.A. (1991). Parent education and support programs for fathers: Outcome effects on paternal involvement. Early Child Development, 67, 73-85.

McBride, B.A. (2001). Time diary form and coding guidelines. Unpublished manuscript, University of Illinois at Urbana.

McBride, B.A., & Mills, G.A. (1993). A comparison of mother and father involvement with their preschool age children. Early Childhood Research Quarterly, 8, 457-477.

Press, J.E., & Townsley, E. (1998). Wives' and husbands' housework reporting: Gender, class, and social desirability. Gender and Society, 12, 199-218.

Robinson, J.P. (1977). How Americans use time: A social-psychological analysis of everyday behavior. New York: Praeger.

Robinson, J.P. (1985). The validity and reliability of diaries versus alternative time use measures. In F.T. Juster & F.P. Stafford (Eds.), Time, goods, and well-being (pp. 1-18). Ann Arbor: Survey Research Center, Institute for Social Research, University of Michigan.

Robinson, J.P. (1996). Time, housework and the rest of life. Journal of Family and Economic Issues, 17, 213-229.

Roggman, L.A., Fitzgerald, H.E., Bradley, R.H., & Raikes, H. (2002). Methodological, measurement, and design issues in studying fathers: An interdisciplinary perspective. In C.S. Tamis-Lemonda & N. Cabrera (Eds.), Handbook of father involvement: Multidisciplinary perspectives (pp. 1-30). Mahwah, NJ: Erlbaum.

Seltzer, J.A. (1991). Relationships between fathers and children who

live apart: The father's role after separation. Journal of Marriage and the Family, 53, 79-101.

Thompson, E., & Williams, R. (1982). Beyond wives' family sociology: A method for analyzing couple data. Journal of Marriage and the Family, 44,999-1008.

Warner, R.L. (1986). Alternative strategies for measuring household division of labor. Journal of Family Issues, 7, 179-195.

This study was funded in part by a grant from the Maternal and Child Health Bureau, Health Services Research Administration, U.S. Department and Health and Human Services (5R40 MC 00141-03).

Correspondence concerning this article should be addressed to Kurt A. Wical, Family Social Science, University of Minnesota, 290 McNeal Hall, St. Paul, MN 55108. Electronic mail: [e-mail].

KURT A. WICAL

WILLIAM J. DOHERTY

University of Minnesota