Development and Validation of a Father Involvement in Health Measure

Abstract

Father involvement has been associated with positive child social, emotional, psychological, developmental, and health outcomes. However, tools for measuring father involvement have not kept pace with the expanding understanding of the roles of fathers, and in the area of child health, are blunt. The purpose of this study was to develop and validate a self-report measure of father involvement in preschooler’s health, the Father Involvement in Health–Pre-School (FIH-PS). In phase 1 item generation, 47 items were developed based on previous qualitative work and vetted through cognitive interviews with 21 fathers of children ages 3–5 (preschool). In phase 2 psychometric validation, 560 fathers of 3–5 year olds (n=392 resident, n=168 non-resident) completed the FIH-PS item bank. Participants were predominantly white (64%), had private health insurance (53%), had a mean age of 33 years, and half were married. Item Response Theory was used to determine measurement scoring. The FIH-PS scale was reduced from a 47-item bank to a total of 20 items supporting a 4-factor scale made up of Acute Illness, General Well-being, Emotional Health, and Role Modeling. Following exploratory (n=280) and confirmatory factor (n=280) analyses, the scale followed a bifactor structure, was internally consistent (Cronbach’s α=0.953), and discriminated among fathers with lower involvement. A sum-to-T-score crosswalk table was produced to standardize the scores along a normal distribution (mean=50, S=10, range 10.8–71.3). Future research and clinical applications of the FIH-PS are discussed.

Over the past 50 years, there has been nearly a threefold increase in the time fathers spend with their children (Livingston, 2013). Father involvement in the United States is now at an all-time high, in part, due to societal expansion of the paternal role. The roles of fathers have shifted from a focus on being a “breadwinner” to involvement in childcare, emotional nurturance, and coparenting (Harrington, Van Deusen, & Ladge, 2010). Since 1989, the number of stay-at-home fathers has nearly doubled to almost 2 million (Livingston, 2014) and the convergence of men and women’s work patterns (Connelly, 2016; Kan, Sullivan, & Gershuny, 2011; Livingston, 2013) has found more fathers involved in childrearing. The 2016 American Academy of Pediatrics report on fathers’ roles in child care and development highlights the surge in research on fathers in families and the benefits for children (Yogman, Garfield, & Committee on Psychosocial Aspects of Child Family Health, 2016). A primary finding from the existing literature is that positive father involvement is associated with an array of beneficial psycho-social-developmental outcomes in infancy through adolescents (e.g., Lamb, 2010; Wilson & Prior, 2011).

The growing body of research on father involvement has both reinforced and reflected a “New Dad” paradigm in the United States (Harrington, Van Deusen, & Humberd, 2011). Many fathers in the 21^st century are combining their workplace responsibilities with being caregivers, and thereby defining a “good father” as one who includes caregiving characteristics such as providing love and emotional support, being a teacher and guide, and contributing to the daily tasks of childcare (Harrington et al., 2011). Growing evidence also points to the utility of using a theory of generativity when working with fathers that focuses on how becoming a father and caring for children can lead to fulfillment and can be a lever for change (Fleming, Hunt & King, 2014). Such scholarship underscores a shift in men’s perceptions, values, and attitudes about fathering from previous generations. Measurement methods need to keep pace with such changes to improve understanding of how those changes impact the health of children and families.

Theory and Measurement of Father Involvement

Research on fathers is now benefiting from maturing conceptualizations of fatherhood and father involvement. The prevailing model of father involvement introduced over thirty years ago (Lamb, Pleck, Charnov, & Levine, 1985) posited that father involvement consists of three main components: accessibility (availability to spend time with child), engagement (father’s direct interaction with his children), and responsibility (planning, monitoring, and supervising roles). Despite the three components, engagement became the most studied construct and the model became synonymous with active caregiving behaviors (Pleck, 2012). Engagement behaviors such as reading to a child and playing games with a child were relatively easier to operationalize through time diary and survey research than more complex concepts around responsibility or accessibility. However, the sole focus on engagement behaviors was critiqued by emphasizing quantity over other constructs and components of father involvement such as quality, parenting style, identity, and emotional nurturance (Wilson & Prior, 2011). Pleck (2010) presented a new father involvement model to reflect current empirical trends about fathering behavior. This model of father involvement consists of three domains – positive engagement activities, warmth and responsiveness, and control as well as two auxiliary domains – indirect care and process responsibility (Pleck, 2010). Lamb and Pleck added depth to fatherhood models by proposing that father involvement is generated by motivation, skills, self-confidence, social supports, and the absence of institutional barriers (Pleck, 2012). With this theoretical maturation, father involvement and fathering research has benefited from a deeper, more nuanced conceptualization base that reflects the lived experiences of fathers. Father involvement has evolved from a unidimensional to a multidimensional construct.

Operationalization of the multiple dimensions of the father involvement construct has been difficult and faces many methodological challenges. Previous father involvement measures have been limited, neglecting contextual factors (e.g., father financial contributions, marital status), focusing solely on the mother’s perspective, relying on a single item (e.g. “Have you taken your child to the doctor?”), using a mother-child template to study father-child involvement, only measuring one component of father involvement (e.g., engagement behaviors) and missing the multiple components and the variety of ways that fathers may be involved with their child (Day & Lamb, 2003; Fagan & Kaufman, 2014; Feinberg, Brown, & Kan, 2012; Hawkins & Palkovitz, 1999; Lu et al., 2010; Sayers & Fox, 2005). Father involvement measures are therefore not reflective of current cultural and gender role shifts in many modern fathers and families (Day & Lamb, 2003; Fagan, Day, Lamb, & Cabrera, 2014). As a result, a need to develop new measures of father involvement that keep pace with current conceptualizations of father involvement has emerged as a priority (Singley et al., in press). New measures that marry theoretical definitions and valid measurement approaches hold the most promise for advancing the science and understanding the current state of fathers and families (Heppner et al., 2016).

Father Healthcare Involvement and Preschool-Aged Children

In addition to emphasizing the need to develop multidimensional measures of father involvement, fatherhood scholars have called for focused examination of fathering in specific domains such as the health and healthcare of their child (Garfield & Isacco, 2006). A measure of father involvement in the health and healthcare of preschool-age children is needed as the environmental milieu for these children begins to extend beyond the home and into daycare and preschool, with more complex social needs of attachment and exploration (Cath, Gurwitt, & Ross, 2013; Yogman, 1994). Father involvement at this age is associated with a number of psychosocial and biobehavioral outcomes, including cognitive, emotional and social development and adjustment capacities (Brown, Mangelsdorf, & Neff, 2012; Cabrera, Shannon, & Tamis-LeMonda, 2007; Cook, Roggman, & Boyce, 2011; Downer & Mendez, 2005; Jeynes, 2015; Lamb, 2004; Lamb & Lewis, 2013; McWayne, Downer, Campos, & Harris, 2013; Pancsofar & Vernon-Feagans, 2006; Pougnet, Serbin, Stack, & Schwartzman, 2011). Fathers’ childrearing practices related to the health and healthcare domain have been linked to an array of child and parent health outcomes including reading, discipline and punishment, language development, and nutrition (e.g., Garfield & Isacco, 2012; Garfield et al., 2014; 2015).

Earlier research involving fathers in the national, longitudinal birth cohort study Fragile Families and Child Wellbeing study qualitatively categorized the myriad of ways fathers promote the health and wellbeing of their pre-school age children. The conceptualization emerging from these data indicated a categorization of various father-child healthcare domain activities according to the three components of Lamb’s foundational father involvement model: engagement, responsibility, and accessibility (Garfield & Isacco, 2012). Accessibility in the healthcare domain consisted of the father maintaining a physical presence in his child’s life, which enabled the father to accrue a baseline knowledge of his child’s health and notice any detrimental changes. Engagement in the healthcare domain entailed the father administering medications to the child, taking the child to sick visits, playing/exercising with their child, and providing nutritious meals. Responsibility in the healthcare domain consisted of the father tracking immunizations, providing health insurance, selecting a doctor, and monitoring their child’s health. The theoretical underpinning for the FIH-PS measure of father involvement draws directly from the conceptualization advanced in our prior qualitative findings.

In addition, an increased awareness of the contextual factors that impact father involvement in health and healthcare underscore the need for more advanced involvement measures. Fathers’ residency status is known to effect involvement. Most research has found that residential fathers are more accessible to their children and have a more positive relationship with the child’s mother than non-residential fathers, which increases their opportunities for involvement (Goldberg, 2013). Non-residential fathers may experience substantial role ambiguity and little decision-making power, which further decreases their child involvement (Insabella, Williams, & Pruett, 2003; Leite & McKenry, 2002; Madden‐Derdich & Leonard, 2000). Accounting for residential status in any measurement of father involvement is important to track how various family structures impact child and adolescent health outcomes (Carlson, 2006). Yet, a psychometrically-sound measure of father involvement in their child’s health and healthcare has not yet been developed for non-resident and resident fathers.

As a counterpoint to the rationale for a distinct father involvement measure, Fagan and colleagues (2014) argued for a move towards gender-neutral models of parenting. The basic rationale is that fathers and mothers assume overlapping roles in families and there is not enough empirical support for separate involvement constructs for mothers and fathers (Fagan et al., 2014). Nevertheless, similar to other recent scale development efforts of father involvement (Singley et al., in press), we describe how our study is situated within an “essential father” perspective. We acknowledge that fathers and mothers may play similar, overlapping roles in some families, which create a shared effect of parenting on child psych-social-developmental outcomes. Yet, mothers and fathers also assume distinct roles in their families and make unique contributions to their child’s development. We disagree with scholars such as Silverstein and Auerbach (1999) and assert that fathers have a unique and essential role in child development. Indeed, the importance of fathers’ unique and essential role in child development has been underscored by the American Academy of Pediatrics (Yogman & Garfield, 2016). Similarly in the field of psychology, fathers have been identified as contributing to child psychopathology and treatments in unique ways (e.g., Bogels & Phares, 2008). The essential father perspective has some empirical support; longitudinal studies have found statistically significant correlations between father constructs and child outcomes even when controlling for mother involvement, mother engagement, and mother sensitivity (e.g., Amato & Rivera, 1999; Ramchandani et al., 2013; Tamis-LeMonda, Shannon, Cabrera, & Lamb, 2004). The extant literature has found that fathers may play unique roles with their children through active play and gross motor movements that foster body control, risk-taking, and secure attachment relationships that are separate from mothers (e.g., Berger, 2009; Grossman et al., 2002; Paquette & Dumont, 2013). The unique role(s) fathers play with their child and in their family are diverse and likely to reflect a number of factors such as culture, personality, coparenting relationship, and division of labor within the household. Thus, we conclude an essential father perspective based on solid theoretical and empirical underpinnings can contribute to developing a richer understanding via measurement of fathers’ involvement in the unique domain of health.

Summary and Purpose of the Current Study

Taken together, we have reasoned that a new self-report measure of father involvement in their child’s health is needed for several reasons: 1) to match current conceptualizations of father involvement as a multidimensional construct, 2) to focus on the important healthcare domain for preschool-aged children, 3) to account for contextual factors such as father residential status, and 4) to examine the unique contributions that fathers make to child development. A measure with sufficient validity would have clinical and research utility as a way to associate current father involvement with child or parent outcomes as well as a way to assess the effectiveness of future interventions and programs on father involvement in health. The purpose of this study is to develop the Father Involvement in Health – Pre-School (FIH-PS) tool, as a psychometrically-sound, self-report measure of father involvement in preschooler’s health, which is based on and evolves from prior qualitative work and reflects current conceptualizations of father involvement.

Method

Phase 1: Item Development, Constructs, and Initial Questionnaire

Based on prior published work involving a multi-step content analysis study analyzing qualitative data from 32 urban fathers of pre-school aged children (Garfield & Isacco, 2012), four specific, developmentally-appropriate, clinically-relevant domains of father involvement were developed: acute illness (non-life threatening short-term illness), general well-being (health aspects necessary for human survival and homeostasis), emotional health (child’s emotional well-being), and role modeling (socializing behaviors that promote child health, are observable, and can be internalized by the child). Each domain was created within Lamb’s fathering domains (engagement, responsibility, and accessibility) that specifically reflected paternal involvement in child health and healthcare. Engagement by the father in their child’s healthcare was defined by hands-on healthcare activities such as administering medication. Responsibility was defined by the father’s overall internalization of the duty to be involved in the planning, monitoring, supervising roles that maintain the child’s health, ensuring that their child’s healthcare is covered (e.g., health insurance), and making healthcare decisions. Accessibility was defined by the father’s availability to spend time with his child to understand their child’s health, recognize changes to the child’s health, and to be available to take action related to the child’s health care.

A panel of fathering researchers and practitioners (all are authors of this study) with extensive knowledge in paternal involvement created the FIH-PS items based on this paradigm. The panel met regularly for 4 months to translate qualitative themes from the a priori paradigm into sample test items. The panel discussed each item, edited as necessary, and ensured that the items accurately reflected the qualitative themes. Consensus among the panel members was needed to keep an item in the bank for further review in the cognitive interviews. Item language was tested at a 5^th grade readability level using a Flesch-Kincaid Readability Test (Kincaid, Fishburne, Rogers, & Chissom, 1975). The initial version of the scale contained 47 items of acute illness (n=13), general well-being (n=14), emotional health (n=9), and role modeling (n=11). Our Institutional Review Board approved this and every stage of the study.

Preliminary Item Analysis

Following the initial item development, cognitive interviews with fathers were used to detect if potential measure items were understood and comprehendible to fathers (Ryan, Gannon-Slater, & Culbertson, 2012). Fathers (n=21) of a preschool-aged child (3 to 5-years-old) were asked to complete the initial list of FIH-PS items and participate in a follow-up phone interview. Participants were recruited through online networks via email list serves, which directed them to the study website (http://fatherinvolvement.weebly.com/). Subjects completed the questionnaire through REDCap, an electronic database (Harris et al., 2009). Follow-up interviews lasted between 40 and 60 minutes. Participants were compensated with a nominal gift card for the entire process. Eighty-six percent (n=18) identified as Caucasian and 14% (n=3) as African-American, with an average age of 38-years-old.

All cognitive interviews were audiotaped over the phone by an experienced qualitative interviewer and transcribed verbatim. Each interview was semi-structured using a standardized set of questions to assess each item and item response wording, literacy, and comprehension. Item responses and cognitive interview transcriptions were analyzed to determine necessary modifications to the initial set of FIH-PS items by examining participant comments, item response percentages, and presence of variance in item selection, mean, and standard deviation. The research team met to make revisions on items that were misinterpreted by participants. For example, the item, “I ask my child how he/she is feeling” was revised to “I ask my child how he/she is feeling emotionally.” This was to clarify that the item assesses the child’s general emotions instead of how the child is feeling when sick. The items were then finalized for reliability and validity evidence.

Phase 2: Measure Finalization and Data Analysis

Participants

A nationally representative sample of fathers (N=560) completed the survey online. The sample was drawn from an online panel of individuals across the US who are eligible to participate in a limited number of research studies per month for small monetary compensation (Qualtrics, Provo, UT). Participants needed to meet three inclusion criteria to be eligible to complete the survey: 1) male, 2) at least 18-years-old at the time of recruitment, and 3) a father of a child between the ages of 3–5 years old. Prior to completing the FIH-PS, fathers completed basic demographic questions (race and ethnicity, income, age, and education), residential status, their relation to their child (biological father, step father, legal guardian, or adoptive father), their current physical and mental health, the health of their child, and the status of their relationship with their child’s mother. Descriptive statistics are found in Table 1.

Table 1.

Sociodemographic Data of the Sample of Fathers

	Total Sample (N = 560)	Resident Fathers (N = 392)	Non-resident Fathers (N = 168)
Age	33.6 (7.0)	34.4 (6.9)	31.8 (6.8)
Child’s age
3-years-old	135 (24%)	98 (25%)	37 (22%)
4-years-old	236 (42%)	161 (41%)	75 (44%)
5-years-old	189 (34%)	133 (34%)	56 (33%)
Child’s sex
Male	310 (55%)	224 (57%)	86 (51%)
Female	250 (45%)	168 (43%)	82 (49%)
First Child
Yes	325 (58%)	226 (58%)	99 (59%)
No	235 (42%)	166 (42%)	69 (41%)
Sexual Orientation
Heterosexual	537 (96%)	366 (96%)	161 (96%)
Homosexual	11 (2%)	7 (2%)	4 (2%)
Bisexual	12 (2%)	9 (2%)	3 (2%)
Current Relationship
Single	155 (28%)	59 (15%)	96 (57%)
Married or Civil Union	281 (50%)	269 (69%)	15 (9%)
Separated/Divorced	3 (1%)	27 (7%)	51 (30%)
Lives with Partner	38 (7%)	33 (8%)	5 (3%)
Widowed	5 (1%)	4 (1%)	1 (1%)
Health Insurance
Private	299 (53%)	248 (63%)	51 (30%)
Medicare	71 (13%)	50 (13%)	21 (13%)
Medicaid	34 (6%)	21 (5%)	13 (8%)
Military Health Care	18 (3%)	13 (3%)	5 (3%)
Indian Health Service	3 (1%)	1 (1%)	2 (1%)
State-Sponsored Health	31 (6%)	13 (3%)	18 (11%)
Other Government Plan	14 (3%)	9 (2%)	5 (3%)
Single Service Plan	12 (2%)	9 (2%)	3 (2%)
None	62 (11%)	22 (6%)	40 (24%)
Don’t Know	16 (3%)	6 (2%)	10 (6%)
Race
White	356 (64%)	255 (65%)	101 (60%)
Black	83 (15%)	49 (13%)	34 (20%)
Native American	7 (1%)	5 (1%)	2 (1%)
Asian	24 (4%)	18 (5%)	6 (4%)
Latino	51 (9%)	37 (9%)	14 (8%)
Multiracial	39 (7%)	28 (7%)	11 (7%)
Education
Some High School or less	37 (7%)	17 (4%)	20 (12%)
High School Diploma or GED	133 (24%)	75 (19%)	58 (35%)
Some College	142 (25%)	92 (23%)	50 (30%)
Technical/Trade School/Adult Basic Education	38 (7%)	25 (6%)	13 (8%)
Bachelor’s Degree	142 (25%)	125 (32%)	17 (10%)
Graduate or Professional School	68 (12%)	58 (15%)	10 (6%)
Employment
Employed	445 (79%)	332 (85%)	113 (67%)
Unemployed	115 (21%)	60 (15%)	55 (33%)
Total Household Income
Less than $25,000	130 (24%)	68 (17%)	62 (37%)
$25,000 to $74,999	248 (44%)	168 (43%)	80 (48%)
$75,000 to $149,999	148 (26%)	124 (32%)	24 (14%)
$150,000 or more	34 (6%)	32 (8%)	2 (1%)

Procedures

The initial measure of father involvement in health was comprised of 47 items. Responses were coded 1 through 5 with 1 representing “Not at all” and 5 representing “Very much”. Higher scores indicated a higher level of father involvement in child health on the measure. Psychometric methods were used to evaluate item and scale properties. Internal consistency was measured using Cronbach’s α and item-total correlations with α values ≥ 0.8 indicating internal consistency and correlations ≥ 0.4 indicating internal consistency (Tavakol & Dennick, 2011; Clark & Watson, 1995). These values were estimated using PROC CORR in SAS 9.4.

Item Response Theory (IRT) was used to develop the measurement. IRT assesses test item data by providing a unified statistical process for estimating the traits of items and test takers (Brennan et al; 2006). Applications such as item and test construction, adaptive test administration, scaling and equating, are among others uses of IRT. A focus of IRT analyses is relating the characteristics of individual items to the whole test. One of the foundations of IRT is its assumption that the probability that a test taker with a given latent trait will endorse a particular response to a particular item. Thus, IRT models relate item scores to the test taker’s latent trait using nonlinear, probabilistic functions. The predictions provided by IRT analysis are flexible and typically expressed using scores (such as the T-score) rather than tallies of correct scores. Once the IRT assumptions were assessed, a bifactor IRT model was fit to the data using a graded response model in IRTPro 2.1. Raw to T-score conversion tables were created from these IRT parameters.

A bifactor model structure of a general father involvement and four subscales was tested and confirmed using exploratory and confirmatory bifactor analyses, respectively. The analyses were chosen on the grounds that the data appeared to conform to a bifactor structure. A general “fatherhood involvement” factor informed the individual items as well as separate, related factors connected to involvement in health. A weighted subset of 280 participants was selected from the 560 total participants for the exploratory bifactor analysis and the 280 remaining participants were used for the confirmatory bifactor analysis. The weighting scheme followed the sampling scheme where 70% (n = 196) of the 280 participants in each sample were resident fathers and 30% were non-resident fathers. The exploratory bifactor analysis used a bifactor rotation (Jennrich & Bentler, 2011). Items with factor loadings that were ≥ 0.450 for both the general factor and one specific factor were included in the confirmatory BFA model. Items that met this criterion for only the general factor were excluded from the confirmatory model since they were not strongly associated enough with any of the four specific factors. No items met the criterion for the specific factor or the general factor and no items loaded onto more than one specific factor. All items with factor loadings ≥ 0.450 on both the general factor and one specific factor had factor loadings < 0.20 for all other factors. This threshold was chosen based on Hair, et al.’s (1998) criteria of being practically significant in tandem with Comrey & Lee’s (1992) criterion for a factor loading being of a “fair” degree. In addition to the compromise between Hair (1998) and Comrey & Lee (1992), a less stringent cut-off of 0.450 accommodated the strong covariance between several of the items and the skewed distribution of the item responses.

Local independence was tested by examining the residual correlations from the confirmatory bifactor analysis where r ≥ |0.15| indicated local dependence (Reeve et al., 2007). Model fitness was tested using the comparative fit index (CFI) with a good-fit criterion of ≥ 0.95, the Tucker-Lewis index (TLI) with a good-fit criterion of ≥ 0.95, the root mean square error of approximation (RMSEA) with a good-fit criterion of ≤ 0.06, and the weighted root-mean-square residual (WRMR) with a good-fit criterion of ≤ 1.0 (Kenny, 2015). Exploratory and confirmatory bifactor analyses were estimated using Mplus 6.12. Weighted least squares mean-and variance-adjusted (WLSMV) estimation was used for rotation. Residual correlations were estimated using the psych package in R 3.1.1. Monotonicity was assessed using Mokken scale analysis (MSA). The tests for monotonicity are used to assess whether the IRT function is strictly non-increasing or non-decreasing (i.e. the IRT function cannot follow a sine function or any other oscillating function). This assumption is important for IRT because it is consistent with one of the IRT assumptions—the probability of endorsing an item response changes directly with the increase in trait level. That is, as the trait level increases, then the probability of endorsing a specific response level will look one of two ways:

$$\mathrm{P}(\mathrm{X}=1)\ge \mathrm{P}(\mathrm{X}=2)\ge \mathrm{P}(\mathrm{X}=3)\ge \mathrm{P}(\mathrm{X}=4)\ge \mathrm{P}(\mathrm{X}=5)$$

or:

$$\mathrm{P}(\mathrm{X}=1)\le \mathrm{P}(\mathrm{X}=2)\le \mathrm{P}(\mathrm{X}=3)\le \mathrm{P}(\mathrm{X}=4)\le \mathrm{P}(\mathrm{X}=5).$$

MSA values of H ≥ 0.3 indicated at least weak monotonicity. The MSA was conducted using the mokken package in R 3.1.1.

Comparative measure for validation

The FIH-PS was administered alongside the Parenting Self-Efficacy Scale (PSES; Suzuki, Holloway, Yamamoto, & Mindnich, 2009) and the Inventory of Father Involvement (IFI; Hawkins et al, 2002) for validity analyses.

The PSES is a 25-item measure of parent’s self-confidence in their parenting abilities with preschool-aged children. Participants are instructed to rate how confident they feel in doing a various parenting task specified in each item, on a 6-point scale, ranging from 1 (not at all confident) to 6 (very confident). Higher scores indicate higher levels of parenting self-efficacy. In previous studies, internal consistency ranged from .87 to .92 (Suzuki et al., 2009); the internal consistency range from .94 – .96 in new versions of the scale (Holloway et al., 2016). In this study, the obtained mean score was 131.04 (SD = 14.73), and Cronbach’s Alpha = 0.9501. Two subscales of the PSES (Accepting as an Individual and Positive Evaluation) were used in the analyses of validity.

The IFI is a 26-item measure of father involvement that reflects a multi-dimensional concept of father involvement developed using a sample of 723 fathers with children of all ages who were largely married and White. Fathers were asked to rate “how good a job” they were doing on each indicator of father involvement with response choices 0 through 6, with 0 anchored by “Very Poor” and 6 anchored by “Excellent.” The IFI included 9 dimensions: Discipline and Teaching Responsibilities, School Encouragement, Mother Support, Providing, Time and Talking Together, Praise and Affection, Developing Talents and Future Concerns, Reading and Homework Support, and Attentiveness. Cronbach’s alpha ranged from .69–.87. Six IFI subscales (Discipline, Providing, Time and Talk, Praise and Affection, Developing Talent, and Attentiveness) were used in the analyses of validity.

As an understudied area, few fatherhood involvement measures exist that are developmentally appropriate for the pre-school age range. Therefore to examine evidence of concurrent validity for the FIH-PS subscale scores (Acute Illness, General Well-being, Emotional Health and Role Modeling), we explored the correlations between the four subscale scores and the PSES and the IFI measuring theoretically-related constructs. However, since most previous research has used full-scale father involvement scores and few studies have specifically examined father healthcare involvement, there is no clear theoretical or empirical guidance on how specific dimensions of father healthcare involvement measured on the FIH-PS will or will not relate to these criterion variables. Previous research has found a positive correlation between father involvement and parenting self-efficacy (Coleman & Karraker, 2003). A qualitative study (Garfield & Isacco, 2012) found that self-efficacy contributed to the likelihood that fathers administered medications to their child, providing some preliminary rationale for a positive relationship between Acute Illness involvement and parenting self-efficacy. Similarly, qualitative findings (Garfield, Isacco, & Bartlo, 2010) indicated that fathers appeared more confident to role model health behaviors to their child if they held positive perceptions of their own health; thus, we hypothesized a positive relationship between parenting self-efficacy and Role Modeling involvement. As no similar qualitative findings were available for General Wellbeing or Emotional Health, to remain consistent with literature on overall father involvement and the reported qualitative findings, we hypothesized that General Wellbeing and Emotional Health would be positively associated with parenting self-efficacy. Convergent and discriminant validity were assessed using Evans’ (1996) empirical classifications of correlation strength. In Evans’ classification system, 0.0 ≤ r < 0.20 is considered very weak, 0.20 ≤ r < 0.39 is considered weak, and 0.40 ≤ r < 0.60 is considered moderate. Very weak correlations were classified as having evidence of discriminant validity, weak correlations had indeterminate concurrence (i.e. the conclusion was no evidence of convergent or discriminant validity existed), and moderate correlations were classified as having evidence of convergent validity. The assumptions about the relationships between the FIH-PS subscales and the PSES and IFI subscales were based on no correlations being strong, but some correlations being very weak.

The general well-being subscale was hypothesized to have evidence of convergent validity with both PSES subscales, and to show evidence of discriminant validity with all of the IFI subscales. The acute illness subscale was hypothesized to show evidence of discriminant validity or no evidence of any convergent or discriminant validity with any of the PSES or IFI subscales, as neither the PSES nor IFI directly measure involvement in health. The emotional health subscale was hypothesized to show evidence of convergent validity with both PSES subscales and all of the subscales of the IFI except for discipline and providing. The role modeling subscale was hypothesized to show convergent validity with both PSES subscales and the time and talk subscale of the IFI.

Results

Table 1 provides the sociodemographic characteristics of the sample. The mean age of the sample was 33.6 with resident fathers (x̄= 34.4) being a few years older than non-resident fathers (x̄= 31.8). Cronbach’s α for the scale was 0.953 and the mean item-total correlation was 0.53 (range 0.40 – 0.63). Items were primarily skewed towards greater involvement in fatherhood (i.e. primarily endorsed “quite a bit” or “very much” options). Three items from the 47 were initially removed from scale production for low α values or low item-total correlations.

The exploratory bifactor analysis (Table 2) identified a general factor and four other factors whose items followed the four hypothesized subscales (i.e. acute illness, general well-being, emotional health, and role model). A total of 20 items had standardized loadings ≥ 0.45 and were kept for use in the confirmatory bifactor analysis. Table 3 summarizes the results from the confirmatory bifactor analysis, which confirmed the hypothetical structure of a general factor of fatherhood involvement in health and its four subscales. Of the 20 items used in the confirmatory analysis, 11 had a factor loading ≥ 0.500, and 5 had a factor loading ≥ 0.600. All of the fit indices indicated a good fit (CFI = 0.968, TLI = 0.959, WRMR = 0.954) except the RMSEA, whose value was 0.07. Figure 1 provides a graphical illustration of the results of the confirmatory bifactor analysis.

Table 2.

Exploratory Bifactor Analysis of Father Involvement in Health-Pre-school FIH-PS items

Items	General Factor	Factor 1	Factor 2	Factor 3	Factor 4
I feed my child healthy food.	0.400	−0.027	−0.072	0.435	0.373
I actively play with my child.	0.472*	0.051	0.123	0.153	0.289
I wash my child’s hands.	0.467*	0.052	0.109	0.077	0.364
I dress my child appropriately based on the weather.	0.478*	0.020	0.102	−0.071	0.548*
I give my child healthy fluids to drink.	0.451*	−0.005	0.034	0.134	0.459*
I put my child to bed at a decent hour.	0.464*	0.049	−0.058	0.177	0.463*
I ensure that my child has health insurance either from myself or his/her mother.	0.361	0.196	−0.064	−0.014	0.367
I keep track of my child’s immunizations.	0.495*	0.453*	−0.060	0.078	0.139
I make appointments for my child to see his/her doctor for the checkups.	0.485*	0.611*	−0.099	0.088	−0.031
I maintain a list of health provider names and numbers for my child’s health.	0.505*	0.519*	−0.055	0.136	0.012
I am accessible to address my child’s health needs.	0.499*	0.416	0.003	−0.029	0.178
I understand how my child is developing.	0.592*	0.197	0.241	0.044	0.169
I am aware of changes to my child’s health.	0.578*	0.224	0.219	0.034	0.155
I am available to observe my child’s general health.	0.557*	0.358	0.157	−0.051	0.113
I check my child’s temperature when he/she is sick.	0.596*	0.366	0.201	−0.075	0.109
I take care of my child when he/she is sick	0.591*	0.426	0.208	−0.113	0.041
I take my child to the doctor when he/she is sick.	0.513*	0.548*	0.029	−0.025	−0.021
I give medication(s) to my child when he/she is sick.	0.468*	0.458*	0.094	−0.046	−0.050
I follow medical advice as recommended when my child is sick.	0.555*	0.260	0.150	−0.004	0.215
I make sure someone (including myself) is available for childcare when my child is sick.	0.456*	0.156	0.175	−0.091	0.241
I monitor my child’s health when he/she is sick.	0.612*	0.326	0.279	−0.138	0.113
I select where my child goes for healthcare when he/she is sick.	0.499*	0.597*	−0.051	0.074	−0.057
I make sure medication (e.g., cold medication) and medical supplies (e.g., thermometer, bandaids) are in the home.	0.575*	0.293	0.164	−0.011	0.182
I recognize when my child becomes sick.	0.610*	0.160	0.336	0.009	0.124
I am accessible to take my child to the doctor when he/she is sick.	0.500*	0.436	−0.013	0.148	0.040
I am accessible to give my child medicine when he/she is sick.	0.586*	0.347	0.144	−0.017	0.162
I adjust my schedule when my child is sick.	0.455*	0.290	0.122	0.101	−0.007
I comfort my child physically.	0.483*	−0.045	0.455*	−0.031	0.058
I comfort my child verbally.	0.492*	−0.061	0.495*	−0.047	0.040
I explain the reasons behind health decisions to my child.	0.484*	0.145	0.323	0.152	−0.129
I ask my child how he/she is feeling emotionally.	0.539*	0.066	0.453*	0.078	−0.092
I monitor my child’s mood and emotional behaviors.	0.596*	0.026	0.451*	0.040	0.073
I ask other caregivers (e.g., child’s mother, teachers) about my child’s mood and emotional behaviors.	0.506*	0.204	0.270	0.120	−0.066
I recognize my child’s emotional changes.	0.576*	−0.011	0.440	0.147	0.028
I understand my child’s emotional needs.	0.580*	−0.021	0.465*	0.172	−0.013
I am available to provide emotional comfort to my child.	0.596*	−0.011	0.504*	0.052	0.030
I model healthy exercise for my child.	0.468*	−0.025	0.089	0.655*	−0.017
I model healthy eating for my child.	0.467*	−0.010	0.034	0.656*	0.105
I talk to my child about the importance of exercise.	0.471*	0.126	0.073	0.596*	−0.083
I talk to my child about the importance of healthy eating.	0.509*	0.115	0.156	0.477*	−0.050
I model a healthy lifestyle for my child.	0.464*	0.046	−0.001	0.729*	0.024
I model appropriate safety behaviors for my child (e.g. looking both ways before crossing the street, wearing a seatbelt).	0.514*	0.019	0.298	0.043	0.204
I recognize opportunities to be a role model for my child’s health.	0.574*	−0.018	0.335	0.194	0.162
I am available to be a role model to my child.	0.530*	0.100	0.258	0.191	0.065
I am physically present in my child’s life.	0.432	0.260	0.133	0.095	−0.012

Note:

^*p<.0

Table 3.

Confirmatory Bifactor Analysis of the Father Involvement in Health-Pre-school (FIH-PS) Measure

	General Factor	General Well-Being	Acute Illness	Emotional Health	Role Model
I dress my child appropriately based on the weather. (q1)	0.451	0.715
I give my child healthy fluids to drink. (q2)	0.516	0.504
I put my child to bed at a decent hour. (q3)	0.457	0.462
I keep track of my child’s immunizations. (q4)	0.578		0.524
I make appointments for my child to see his/her doctor for the checkups. (q5)	0.56		0.754
I maintain a list of health provider names and numbers for my child’s health. (q6)	0.606		0.497
I take my child to the doctor when he/she is sick. (q7)	0.529		0.519
I give medication(s) to my child when he/she is sick. (q8)	0.534		0.468
I select where my child goes for healthcare when he/she is sick. (q9)	0.558		0.498
I comfort my child physically. (q10)	0.468			0.801
I comfort my child verbally. (q11)	0.492			0.886
I ask my child how he/she is feeling emotionally. (q12)	0.655			0.481
I monitor my child’s mood and emotional behaviors. (q13)	0.634			0.451
I understand my child’s emotional needs. (q14)	0.69			0.477
I am available to provide emotional comfort to my child. (q15)	0.652			0.464
I model healthy exercise for my child. (q16)	0.703				0.473
I model healthy eating for my child. (q17)	0.665				0.558
I talk to my child about the importance of exercise. (q18)	0.655				0.537
I talk to my child about the importance of healthy eating. (q19)	0.642				0.592
I model a healthy lifestyle for my child. (q20)	0.639				0.634
CFI	0.967
TLI	0.958
RMSEA (90% CI)	0.07 (0.061, 0.080)
WRMR	0.954

Figure 1. Confirmatory factor analysis results.

CFI = Comparative Fit Index, TLI = Tucker-Lewis Index, RMSEA = Root Mean Square Error of Approximation and WRMR = Weighted Root Mean Square Residual

Results from factor analysis revealed a four-factor solution performed best when compared to a three-factor and a five-factor solution. The overall fit for the three-factor solution was poorer (CFI = 0.949, TLI = 0.937, WRMR = 1.097, RMSEA = 0.08) with a few items loading across multiple factors. Likewise the five-factor solution had poorer overall fit than the four-factor solution (CFI = 0.932, TLI = 0.921, WRMR = 1.102, RMSEA = 0.11). AIC with the four-factor bifactor model was compared to the correlated factors measurement model. The bifactor model had a lower AIC (6361.81) than did the correlated factors model (AIC = 6821.90), providing evidence of better fitness than a correlated factors model.

Monotonicity was achieved with the whole scale being weakly monotonic (H = 0.378), the acute illness subscale being moderately monotonic (H = 0.472), the general well-being scale being weakly monotonic (H = 0.360), the emotional health subscale being strongly monotonic (H = 0.521), and the role model subscale being moderately monotonic (H = 0.454).

Table 4 provides correlations between subscales from the PSES and IFI with the FIH-PS subscales that assess convergent and discriminate validity. Diagonal elements in the table represent Cronbach’s alpha levels for the items within each subscale. Non-diagonal elements represent the correlation between the subscales. Using Evans’ classification method, general well-being showed indeterminate concurrence with both PSES subscales but exhibited evidence of discriminant validity with four of the IFI subscales. Acute illness showed some evidence of discriminant validity with PSES and IFI subscales, but most concurrence between the subscales was indeterminate. The emotional health subscale of FIH-PS had evidence of convergent validity with both subscales of the PSES (accepting as an individual and positive evaluation of child) and with the time and talk subscale of the IFI. Some weak evidence of convergent validity was seen for emotional health of FIH-PS and attentiveness subscale of IFI (r = 0.398). Role modeling exhibited some evidence with the PSES subscale of accepting as individual. Role modeling showed evidence of discriminant validity with IFI providing subscale and praise and affection subscale.

Table 4.

Convergent and Discriminant Validity Matrix between Parenting Self-Efficacy Scale (PSES), Inventory of Father Involvement (IFI) and Father Involvement in Health – Preschool (FIH-PS) Measures¹

		PSES		IFI						FIH-PS
		Accepting as Individual	Positive Evaluation	Discipline	Providing	Time and Talk	Praise and Affection	Developing Talent	Attentiveness	General Well-being	Acute Illness	Emotional Health	Role Modeling
PSES	Accepting as Individual	0.852
	Positive Evaluation	0.650	0.758
IFI	Discipline	0.500	0.435	0.709
	Providing	0.256	0.343	0.217	0.697
	Time and Talk	0.565	0.519	0.476	0.334	0.736
	Praise and Affection	0.229	0.349	0.152	0.317	0.294	0.779
	Developing Talent	0.447	0.393	0.449	0.300	0.527	0.279	0.531
	Attentiveness	0.432	0.462	0.386	0.370	0.485	0.326	0.518	0.721
FIH-PS	General Well-being	0.327	0.329	0.199	0.157	0.173	0.154	0.209	0.250	0.596
	Acute Illness	0.202	0.189	0.199	0.225	0.255	0.207	0.252	0.350	0.354	0.853
	Emotional Health	0.476	0.484	0.287	0.209	0.454	0.247	0.323	0.398	0.394	0.412	0.854
	Role Modeling	0.405	0.239	0.355	0.044	0.360	0.114	0.325	0.244	0.349	0.422	0.471	0.886

Note:

¹= all p-values are significant at the 0.05 level

The estimated IRT parameters for the fatherhood involvement in health scale are provided in Table 5. Slopes for the general factor are provided in the first column and slopes for the specific subscale are given in the second column. The smallest general factor slope was 0.820 (I put my child to bed at a decent hour) and the largest was 3.171 (I make appointments for my child to see his/her doctor for checkups). The thresholds for the items were primarily negative, demonstrating that the FIH-PS scale discriminates well among lower levels of fatherhood involvement.

Table 5.

Item Response Theory Parameters for the Father Involvement in Health-Pre-school (FHI-PS) Measure

Item	Slope (General)	Slope (Subscale)	Threshold 1	Threshold 2	Threshold 3	Threshold 4
I dress my child appropriately based on the weather. (q1)	1.404	1.591	−5.069	−4.469	−3.385	−1.188
I give my child healthy fluids to drink. (q2)	1.299	1.278	−6.219	−4.653	−2.627	−0.084
I put my child to bed at a decent hour. (q3)	0.82	0.696	−7.598	−5.191	−2.428	0.282
I keep track of my child’s immunizations. (q4)	1.521	1.839	−3.944	−3.268	−1.860	−0.431
I make appointments for my child to see his/her doctor for the checkups. (q5)	3.171	6.05	−3.364	−2.481	−1.477	−0.091
I maintain a list of health provider names and numbers for my child’s health. (q6)	1.318	1.675	−3.259	−2.522	−1.094	0.172
I take my child to the doctor when he/she is sick. (q7)	1.302	1.164	−4.280	−2.869	−1.343	0.270
I give medication(s) to my child when he/she is sick. (q8)	1.288	0.858	−4.743	−3.058	−1.564	0.095
I select where my child goes for healthcare when he/she is sick. (q9)	1.408	1.62	−3.641	−2.979	−1.688	−0.378
I comfort my child physically. (q10)	2.47	2.981	−3.856	−2.594	−0.708
I comfort my child verbally. (q11)	1.106	1.61	−3.926	−2.567	−0.656
I ask my child how he/she is feeling emotionally. (q12)	1.868	0.352	−3.466	−2.369	−1.041	0.210
I monitor my child’s mood and emotional behaviors. (q13)	2.421	0.765	−3.403	−2.800	−1.533	−0.082
I understand my child’s emotional needs. (q14)	2.634	0.423	−3.546	−2.568	−1.324	0.073
I am available to provide emotional comfort to my child. (q15)	2.439	0.935	−3.765	−2.895	−1.712	−0.335
I model healthy exercise for my child. (q16)	1.861	1.972	−3.482	−2.157	−0.587	0.781
I model healthy eating for my child. (q17)	2.151	2.475	−3.479	−2.334	−0.777	1.007
I talk to my child about the importance of exercise. (q18)	1.753	1.656	−2.602	−1.600	−0.302	1.167
I talk to my child about the importance of healthy eating. (q19)	1.904	1.622	−3.418	−2.257	−0.861	0.687
I model a healthy lifestyle for my child. (q20)	2.189	3.105	−3.981	−2.540	−0.707	1.154

IRT pattern scoring produced a raw to T-score conversion table (Table 6) from the parameters shown in Table 5. A T-score is calculated where T = 50 represents the mean of the population and one standard deviation is equal to 10. As can be seen from the range, this scale measures from approximately 4 standard deviations below the mean to 2 standard deviations above it (T = 10.8 – 71.3). Items that perform better at discriminating the fatherhood involvement in health at various trait levels will have larger slopes than those that discriminate less well. For example, a slope of 2 would indicate that the item discriminates within a trait level better than an item with a slope of 0.5. An item with better discrimination indicates an item that better predicts changes in trait levels given an item’s response endorsement. The thresholds represent the average trait level of an item’s response endorsement. For example, a theta of −4 for threshold 1 indicates that the average fatherhood involvement level for those who endorsed the lowest response for that item was −4 (i.e. 4 standard deviations below average fatherhood involvement).

Table 6.

Raw to T-Score Crosswalk for the Father Involvement in Health Pre-school (FIH-PS) Measure

Raw Score	T-Score	SE	Raw Score	T-Score	SE	Raw Score	T-Score	SE
20	10.8	0.8	58	29.6	2.5	96	65	4.1
21	10.9	0.8	59	30.3	2.5	97	67.6	4.5
22	10.9	0.9	60	31	2.5	98	71.3	5.2
23	11	0.9	61	31.6	2.6
24	11.1	1	62	32.3	2.6
25	11.2	1.1	63	33.1	2.6
26	11.4	1.2	64	33.8	2.6
27	11.5	1.3	65	34.5	2.6
28	11.7	1.4	66	35.2	2.6
29	12	1.5	67	35.9	2.6
30	12.2	1.7	68	36.6	2.6
31	12.6	1.8	69	37.4	2.6
32	13	1.9	70	38.1	2.6
33	13.4	2.1	71	38.9	2.6
34	13.9	2.2	72	39.6	2.6
35	14.5	2.3	73	40.4	2.6
36	15.1	2.4	74	41.1	2.6
37	15.7	2.4	75	41.9	2.6
38	16.4	2.4	76	42.7	2.6
39	17	2.4	77	43.4	2.6
40	17.7	2.4	78	44.2	2.6
41	18.4	2.4	79	45	2.6
42	19	2.4	80	45.8	2.6
43	19.7	2.4	81	46.7	2.6
44	20.3	2.4	82	47.5	2.6
45	21	2.4	83	48.4	2.6
46	21.6	2.4	84	49.3	2.7
47	22.3	2.4	85	50.2	2.7
48	22.9	2.4	86	51.1	2.7
49	23.6	2.4	87	52.1	2.8
50	24.2	2.5	88	53.2	2.9
51	24.9	2.5	89	54.2	2.9
52	25.6	2.5	90	55.4	3
53	26.2	2.5	91	56.6	3.1
54	26.9	2.5	92	57.9	3.2
55	27.6	2.5	93	59.4	3.4
56	28.2	2.5	94	61	3.6
57	28.9	2.5	95	62.8	3.8

Discussion

The purpose of this study was to develop an instrument that measures father involvement in the health and healthcare of their preschool-aged child using the current, multidimensional, contextual conceptualization of father involvement. The FIH-PS assesses multiple components of the construct, reflects prevailing theory and research, and accounts for context (residency status) and shifting gender roles among fathers in the area of child health and healthcare. This study contributes to the literature by providing practitioners and researchers with a brief self-report measure of father involvement in health that can feasibly be used in research and clinical settings and administered easily via the Internet or a computer.

The FIH-PS measure with its four factors – Acute Illness, General Well-being, Emotional Health, and Role Modeling – can be used as general measure of overall father involvement in health or via the specific subsets. Many of the 47 items first proposed during the qualitative development of the instrument provided insight into the nature of fatherhood involvement in health for pre-schoolers. However, the final selection of 20 items provided sufficient model fitness for evaluating the hypothesized domains—general well-being, acute illness, emotional health, and role modeling. Inclusion of more of the originally developed items would likely not have contributed significantly more information to the research or measure and may have increased respondent test fatigue. The general factor in the bifactor model envelopes fatherhood involvement typically expressed in daily parenting: spending time with one’s children, feeding one’s children, tending to activities that impact health and wellbeing beyond the historic measurement of attendance at doctor visits. The four factors were shown through confirmatory factor analysis to have variance meaningfully explained by these four distinct constructs. This model of father involvement is consistent with current multidimensional conceptualizations of fathering (Garfield & Isacco, 2012; Pleck, 2012) and similar constructs (e.g., paternal self-efficacy) (Sevigny, Loutzenhiser, & McAuslan, 2016), thereby providing an updated and useful measure. The FIH-PS is a more comprehensive self-report measure of how contemporary fathers are involved with their child compared to previous measures that accounted narrowly for financial contributions, residential status, or time spent with their child (Harrington et al., 2010).

The four factors of the FIH-PS provide specific markers of the multidimensional nature of father involvement. The Acute Illness and General Well-being factors, for instance, are distinct to the FIH-PS and unique to the existing literature on father involvement focused on health and healthcare. For example, when considering families that often consist of dual-earner couples, it is plausible that when an acute, unexpected illness in a pre-school aged child arises, fathers may be as likely as mothers to adjust their routines and responsibilities of caring for the child. This adjustment may provide a benefit in both additional time spent between father and child as well as reduced stress across the family unit as two parents are capable of confidently caring for the sick child. The FIH-PS could capture these changes and examine associations with child and family health outcomes. Such a perspective using a tested measure may have been missing as the few previous studies examining father healthcare involvement have been qualitative in design, small in participants, and conducted to identify themes for theoretical conceptualization purposes (Garfield & Isacco, 2006; 2012).

Another unique contribution from the development of the FIH-PS measure was the consideration and inclusion of resident and non-resident fathers throughout the process. Over 40% of births today are to unmarried couples with a variety of cohabiting situations (Martin et al., 2017). Fathers’ residential status remains a key contextual variable that impacts the level of child involvement. The means for the sample were similar between resident and non-resident fathers, suggesting that the FIH-PS may have a sufficient level of validity to act as a measure for both groups of fathers. While previous research has found differences in involvement and decision-making in their child’s healthcare based on father residency (Isacco & Garfield, 2010; Jones & Mosher, 2013), future research can now use the FIH-PS to explore these differences between non-resident and resident fathers within the healthcare domain.

The range of IRT thresholds allows this scale to examine levels of fatherhood involvement up to two standard deviations above the mean and four standard deviations below the mean. This measurement is particularly useful when examining lower levels of fatherhood involvement in health, with specificity of the degree of low involvement (mildly, moderately, and severely low) able to be distinguished and quantified. All of the first thresholds have theta levels smaller than −3 (or 3 standard deviations below the mean) with the exception of “I talk to my child about the importance of exercise” whose theta = −2.602. All of the largest thresholds are less than 1.20 (or 1.2 standard deviations above the mean). Thus, the FIH-PS measurement primarily measures fatherhood involvement at its lowest levels. This result indicates that the scale’s most efficient use is determining fathers who have low involvement in their children’s health, which might be associated with less optimal outcomes. In turn, the ability to measure change in involvement among these low involvement fathers (as the result of some program or intervention) may prove especially useful for the field.

This study does have limitations. The FIH-PS has been scaled to a T-score (range 10.5 to 70) to facilitate generalizability to a population of English-speaking fathers in the U.S. with 3-to-5 year old children. Thus, the current scale is limited to community-based father involvement in preschool children ages 3–5. Fathers were also recruited for phase 2 of this study (FIH-PS testing) online and completed their involvement entirely online, thereby limiting the generalizability to fathers with Internet access. Performance of the FIH-PS in other populations (e.g., unique clinical settings, child with chronic illness, adolescent fathers, immigrant fathers) or with children in other age groups is unknown. As research grows in understanding the unique and diverse contributions of fathers to families, reliable measures are necessary to accurately measure and distinguish father involvement. For instance, father involvement with children who have specific childhood chronic illnesses like cystic fibrosis, cancer or diabetes mellitus would require attention to the unique needs of those children and families, dealing with multiple medications, frequent healthcare encounters, and much closer monitoring of the child to optimize health outcomes. Development of involvement measures in other age groups will need to test current items, alter items or generate new items that are developmentally and age-appropriate. Likewise, future longitudinal assessment would help to identify how fathers shift their health involvement to meet the health needs of their child at different developmental stages to meet the psychological and physical growth of their child. While this measure discriminates well at lower levels of father involvement, it does not discriminate very well at higher levels of fatherhood involvement, making distinction within highly involved fathers more difficult to discern. Further research can be conducted to better develop this measure, in particular, using structural equation model (SEM) within a latent model framework, which might help in examining the evidence for convergent and discriminant validity between the FIH-PS and other similar scales.

Practical Implications and Conclusions

As a tool developed to measure fatherhood involvement in health, the FIH-PS has several practical implications. From a fathering research and family systems approach, just as higher levels of father involvement have been associated with positive child outcomes, lower levels of father involvement have had an adverse effect on children (Garfield & Isacco, 2012). Since the FIH-PS performed well at distinguishing lower levels of father involvement (FIH-PS score of 10.5–40 or 1–4 SDs below the mean), interventions designed for fathers and families may benefit from using the instrument to identify areas for which fathers feel they need support and offering interventions to those fathers who are less involved. From an interventions and evaluations perspective, programs designed to engage and support fathers in their roles in families are more often required to point to specific metrics for success. Using the FIH-PS as either a simple pre-post tool or a longitudinal measure collected several times over a length of time can identify changes attributable to the program or intervention. Finally, for practitioners who engage with fathers, the FIH-PS can be used to determine which health domains fathers are currently involved in and which fathers may need more support, education, or guidance in order to increase their involvement.

The FIH-PS is a novel and promising tool to assess how fathers are involved in their child’s healthcare, which reflects current conceptualizations of the construct and the lived experience of today’s residential and non-residential fathers. This study fills important gaps in the existing literature, providing a measure of father involvement for future use across research and clinical settings.