Methodological considerations in measuring paternal identity.
Key Words: fathers, parenting, identity, methodology, pie chart
Recent research on fatherhood has employed multiple methodologies to assess several different aspects of fatherhood. The most frequently assessed aspect is paternal involvement or behavior (e.g., Bonney, Kelley, & Levant, 1999; Bruce & Fox, 1999; NICHD Early Child-Care Research Network, 2000; Yeung, Sandberg, Davis-Kean, & Hofferth, 2001). Less frequently assessed, though receiving more attention recently, is paternal identity (e.g., Burke & Cast, 1997; Ihinger-Tallmann, Pasley, & Buehler, 1993; Maurer, Pleck, & Rane, 2001).
Research on paternal identity evolved out of prior, more general work in identity theory. Identity theory posits that identity is "internalized sets of role expectations" (Stryker, 1987, p. 90). As applied to paternal identity, a man's identity as a father comprises all the expectations for his behavior that he has internalized as being associated with being a father (e.g., being a breadwinner, being a caregiver, etc.). To the extent that a man internalizes those expectations for the role of father, one would expect his behavior to be consistent with those expectations. For example, a man who internalizes the expectations of being a caregiver to his child into his identity as a father should in fact provide care to his child (thus matching his expectations to his actual behavior). In this sense, paternal identity theory posits that fathers' identification with the role of father should predict their enactment of expected behaviors (i.e., their involvement). Indeed, prior research on paternal identity has found fathers' self-reports of paternal identity to be linked to paternal involvement (e.g., Bruce & Fox, 1999; Ihinger-Tallmann et al., 1993; Maurer et al., 2001). Much of this research has been based on assessments of paternal involvement at the role level, rather than investigating the connection between paternal identity and more specific areas of paternal behavior, such as caregiving.
Most research on paternal identity has also employed closed-ended, multiple-item scales to assess both paternal identity and paternal involvement. However, recent scholarship evaluating the last decade of fatherhood research has suggested that new and improved methods for measuring these constructs are needed (Day & Lamb, in press; Marsiglio, Amato, Day, & Lamb, 2000). This paper explores the advantages and disadvantages of scale and pie-chart methodologies for assessing identity at both the role and domain levels. It also compares the extent to which these measures of paternal identity predict paternal involvement.
Past studies have demonstrated that multi-item scales have adequate internal reliability and significantly predict paternal involvement (e.g., Maurer et al., 2001). In contrast to scales, an alternative new method of evaluation, the identity pie chart, is relatively easier to use, and often requires significantly less time to complete. Pie-chart measures of identity provide parents with a list of roles (e.g., parent, worker) and ask parents to divide a pie according to the relative importance of the listed roles to them (Cowan et al., 1985; Futris, Pasley, Kerpelman, & Goodman, 1998; Strauss & Goldberg, 1999). For example, if a father is asked to fill in a pie chart illustrating how important each of four social roles (e.g., worker, spouse, parent, other) is to him, then the larger he draws one section (indicating it is more important), the smaller he must draw the other three (indicating they are less important). The size of each slice of the pie is measured in degrees, and since any role or domain can only have one slice, however large, the pie chart is effectively a single-item measure. Prior work with pie charts has revealed a positive association between the size of fathers' parent role slice of the pie and their involvement in routine child-care activities (Futris et al., 1998; Strauss & Goldberg, 1999), consistent with the predictions of paternal identity theory.
The advantage of ease of use that pie charts offer, however, also presents a significant drawback to this method: pie charts are single-item measures, and as such, their reliability is much more difficult to evaluate. Since the use of pie-chart methodology in investigations of paternal identity is increasing and researchers are drawing substantive conclusions about paternal involvement from pie chart data, it is important to evaluate the efficacy and validity of this methodology before more fully endorsing it as either a supplement to, or replacement for, traditional scale measures of identity.
Another methodological difference in measures of paternal identity concerns the conceptual level at which identity is assessed. Much prior research has investigated paternal identity at the role level. Role-level identity concerns the major roles in a father's life, such as his role as a worker, parent, or husband (Ihinger-Tallmann et al., 1993; Minton & Pasley, 1996). However, Ihinger-Tallmann et al. have pointed out that within a role such as fatherhood, there are theoretically distinct conceptual areas, such as caregiving or being a companion to one's child. These areas can be considered equivalent to Stryker's (1987) "sets of role expectations," and for clarity and ease of reference we use the term domains to refer to these sets of expectations within the father role. Thus, assessing identity at the role level means assessing the importance of being a father (or identifying oneself as a father), while assessing identity at the domain level means assessing the importance of being a caregiver, teacher, or any of the other sets of role expectations associated with being a father.
Recent scale measures of paternal involvement behavior have distinguished multiple domains. Bruce & Fox (1997) developed a measure of paternal involvement comprised of four domains: custodial caretaking functions, socioemotional functions, teaching functions, and executive functions, although only a summary score of all four domains was used in their analyses. While prior research has established that both domain-level measures of identity (Maurer et al., 2001) and role-level measures of identity (Bruce & Fox, 1999) can predict role-level measures of behavior, these measures of identity have not been tested for their relationship with domain-level measures of behavior.
Based on these prior findings and the theoretical predictions of paternal identity theory, we present the following hypotheses and research questions:
* Hypothesis 1A: Role-level measures of identity will be significantly, positively correlated with role-level measures of behavior.
* Hypothesis 1B: Domain-level measures of identity will be significantly, positively correlated with role-level measures of behavior.
* Hypothesis 2A: Role-level measures of identity will be significantly, positively correlated with domain-level measures of behavior (i.e., according to paternal identity theory, fathers' identities as fathers should be related to paternal involvement).
* Hypothesis 2B: Domain-level measures of identity will be significantly, positively correlated with domain-level measures of behavior (i.e., according to paternal identity theory, fathers' sets of role expectations should be consistent with the enactment of the associated behaviors).
* Research Question 1: What is the relationship between pie-chart and scale measures of identity?
* Research Question 2: Are pie-chart measures of identity, at both the role and domain level, equivalent (in magnitude of correlation with behavior) to scale measures of identity?
Fathers in two-parent families with a child aged two to five years were recruited at three sites. Two university child-development laboratories, one in the Midwest, one in the Pacific Northwest, and a daycare center in the Pacific Northwest, were used. The combined analysis sample of 64 fathers from three sites represented 35% of all eligible fathers in two-parent families. All fathers who participated in this study were married, though no attempt was made to exclude cohabiting fathers. At one site, administrative data were available for the entire pool of prospective participants, so a limited comparison between respondents and non-respondents was conducted. Results indicated no significant differences between respondents and non-respondents (e.g., average age and education of respondent and non-respondent fathers were not significantly different).
Eighty-four percent of the sample was white (54 of 64 fathers). The average age of fathers was 38.3 years (SD = 3.6), with a mean education level of 18.0 years (SD = 2.2). Of the 64 fathers, 90.6% (58) were employed. The mean number of weekly paid and unpaid work hours was 39.4 (SD = 19.5). The average number of children per family was 2.0 (SD = 0.9), and the average age of the children was 4.3 years (SD = 0.9). Of the children, 44.4% (28) were boys and 55.6% (35) were girls, with one child's gender unreported.
Role level identity scale. The first instrument used to measure identity was the Bruce & Fox (1999) Father Role Salience Scale (RSS), a 10-item measure (five reversed items) with a reported internal reliability of alpha = .61. This role-level scale consists of 10 statements about the paternal role. Respondents must indicate which statements are not true, somewhat true, or very true of them. Examples of the statements are "Being a father has changed me a lot"; "I would rather work overtime than watch my kids for the evening" (reversed); "I like being known as a father." The scale points are 1,2, and 3, and a mean score is computed (range = 1-3). In this investigation, internal reliability was alpha = .41.
Domain level identity scale. The Maurer et al. (2001) Caregiving and Breadwinning Identity and Reflected-Appraisal Inventory was the second instrument used to assess parents' identities. This measure was developed for use with both mothers and fathers and is a domain-level measure. It consists of six sections, three for caregiving and three for breadwinning. In each parenting domain (caregiving and breadwinning), the three sections consist of identity, partner's reflected appraisal, and perceived reflected-appraisal items. For the purposes of this investigation, only the data from the 14-item caregiving identity subsection (CGI) were analyzed. All items use a five-point Likert format (six items are reversed), with 11 items employing an agree/disagree scale, one item asking "how important" caregiving is to them (1 = not at all important, 5 = extremely important), and two items asking participants to select one of five answers to indicate the relative importance of an item with regard to another role or to one's spouse. Examples of the items are "I would like to be remembered for the quality of care I gave my child"; "When I do caregiving, the main reason is to help out my spouse" (reversed); "When my child is sick on a work day, I have a responsibility to stay home and care for him/her if I can get time off." The CGI has a reported internal reliability of alpha = .70 for fathers, and that alpha was replicated in this analysis.
Role level identity pie chart. The third instrument used to measure identity was the Futris et al. (1998) Role Salience Pie Chart. This role-level pie chart measures the relative salience of four role identities (marital, paternal, work, and other) by asking participants to divide the pie according to "how important" each of the roles is to them (e.g., if all four are equally important, they would divide the pie into four equal slices). This pie-chart measure of paternal identity (PIE-P) is assessed by measuring the number of degrees of the pie allocated to the parenting slice.
Domain level identity pie chart. The fourth instrument used to measure identity was the Parenting Domain Pie Chart, developed for this study. This domain-level pie chart measures the relative importance of five parenting domains (the four domains measured by Bruce & Fox, 1997, plus breadwinning) in the same manner as the PIE-P, and it is scored in the same way. Because the caregiving slice is the only domain of interest to this investigation, this portion of the pie chart will be referred to as the PIE-CG.
Role level behavior instrument. The measure selected was the Bruce and Fox (1997) Paternal Involvement Inventory (PII). This role-level measure has a reported reliability of alpha = .95 and assesses paternal behaviors in four domains: custodial caretaking functions, socioemotional functions, teaching functions, and executive functions. The 21 items ask how frequently the parent engages in specific parenting activities (e.g., assisting the child in bathing) with a four-point response scale ranging from never or hardly ever to almost daily. Since the mean summary score (range = 1-4) is aggregated across all four domains, and since the PII was designed as a measure of father involvement (instead of domain-level caregiving involvement specifically), the PII is interpreted as a role-level measure. In this investigation, internal reliability was alpha = .89.
However, because the caregiving domain is specifically assessed in this measure, it is possible that any observed correlation between domain-level measures of identity (i.e., caregiving) and this role-level measure of behavior may be inflated by the presence of items from the caregiving domain in this measure (i.e., any observed relationship might actually be between identity and the caregiving subscale, rather than identity and the full role-level measure). To prevent such an inflation of the observed relationships, the six-caregiving domain items were dropped from the measure. The remaining 15 items obtained an internal reliability of alpha = .87, suggesting that dropping the caregiving items did not significantly compromise the measure. As a result, this 15-item PII was used as a role-level measure of paternal behavior.
Domain level behavior instrument. The measure of caregiving was the six-item domain-level custodial caretaking subscale, which was dropped from the PII. As one would expect, the same four-point response scale is used in this subscale, and the scale score is computed in a similar fashion (the mean of the six items, range = 1-4). The subscale has a reported reliability of alpha = .89. In this investigation, internal reliability was alpha = .77. Because this subscale refers to caregiving, it will be abbreviated PII-CG.
Controls. Data on demographic characteristics were collected from the participants to be used as potential control variables. Questions asked concerned fathers' and children's ages and gender, fathers' level of education, birth order of the child, number of children in the family, and fathers' ethnic background. Aside from intuitive reasons to include these characteristics as control variables (e.g., younger children require more care, so one would expect a possible confound with caregiving behavior), prior investigations of paternal identity and involvement have attempted to control for these characteristics because of significant observed relationships with the dependent variables (e.g., Maurer et al., 2001, reported that fathers engaged in more caregiving behaviors when there were more young children in the home).
Means and standard deviations for all of the identity and behavior variables measured were computed and appear in Table 1. The mean score of 2.63 out of 3.0 for the RSS indicates that fathers were most likely to say that the scale items were "very true" of them. The mean score of 3.66 out of 5.0 for the CGI indicates that fathers were most likely to "agree" with the scale items. The mean score of 121.60 degrees (out of 360) for the PIE-P equates to roughly 34% of the pie (which could be divided among four roles). The mean score of 67.97 degrees for the PIE-CG equates to roughly 19% of the pie (which could be divided among five domains). The mean score of 3.06 out of 4.0 for the PII indicates that fathers were most likely to report engaging in the behaviors "often," while the mean score of 3.03 for the PII-CG can be interpreted in the same way.
To test the hypotheses, correlational matrices were computed in two ways: zero-order correlations and partial correlations. The partial correlations removed any possibly confounding effects of the control variables. Z-transformations of the zero-order and partial correlations, and z-score comparisons, were conducted, and the results indicated that partialling out the effects of the control variables did not result in any significant differences from the zero-order correlations. Therefore, while both correlations will appear in Tables 2 and 3, only the zero-order correlations will be discussed in the text.
HYPOTHESES AND RESEARCH QUESTIONS
Hypothesis 1A. Role-level measures of identity will be significantly, positively correlated with role-level measures of behavior. This hypothesis was partially supported. Fathers' identity as measured by the role-level RSS was significantly correlated with the role-level PII, Pearson's r = .41, p < .01, but fathers' identity as measured by the role-level PIE-P was not significantly correlated with the PII, r = .09, ns. The results for all the hypotheses are presented in Table 2.
Hypothesis 1B. Domain-level measures of identity will be significantly, positively correlated with role-level measures of behavior. This hypothesis was fully supported. Fathers' identity as measured by the domain-level CGI was significantly correlated with the PII, r = .33, p < .01, as was fathers' identity as measured by the domain-level PIE-CG, r = .31, p < .05.
Hypothesis 2A. Role-level measures of identity will be significantly, positively correlated with domain-level measures of behavior. This hypothesis was not supported. Fathers' identity as measured by the role-level RSS was not significantly correlated with the domain-level PII-CG, r = .07, ns, nor was fathers' identity as measured by the role-level PIE-P, r = .04, ns.
Hypothesis 2B. Domain-level measures of identity will be significantly, positively correlated with domain-level measures of behavior. This hypothesis was fully supported. Fathers' identity as measured by the domain-level CGI was significantly correlated with the PII-CG, r = .40, p < .01, as was fathers' identity as measured by the domain-level PIE-CG, r = .52,p < .01.
Research question 1. What is the relationship between pie chart and scale measures of identity? To address this question, convergent and discriminant validity of the measures of paternal identity were examined, and the results are presented in Table 3. Scale measures of paternal identity were compared to pie-chart measures at both the role and the domain level. The correlation between the role-level scale measure RSS and the role-level pie-chart measure PIE-P was only r = .20, ns. This low convergent validity suggests that the RSS and the PIE-P may not be tapping into the same concept, and that any comparisons between them should be made with caution. (However, it should be noted that the low alpha of the RSS in this investigation has already called into question its appropriateness as a measure of paternal identity.)
The correlation between the domain-level scale measure CGI and the domain-level pie-chart measure PIE-CG was r = .59, p < .01. This high convergent validity suggests that the CGI and the PIE-CG are both tapping into the same aspect of paternal identity. Thus, while it appears that the role-level scale and pie-chart measures of paternal identity are only minimally related, the domain-level scale and pie-chart measures of paternal identity appear to be closely related.
Research question 2. Are pie-chart measures of identity, at both the role and domain level, equivalent (in magnitude of correlation with dependent variables) to scale measures of identity? To address this question, the correlations between scale measures of identity and the PII and PII-CG, and the correlations between pie chart measures of identity and the PII and PII-CG, were compared at the role and the domain level (but not across levels). Z-transformations of the correlations and z-score comparisons, were used to examine the significance of any difference between the correlations. Only one statistically significant difference emerged: the correlation between the role-level RSS and the role-level PII was significantly larger than the correlation between the role-level PIE-P and the PII, z = 2.69, p < .05. The RSS was significantly correlated with the PII, while the PIE-P appeared to be orthogonal to it. (However, as noted above, the RSS had low internal reliability, so this comparison may not be as substantive as it appears.)
The remaining three comparisons, RSS vs. PIE-P with PII-CG, CGI vs. PIE-CG with PII, and CGI vs. PIE-CG with PII-CG, were not statistically significant. Thus, it appears that pie-chart measures of identity at the domain level are equivalent to scale measures of identity, for both role- and domain-level measures of behavior. In contrast, while pie-chart measures of identity at the role level appear to be equivalent to scale measures of identity in correlations with domain-level behavior, pie-chart measures of identity at the role level do not appear to be equivalent to scale measures of identity in correlations with role-level behavior.
Two patterns emerged from the results of this study. First, the measures of role-level paternal identity, with one exception, appeared unrelated to paternal behavior at either the role or the domain level. Second, at the domain level, both scale and pie-chart measures of paternal identity were significantly correlated with paternal behavior, at both the role and the domain level, and the magnitude of the correlations did not significantly differ between scale and pie-chart measures.
With respect to the first pattern, these results call into question the efficacy of using role-level measures of paternal identity to predict paternal involvement. The role-level pie chart PIE-P, which asked participants to indicate how important the father role was to them, appeared completely orthogonal to both the role-level and the domain-level measures of involvement. The role-level scale RSS, though significantly correlated with the role-level measure of involvement, had low internal reliability and also appeared orthogonal to the domain-level measure of involvement. These findings are especially problematic when interpreted in light of paternal identity theory. According to the theory, high identification with the father role should equate to high levels of enactment of the behaviors associated with that role. Clearly, such was not the case in this investigation.
There are at least two possible explanations for this discrepancy. The first is that the role-level measures of paternal identity lacked construct validity and were not actually measuring the extent to which participants identified with the father role. Although this could explain the results from the PIE-P, it cannot fully explain the results from the RSS, which was significantly correlated with role-level involvement. Although the correlation between the RSS and the PII was suspect because of the low reliability of the RSS, it is possible that the RSS was, at least in part, tapping into some aspect of the father role.
The second possible explanation for the discrepancy between the theoretical predictions and the results is as follows: the results from the RSS may suggest that the modern father role is too ambiguous to be associated at the role-level with specific domain-level paternal behaviors. Traditionally, parenting role expectations have been defined by gender, with men being expected to contribute little in the way of child-oriented tasks, such as caregiving (Aryee & Luk, 1996; Lamb, 1995). However, the nature of modern American fatherhood is changing, especially with the changes in patterns of women's employment, and many traditionally "female" tasks, such as caregiving, can now be internalized by fathers as one of the sets of role expectations for the role of father (Parke, 2002), with "can" being the operative word. Precisely because the father role is changing, the meaning of being a father, and the expectations for paternal involvement, may vary widely from one father to another. This could certainly explain how in this investigation, fathers who reported a high father role identity on the RSS were also likely to report high levels of involvement on the role-level PII (which included behaviors in three domains), but those same high levels of identity were unrelated to domain-level caregiving involvement (PII-CG). These fathers may identify with the domains of the father role other than caregiving that were assessed in the PII (i.e., teaching functions, socioemotional functions, executive functions), but not in the PII-CG. This interpretation suggests that future research utilizing role-level measures of paternal identity use caution in selecting measures of paternal involvement so that the multiple domains that comprise the role of father can be more fully assessed.
Alternatively, future research could focus more exclusively on the domain level. In this investigation, domain-level measures of paternal identity were significantly correlated with both role- and domain-level measures of involvement, using both scale and pie-chart measures. Our analyses indicated no significant differences in the magnitude of correlations between the identity and involvement measures when comparing scale and pie-chart measures of identity. These results have two primary implications. First, they suggest the efficacy of using pie charts as replacements for, or supplements to, more traditional measures of paternal identity, such as scale measures. Second, they suggest that using domain-level measures of paternal identity to predict role-level paternal involvement may be more prudent than using role-level measures of identity to predict domain-level involvement.
There is one additional advantage to assessing paternal identity and behavior at the domain level: a popular question both in the literature, and among the public and in government, is how to get fathers more involved in domain-level activities (e.g., caregiving, teaching functions) with their children. Assessing paternal identity and behavior at the domain level appears to be able to more directly answer this question than a more global, role-level assessment.
Several limitations of this investigation should also be discussed. The RSS, which had a reported reliability of alpha = .61 (Bruce & Fox, 1999), only achieved an alpha of .41 in our analyses. As discussed above, this low reliability calls into question any results associated with the RSS. Further, the three-point response format for the RSS may have been responsible for both the low internal reliability and the small correlations with the involvement measures by reducing the available variance. Since the only middle ground between the response points of "not true" and "very true" was "somewhat true," respondents were forced to choose between the two extreme ends of the scale and a single middle point; more variance may have appeared in a five-point response format that would have allowed fathers to answer above the middle point, but below the highest extreme. Future research on role-level identity measures should incorporate a response format that allows for greater variance in respondents' answers to more fully investigate the potential of assessing identity at the role level.
This investigation is also limited by shared-method variance (Marsiglio et al., 2000), a type of measurement error variance attributable to respondents filling out both independent and dependent measures, without outside sources of data to verify the relationship (e.g., correlating father identity and father involvement, both as reported by the father, instead of correlating self-reported father identity and mother-reported father involvement). This type of measurement error can be responsible for inflating observed correlations between independent and dependent variables, and should be noted as a possible limitation of this investigation.
Third, the sample in this investigation was 64 middle-aged, college-educated fathers, most of whom were white. Although this is in some ways beneficial, since it reflects the types of convenience samples that are often used in fatherhood research (e.g., Maurer et al., 2001) and as such allows for a more direct comparison and evaluation of the measures used, it clearly is disadvantageous in underrepresenting younger and older fathers, fathers with less education, and fathers of color. The limited external validity of this investigation is a serious shortcoming and should be addressed in future research.
In addition, this investigation in general raises several more questions for future research. First, measures of paternal involvement, like measures of paternal identity, are frequently scale measures. Would pie-chart measures of paternal involvement demonstrate the same potential as pie-chart measures of paternal identity? Would the combination of pie-chart measures of both identity and involvement be more or less advantageous than the combination of scale measures, or one scale and one pie-chart measure? Second, would investigating another domain within the father role (e.g., breadwinning) yield a similar pattern of results, or are the findings unique to the caregiving domain? Finally, would research on the mother role and domains yield a similar pattern of results, suggesting a larger parental model, or is this approach uniquely effective for fathers? Answers to these questions may help to more fully illuminate the relationship between paternal identity and paternal involvement.
Table 1 Means and Standard Deviations for Identity and Behavior Variables Standard Construct Means Deviations Identity Scale RSS (role-level) 2.63 (.20) CGI (domain-level) 3.66 (.38) Pie Chart PIE-P (role-level) (a) 121.60 (31.70) PIE-CG (domain-level) (a) 67.97 (24.82) Behavior PII (role-level) 3.06 (.45) PII-CG (domain-level) 3.03 (.66) (a) number of degrees (out of 360) Table 2 Correlations Between Identity and Behavior Variables RSS CGI PIE-P PIE-CG (role- (domain- (role- (domain- Variable level) level) level) level) Zero-Order Correlations PII (role-level) .41 ** .33 ** .09 .31 * PII-CG (domain-level) .07 .40 ** .04 .52 ** Partial Correlations PII (role-level) .38 ** .35 * .10 .31 * PII-CG (domain-level) .04 .42 ** .05 .55 ** * p < .05 ** p < .01 Table 3 Intercorrelations Between Identity Variables Scale Pie Chart RSS CGI PIE-P PIE-CG Scale RSS (role-level) -- .13 .23 .05 CGI (domain-level) .12 -- .02 .59 ** Pie Chart PIE-P (role-level) .20 .03 -- .08 PIE-CG (domain-level) .07 .59 ** .13 -- ** p<.01. Note. Correlations below the diagonal are zero-order correlations; correlations above the diagonal are partial correlations.
This material is based upon work supported by the Cooperative State Research, Education and Extension Service, U.S. Department of Agriculture, under Project NO. ILLU-45-0329 to Joseph Pleck and by a Jonathan Baldwin Turner Fellowship from the University of Illinois to Trent Maurer. We would like to acknowledge the following for their contribution to this project: David Hansen and Jeff Stueve; Linda Culton, Terrina Ellerson, Nancy Plane, and Angela Smith; and Elizabeth Crawford, Jenna Grell, Corrie Haas, Sue Jurgovan, Tiffany May, Vicki Nolan, Iraida Rios, and Tina Siler.
Aryee, S., & Luk, V. (1996). Balancing two major parts of adult life experience: Work and family identity among dual-earner couples. Human Relations, 49, 465-487.
Bonney, J.F., Kelley, M.L., & Levant, R.F. (1999). A model of fathers' behavioral involvement in child-care in dual-earner families. Journal of Family Psychology, 13, 401-415.
Bruce, C., & Fox, G.L. (1997, November). Measuring father involvement among lower-income white and African-American populations. Poster session presented at the annual conference of the National Council on Family Relations, Arlington, VA.
Bruce, C., & Fox, G.L. (1999). Accounting for patterns of father involvement: Age of child, father-child coresidence, and father role salience. Sociological Inquiry, 69, 458-76.
Burke, P.J., & Cast, A.D. (1997). Stability and change in the gender identities of newly married couples. Social Psychology Quarterly, 60, 277-290.
Cowan, C.P., Cowan, P.A., Heming, G., Garrett, E., Coysh, W.S., Curtis-Boles, H., et al. (1985). Transitions to parenthood: His, hers, and theirs. Journal of Family Issues, 6, 451-481.
Day, R., & Lamb, M. (in press). Conceptualizing and assessing father involvement. Mahwah, NJ: Lawrence Erlbaum & Associates.
Futris, T.G., Pasley, K., Kerpelman, J., & Goodman, R.M. (1998, November). The effects of work and paternal identity prominence on behavior and life satisfaction of fathers. Poster session presented at the annual conference of the National Council on Family Relations, Milwaukee, WI.
Ihinger-Tallman, M., Pasley, K., & Buehler, C. (1993). Developing a middle-range theory of father involvement postdivorce. Journal of Family Issues, 14, 550-571.
Lamb, M.E. (1995). The changing roles of fathers. In DJ. Sonkin (Series Ed.) & J.L. Shapiro, M. J. Diamond, & M. Greenberg (Vol. Eds.), Springer series, Focus on men: Vol. 8. Becoming a father: Contemporary, social, developmental, and clinical perspectives (pp. 18-35). New York: Springer Publishing.
Marsiglio, W., Amato, P., Day, R.D., & Lamb, M.E. (2000). Scholarship on fatherhood in the 1990s and beyond. Journal of Marriage and Family, 62, 1173-1191.
Maurer, T.W., Pleck, J.H., & Rane, T.R. (2001). Parental identity and reflected-appraisals: Measurement and gender dynamics. Journal of Marriage and Family, 63, 309-321.
Minton, C., & Pasley, K. (1996). Fathers' parenting role identity and father involvement: A comparison of nondivorced and divorced, nonresident fathers. Journal of Family Issues, 17, 26-45.
NICHD Early Child-care Research Network. (2000). Factors associated with fathers' caregiving activities and sensitivity with young children. Journal of Family Psychology, 14, 200-219.
Parke, R.D. (2002). Fathers and families. In M. Bornstein (Series & Vol. Ed.), Handbook of Parenting: Vol. 3. Being and becoming a parent (2nd ed., pp. 27-73). Mahwah, NJ: Lawrence Erlbaum.
Strauss, R., & Goldberg, W.A. (1999). Self and possible selves during the transition to fatherhood. Journal of Family Psychology, 13, 244-259.
Stryker, S. (1987). Identity theory: Developments and extensions. In K. Yardley & T. Honess (Eds.), Self and identity: Psychosocial perspectives (pp. 89-103). New York: Wiley.
Yeung, W.J., Sandberg, J.F., Davis-Kean, P.E., & Hofferth, S.L. (2001). Children's time with fathers in intact families. Journal of Marriage and Family, 63, 136-154.
TRENT W. MAURER
JOSEPH H. PLECK
University of Illinois at Urbana-Champaign
THOMAS R. RANE
WASHINGTON STATE UNIVERSITY
Correspondence concerning this article should be addressed to Trent W. Maurer, 167 Bevier Hall MC-180, 905 S. Goodwin Ave., Urbana, IL 61801. Electronic mail: email@example.com.
|Printer friendly Cite/link Email Feedback|
|Author:||Rane, Thomas R.|
|Date:||Jun 1, 2003|
|Previous Article:||From the editor.|
|Next Article:||The inner father.|