At age 1, fathers reported on their own involvement with their children via a series of 5 items. Fathers were asked how often during a typical week they participated in activities with their children such as “Play games like “peek-a-boo” or “gotcha” with [child],” “Put [child] to bed,” and “Hug or show physical affection to [child];” responses for all items ranged from 0–7 days per week. These five items had a Cronbach’s alpha in the present sample of .90, and a measurement model tested in SEM showed good fit to the data (? 2 = , p < .05; CFI = .99; RMSEA = .066), with all items loading onto the latent factor with values greater than .6.
Father-guy relationship high quality
Children responded to five items asking “Does your dad talk over important decisions with you?”, “Does your dad listen to your side of an argument?”, and “How close to you feel to your dad?” Responses could range from 0 (Never/Not very close) to 3 (Always/Extremely close). Responses were coded such that higher values represented better father-child relationships. Cronbach’s alpha for the current sample was .90, and a measurement model tested in SEM showed good fit to the data, ? 2 (5) = , p < .05; CFI = .99; RMSEA = .077, with all items loading onto the latent factor with values greater than .6.
Father citizen position
Dad resident position over time was used since the good moderator and you may are reviewed in two 
means (examined in the independent models). Basic, we determined a dummy coded varying you to definitely indicated if or not fathers was basically constantly nonresident and not had lived through its guy (0), otherwise reported living with their boy will ultimately within the kid’s earliest nine age (Birth, age step 1, years step three, decades 5, or many years nine). Second, we determined about three groups representing dads have been constantly resident in the the swells (2), have been citizen for at least one time area although not in history facts (1), or who had been constantly nonresident across the all the swells (3).
Analyses
Analyses was indeed presented having fun with architectural equation acting inside the AMOS 20 and you may lost study was in fact estimated thru FIML. Studies mostly was indeed lost in the event the a dad wasn’t utilized in a specific revolution of data collection. On Trend step 1 (birth), dos,402 dads was questioned (81%) and at man decades one year, 2,207 dads had been questioned (74%); of the 2,207 fathers interviewed from the ages 12 months, 151 (5%) had not been questioned in the delivery. Only child studies were used in the years 9, and you may household had been only included in the present research in the event that their pupils were questioned in the many years 9. In spite of the quite large amounts of missing research, i decided to include these members and you can estimate shed study as an alternative than just dump the inclusion throughout the analysis totally due to missingness at the among around three waves, such because of the highest-quality rates produced playing with FIML.
A design was estimated one to relevant fathers’ reference to brand new PGF during the childhood along with his reference to the latest PGF at the years 1, both PGF parameters which have fathers’ connections to their particular college students during the years step one, and all of three parameters that have children’s declaration regarding dad-child matchmaking top quality within decades 9, whenever you are dealing with getting father age, degree, race/ethnicity, and you may home income (find Profile step one ). One or two sets of multi-class analyses after that were conducted during the SEM to evaluate whether or not patterns anticipating father-boy matchmaking top quality was basically comparable for consistently citizen, consistently nonresident, and often citizen dads. Design fit is actually reviewed playing with chi-rectangular, CFI, and you can RMSEA opinions; improvement in design fit for multi-classification analyses was reviewed using change in chi-rectangular opinions.
