Preventive Medicine 123 (2019) 84–90

Contents lists available at ScienceDirect

Preventive Medicine
journal homepage: www.elsevier.com/locate/ypmed

Does optimal parenting style help offspring maintain healthy weight into T
mid-life?
Ying Chena,b, , Ichiro Kawachic, Lisa F. Berkmanb,c,d,e, Claudia Trudel-Fitzgeraldc,f,
⁎

Laura D. Kubzanskyc,f
a
Human Flourishing Program, Institute for Quantitative Social Science, Harvard University, United States of America
b
Department of Epidemiology, Harvard T.H. Chan School of Public Health, United States of America
c
Department of Social and Behavioral Sciences, Harvard T.H. Chan School of Public Health, United States of America
d
Department of Global Health and Population, Harvard T.H. Chan School of Public Health, United States of America
e
Center for Population and Development Studies, Harvard University, United States of America
f
Lee Kum Sheung Center for Health and Happiness, Harvard T.H. Chan School of Public Health, United States of America

ARTICLE INFO ABSTRACT

Keywords: An authoritative parenting style is generally associated with healthier body weight in children and adolescents.
Parenting styles However, whether the protective effect of an authoritative style on offspring body weight may persist into
Parental warmth adulthood has seldom been investigated. In this study we examined the longitudinal association between par-
Parental control enting style and body mass index (BMI) change in mid-life. Longitudinal data from the Midlife in the United
Body weight
States Study (N = 3929) were analyzed using generalized estimating equations, adjusting for a range of relevant
Lifecourse
covariates. Parenting styles were assessed at phase I (1995–1996) using items measuring parental warmth and
control, while BMI was assessed at phases I and II (2004–2006). Four parenting styles were derived following
prior research: authoritative, authoritarian, permissive, and uninvolved styles. Compared to an authoritative
style, an authoritarian style was associated with 14% higher increase in the standardized BMI change score
(β = 0.14, 95% confidence interval: 0.03, 0.26). While there was suggestive evidence that an uninvolved versus
authoritative style might also be associated with greater BMI increase, we found no differences between a
permissive and authoritative style. This study suggested that the protective effect of an authoritative parenting
style on offspring body weight may persist well into mid-life, particularly as compared to the authoritarian style
and possibly the uninvolved style. Such work may reinforce the importance of a public health focus on im-
proving parenting practices and suggest the value of implementing parenting programs, as one strategy for
increasing the likelihood that individuals can maintain healthy weight well into adulthood.

1. Introduction on levels of parental warmth and parental control, prior researchers

(Maccoby and Martin, 1983) identified four general parenting styles: the
Obesity has been identified as a public health crisis worldwide (NCD authoritative (high in both warmth and control), authoritarian (low in
Risk Factor Collaboration, 2017). In addition to targeting individual- warmth and high in control), permissive (high in warmth and low in
level risk factors for obesity such as unhealthy diet and sedentary control) and uninvolved style (low in both warmth and control). Em-
lifestyle, the Institute of Medicine has called for a population-based pirical evidence generally suggests that the authoritative style is asso-
integrative approach for obesity prevention that also considers mod- ciated with better offspring health and well-being compared to other
ifiable factors in the broader social environment (Committee on parenting styles (Pinquart, 2016; Pinquart, 2017a; Pinquart, 2017b).
Accelerating Progress in Obesity Prevention, 2012). Recent research, though mostly cross-sectional, has begun to con-
Family represents individuals' immediate social surroundings and is sider parenting styles in relation to childhood obesity. A number of
potentially a critical source of social support that influences health reviews have suggested that an authoritative style is associated with
(Alvarez et al., 2016). In particular, parenting practices may shape off- healthier body weight and weight-related behaviors in children and
spring health and well-being over the lifecourse (Britto et al., 2017). One adolescents, as compared to other parenting styles (Sleddens et al.,
of the most widely-studied parenting practices is parenting style. Based 2011; Sokol et al., 2017; Vollmer and Mobley, 2013). In comparison,

⁎
Corresponding author at: 129 Mt Auburn Street, Cambridge, MA 02138, United States of America.
E-mail address: [email protected] (Y. Chen).

https://doi.org/10.1016/j.ypmed.2019.03.001
Received 10 September 2018; Received in revised form 25 February 2019; Accepted 2 March 2019
Available online 04 March 2019
0091-7435/ © 2019 Published by Elsevier Inc.
Y. Chen, et al. Preventive Medicine 123 (2019) 84–90

only a handful of studies have explored whether the protective effects Affection Scale (Rossi, 2001) (Table S1) was used to assess maternal
of an authoritative style on maintaining healthy body weight may and paternal warmth separately (e.g., “How much love and affection
persist beyond adolescence, but the limited evidence does suggest a did your mother/father give you?”). Response categories ranged from 1
lingering beneficial influence in young adulthood (Fuemmeler et al., (a lot) to 4 (not at all). When appropriate, items were reverse coded so
2012). However, to our knowledge whether the association may extend that a higher score reflected greater warmth. Maternal and paternal
into middle or late adulthood has never been examined. From a life- warmth scores were calculated for participants with valid data on at
course perspective, parental influences may shape offspring health not least half of the scale items, by averaging responses across items.
only in childhood but also well into adulthood (Holt-Lunstad et al., Following prior research, an overall parental warmth score was
2017). Children often model dietary and exercising behaviors from calculated by averaging the maternal and paternal warmth scores
their parents, and such behavioral patterns may persist into adulthood (Rothrauff et al., 2009). The scale had good internal consistency
and exert long-term health effects (Sokol et al., 2017). In fact, obese reliability in this sample (α = 0.89 for the maternal scale, α = 0.91
children are at substantially higher risk of staying obese when they for the paternal scale). It also showed positive associations with
grow up into adults (Simmonds et al., 2016). multiple domains of well-being in prior work (Chen et al., 2018;
Prior studies on parenting styles and offspring body weight often Moran et al., 2018), providing some evidence for construct validity.
considered a limited range of confounders, generally including factors
such as demographic characteristics and socioeconomic status [SES] 2.2.1.2. Parental control. Participants recalled parental control during
(Sokol et al., 2017). Other factors may also have important linkages to their years of growing up at phase I. A three-item Parental Control Scale
parenting styles and body weight but few studies have included them. (Rossi, 2001) (Table S1) was used to assess maternal and paternal
For example, childhood familial characteristics such as family structure control separately (e.g., “How strict was your mother/father with her/
(Sokol et al., 2017), parental abuse (Danese and Tan, 2014), residential his rules for you?”). Response options ranged from 1 (a lot) to 4 (not at
stability (Anderson et al., 2014; Jones, 2015) and family religiousness all). Responses were scored so that a higher score represented greater
(Bornstein et al., 2017; Goeke-Morey and Cummings, 2017) have been control. Maternal and paternal control scores were calculated for
linked with either parenting or children's risk of obesity. However, participants with valid data on at least half of the scale items, by
whether they may confound the association of parenting style with averaging responses across items. Following prior research, an overall
offspring body weight is understudied. parental control score was calculated by averaging the maternal and
This study examined the longitudinal association between parenting paternal control scores (Rothrauff et al., 2009). The scale showed
styles and offspring body weight over a 9-year follow-up in mid-life, acceptable internal consistency reliability in this sample (α = 0.74 for
controlling for a wide range of childhood family environment char- both the maternal and paternal scale). In prior work (Enns et al., 2002),
acteristics as potential confounders (e.g., family SES, family structure, greater paternal control measured with this scale was associated with
parental abuse, residential stability, family religiousness). Parenting lower risk of externalizing disorders in male children, which provided
styles were recalled when offspring were middle-aged. Because parenting some evidence for construct validity.
styles were retrospectively reported, as a sensitivity analysis we ad-
ditionally adjusted for a number of adulthood characteristics (e.g., 2.2.1.3. Parenting style. Four parenting style typologies were created
adulthood SES, depression, chronic health conditions) that may affect based on distinct constellations of parental warmth and control
how parenting styles were recalled (Widom et al., 2004). We hypothe- (Maccoby and Martin, 1983). As specific cut-points for these
sized that the authoritative parenting style would be associated with measures have not been validated, we followed common practice in
healthier body weight in mid-life, compared to other parenting styles. epidemiologic studies of psychosocial characteristics (Kubzansky et al.,
2014) and characterized individuals scoring in the top tertile of the
2. Methods distribution of scores on each subscale as being distinctively high in
warmth and in control (Fig. S1). The authoritative style included
2.1. Study sample participants scoring in the top tertile of both warmth and control
(14.11%); the authoritarian style included respondents scoring in the
Data were from the Midlife in the United States (MIDUS) study. bottom and middle tertiles of warmth but in the top tertile of control
MIDUS was initiated in 1994–1995 to study health and well-being in (18.07%); the permissive style included those scoring in the top tertile
mid-life. At the first wave (MIDUS I), 7108 non-institutionalized in- of warmth but in the bottom and middle tertiles of control (18.37%);
dividuals aged between 25 and 74 years across the United States were and the uninvolved style included those in the bottom and middle
enrolled through a random selection process. Participants were invited tertiles of both warmth and control (49.45%). To evaluate the
to participate in a phone interview, and then received a self-adminis- sensitivity of any associations to these particular cut-points, we also
tered questionnaire (SAQ). A second wave of the study (MIDUS II) took followed another categorization approach from prior literature
place in 2004–2005, which followed up 70% (N = 4963) of the original (Rothrauff et al., 2009) by using a median split to define high versus
participants. Details of the MIDUS recruitment and follow-up proce- low warmth or control (Fig. S2).
dures were reported elsewhere (Brim et al., 2004; Shaw et al., 2004).
Because parenting styles and body weight were only assessed in the 2.2.2. Dependent variable
SAQ, the analytic sample for the present study was drawn from re- 2.2.2.1. Body weight. At both phases participants reported their height
spondents who completed the SAQ at both waves (N = 3929; 569 of and weight, based on which body mass index (BMI, kg/m2) was
them either siblings or twins, and we adjusted for potential clustering calculated. The self-reported BMI showed high concordance
by sibling status in the analyses). This study was approved by the in- (r = 0.92) with BMI measured by clinicians or trained staffs in a
stitutional review board (IRB) at the authors' institution. The original subgroup who participated in a biomarker project at phase II
MIDUS study was approved by the IRB at participating institutions, and (N = 1255, the subgroup was comparable to the full sample in terms
all participants provided informed consent (Radler, 2014). of self-reported BMI and major demographic and health-related
characteristics) (Dienberg Love et al., 2010). We calculated a BMI
2.2. Measures change score (used as a continuous variable) by subtracting BMI at
phase I from BMI at phase II. To minimize influence of extreme outliers,
2.2.1. Independent variables the score was winsorized at the 1st and 99th percentile (i.e.,
2.2.1.1. Parental warmth. At phase I, participants recalled parental respondents with a score < the 1st percentile or > the 99th
warmth during their years of growing up. A six-item Parental percentile were assigned the value for the 1st and 99th percentile).

85
Y. Chen, et al. Preventive Medicine 123 (2019) 84–90

We standardized the BMI change score (mean = 0, standard reanalyzed the primary sets of models with parenting styles created
deviation = 1) and used the standardized score as the dependent using the median split to define high versus low levels of parental
variable in all models, to facilitate comparison of effect estimates warmth and control.
across studies in which BMI may follow a different distribution (Landis, Because results of the primary analyses suggested that the strongest
2005). differences were between parenting styles low in warmth (i.e., au-
thoritarian, uninvolved style) and high in warmth (i.e., authoritative
2.2.3. Covariates style), we performed a post-hoc analysis to explore whether effects of
2.2.3.1. Demographic factors. Demographic covariates included parenting styles were primarily driven by the dimension of parental
participant age (in years), sex (male, female), and race (white, black, warmth. Specifically, we reanalyzed the primary models first including
other races), all self-reported at phase I. parental warmth and then including parental control as the in-
dependent variable in separate models. We also conducted stratified
2.2.3.2. Childhood characteristics. All childhood characteristics were analyses to consider effects of parental warmth within low, moderate,
recalled at phase I. Childhood SES was assessed with the highest or high levels of parental control.
educational attainment of parents (less than high school, high school, In the full analytic sample (N = 3929), 10 participants were missing
some college, college degree or higher) (Miller et al., 2011). Severe data on parenting style, 301 were missing data on body weight at
parental abuse was assessed using one question from the Conflict baseline or the follow-up, and another 261 were missing data on cov-
Tactics Inventory (Straus, 1979): “During your childhood, how often ariates. Complete-case analysis would result in a loss of 14.6%
did your mother/father kicked, bit, or hit you with a fist or an object, (n = 572) of the participants. We performed a multivariate normal
beat you up, choked, burned or scalded you”. Response options ranging multiple imputation procedure (number of imputed datasets = 5) to
from 1 (often) to 4 (never), and responses were coded so a higher score impute missing data on all variables, as it often provides more accurate
represented greater abuse. An overall parental abuse score was created estimates compared to other methods of handling missing data (Sterne
by averaging the maternal and paternal abuse scores, and was used as a et al., 2009). We also performed complete-case analysis as a sensitivity
continuous variable (Savla et al., 2013). Two-parent family structure analysis.
was assessed with a single question: “Did you live with both biological
parents up until you were 16?” (yes, no). Childhood residential area 3. Results
was also queried (rural, small town, medium-sized town, suburbs, city,
moved around). Childhood residential stability was assessed with a Participants were predominantly white (93.48%) and slightly
single question: “How many times during your childhood did you move higher percentage female (55.48%), with the mean baseline age of
to a totally new neighborhood or town?” Participants reporting < 3 47.39 years (SD = 12.43). The mean BMI increase was 1.24 kg/m2
residential moves were considered as having residential stability (SD = 3.14) over an average of 9-years of follow-up. Participants gen-
(Bures, 2003). Family religiousness was also queried: “How important erally reported high levels of parental warmth (mean = 2.97, range: 1
was religion in your home when you were growing up?”, with responses to 4) and parental control (mean = 3.00, range: 1 to 4). Descriptive
ranging from 1 (not at all important) to 4 (very important). The analyses suggested participants with authoritative parents were more
measure was used as a continuous variable. likely to report a two-parent family structure, high family religiousness
and residential stability in childhood, and were more likely to be
2.2.3.3. Adulthood characteristics. Adulthood covariates were assessed married and have no depression in adulthood, compared to those raised
at phase I. Participants reported their current marital status (married, by authoritarian and permissive parents (Table 1).
divorced/separate, widowed, never married) and their own educational Compared to those raised by authoritative parents, participants with
attainment (less than high school, high school, some college, college authoritarian parents were 16% higher in the standardized score of BMI
degree or higher). Household income was also self-reported (in U.S. increase (β = 0.16, 95% confidence interval [CI]: 0.05, 0.28), adjusting
dollars, income greater than $300,000 was recoded as $300,000 to for age, sex and race (Table 2). Additionally adjusting for childhood SES
minimize risk of deductive disclosure), and quartiles of household (β = 0.16, 95% CI: 0.05, 0.28) and other childhood covariates
income were created. Major depression (yes, no) over past year was (β = 0.14, 95% CI: 0.03, 0.26) did not change the association. Notably,
assessed with the validated Composite International Diagnostic the effect size of authoritative versus authoritarian parenting style was
Interview Short Form (CIDI-SF) (Aalto-Setala et al., 2002; Kessler even larger than having the highest versus the lowest category of
et al., 1998). Participants who reported ever having or taking childhood family SES (i.e., parental education as college degree or
medication for any of the following conditions were considered as higher versus less than high school). Moreover, the association re-
having chronic physical health conditions (yes, no): cancer, heart mained robust in the sensitivity analysis that further considered
attack, diabetes, and stroke. adulthood characteristics. To a lesser extent, the uninvolved versus
authoritative style was also associated with a greater BMI increase, but
2.3. Statistical analyses the association was somewhat attenuated in the fully-adjusted model.
In comparison, there was no difference in the permissive versus au-
All statistical analyses were performed in SAS 9.4 (p < .05 sig- thoritative style in any model. Sensitivity analyses using the median
nificance level, two-tailed). Chi-square tests and analysis of variance split to define high versus low levels of parental warmth and control
tests were used to examine distribution of participant characteristics by yielded similar results (Table S2). Further, the complete-case analysis
parenting styles. also yielded somewhat attenuated but largely similar results (Table S3).
Generalized estimating equation models (GEE) were used to ex- When considering individual dimensions of parenting separately,
amine the association between parenting styles and BMI change, ad- parental warmth was associated with substantially less BMI increase
justing for clustering by sibling status. The base model adjusted for age, over time (Table 3), whereas parental control was not associated with
sex and race. A second model further adjusted for childhood SES. The BMI change (p > .05 in all models, results not shown). However, the
third model additionally accounted for a range of other childhood fa- stratified analyses suggested that effects of parental warmth were evi-
milial factors. We performed several sensitivity analyses. Because par- dent only for individuals scoring in the top tertile of parental control
enting styles were retrospectively reported in mid-life, to account for (Table 3). Results of this post-hoc analysis were consistent with our
the possibility that adulthood factors might influence how parenting primary analyses suggesting that the authoritative style was likely as-
styles were recalled, we additionally adjusted for adulthood char- sociated with healthier body weight, particularly as compared to the
acteristics concurrently assessed with parenting styles. We also authoritarian style.

86
Y. Chen, et al. Preventive Medicine 123 (2019) 84–90

Table 1
Participant characteristics by parenting styles in childhood (The Midlife in the United States Study 1995/1996–2004/2005 questionnaire wave, N = 3929).
Participant characteristics Full sample Parenting styles in childhood p-Value
N = 3929
Authoritative Authoritarian Permissive Uninvolved
N = 553 N = 708 N = 720 N = 1938

Baseline age, years, mean (SD) 47.39 (12.43) 47.44 (12.96) 46.88 (11.70) 48.38 (12.97) 46.93 (11.96) 0.04
Male, N (%) 1749 (44.52) 522 (47.07) 294 (38.43) 390 (45.94) 539 (45.07) 0.002
Race, N (%) < 0.001
White 3673 (93.48) 1015 (91.52) 708 (92.55) 802 (94.46) 1140 (95.32)
Black 148 (3.77) 68 (6.13) 30 (3.92) 24 (2.83) 24 (2.01)
Others 108 (2.75) 26 (2.34) 27 (3.53) 23 (2.71) 32 (2.68)
Childhood characteristics
Highest parental education, N (%) 0.06
Less than high school 985 (25.79) 283 (26.37) 220 (29.45) 172 (21.16) 302 (25.95)
High school 1377 (36.05) 379 (35.32) 260 (34.81) 310 (37.48) 427 (36.68)
Some college 301 (15.73) 177 (16.50) 107 (14.32) 138 (16.69) 177 (15.21)
College degree or higher 857 (22.43) 234 (21.81) 160 (21.42) 204 (24.67) 258 (22.16)
Parental abuse score, mean (SD) 1.25 (0.55) 1.13 (0.39) 1.51 (0.75) 1.08 (0.28) 1.30 (0.58) < 0.001
Lived with biological parents, N (%) 3174 (80.83) 969 (87.38) 605 (79.08) 694 (81.84) 897 (75.06) < 0.001
Childhood residential area, N (%) < 0.001
Rural 952 (24.55) 313 (28.58) 200 (26.42) 203 (24.08) 234 (19.91)
Small town 1003 (25.86) 291 (26.58) 185 (24.44) 209 (24.79) 315 (26.81)
Medium-sized town 458 (11.81) 123 (11.23) 75 (9.91) 97 (11.51) 163 (13.87)
Suburbs 606 (15.63) 158 (14.43) 116 (15.32) 127 (15.07) 204 (17.36)
City 697 (17.97) 173 (15.80) 135 (17.83) 169 (20.05) 218 (18.55)
Moved around 162 (4.18) 37 (3.38) 46 (6.08) 38 (4.51) 41 (3.49)
Residential stability, N (%) 2915 (74.84) 873 (79.15) 536 (70.71) 654 (77.58) 846 (71.51) < 0.001
Family religiousness, N (%) < 0.001
Religion not at all important 168 (4.28) 13 (1.17) 44 (5.75) 26 (3.07) 85 (7.12)
Religion not very important 603 (15.37) 86 (7.77) 122 (15.95) 112 (13.21) 283 (23.70)
Religion somewhat important 1380 (35.19) 327 (29.54) 266 (34.77) 317 (37.38) 467 (39.11)
Religion very important 1771 (45.16) 681 (61.52) 333 (43.53) 393 (46.34) 359 (30.07)
Adulthood characteristics
Marital status, N (%) < 0.001
Married 2814 (71.64) 854 (77.08) 532 (69.54) 597 (70.32) 826 (69.06)
Divorced/separated 537 (13.67) 111 (10.02) 124 (16.21) 103 (12.13) 197 (16.47)
Widowed 166 (4.23) 37 (3.34) 32 (4.18) 47 (5.54) 50 (4.18)
Never married 411 (10.46) 106 (9.57) 77 (10.07) 102 (12.01) 123 (10.28)
Educational attainment, N (%) 0.16
Less than high school 264 (6.72) 71 (6.40) 61 (7.97) 43 (5.06) 86 (7.19)
High school 1099 (27.97) 320 (28.85) 209 (27.32) 234 (27.56) 331 (27.68)
Some college 1145 (29.14) 323 (29.13) 224 (29.28) 231 (27.21) 366 (30.60)
College degree or higher 1421 (36.17) 395 (35.62) 271 (35.42) 341 (40.16) 413 (34.53)
Household income quartiles, N (%) 0.95
Bottom quartile ($0–$32,499) 931 (24.24) 254 (23.43) 180 (24.13) 192 (23.13) 298 (25.40)
Second quartile ($32,500–$58,999) 986 (25.67) 280 (25.83) 194 (26.01) 208 (25.06) 303 (25.83)
Third quartile ($59,000–$98,999) 965 (25.12) 269 (24.82) 191 (25.60) 213 (25.66) 292 (24.89)
Top quartile ($99,000–$300,000+) 959 (24.97) 281 (25.92) 181 (24.26) 217 (26.14) 280 (23.87)
Major depression, N (%) 469 (11.94) 89 (8.03) 127 (16.60) 71 (8.36) 181 (15.13) < 0.001
Any chronic health condition, N (%) 593 (15.09) 164 (14.79) 117 (15.29) 140 (16.49) 170 (14.21) 0.55

Note. Percentages refer to the proportion of individuals within each parenting style category with that characteristic. p comes from χ2 or analysis of variance tests.

4. Discussion help increase resilience and reduce unhealthy coping strategies (e.g.,
binge eating) under stressful situations (Holmes, 2014). In addition,
This is the first longitudinal study suggesting that the authoritative authoritative parents may also teach children healthy practices and set
parenting style may exert a protective effect on offspring body weight reasonable expectations on their behaviors. Some of these behavioral
well into mid-life, particularly as compared to the authoritarian style patterns including diet and exercise habits may persist into adulthood
and also possibly the uninvolved style. However, there was no differ- and affect one's weight trajectories (Kwon et al., 2015; Watts et al.,
ence between the authoritative and permissive style. The post-hoc 2018). However, effects may be tempered by influences occurring later
analyses on individual dimensions of parenting also indicated that in life. Thus, while parental influences set a developmental trajectory,
parental warmth may help offspring maintain healthy body weight, social relationships developed in later life may redirect the course. For
only when a high level of parental control was also present. instance, as individuals transition to adolescence and adulthood, peers
Findings of this study are consistent with prior work in younger and partners may reshape one's behavioral perceptions and patterns
populations which generally suggested that the authoritative parenting (e.g., smoking, diet), and offset early parental influences (Scalici and
style was associated with healthier body weight in children and ado- Schulz, 2014).
lescents, as compared to other parenting styles (Sokol et al., 2017). This Compared to the authoritative style, the elevated BMI increase as-
study expands prior literature by adding evidence that the protective sociated with the authoritarian style was particularly pronounced, and
effect of the authoritative style may persist into mid-life. Parental this finding may be attributable to the synergistic effects of low parental
warmth may provide children with a sense of emotional security and warmth and high parental control. Although the dimension of parental
self-worth. Such experiences could facilitate the formation of develop- control was not associated with BMI change in this middle-aged sample,
mental assets such as social integration and self-regulation, which may prior studies have separately linked low parental warmth and excessive

87
Y. Chen, et al. Preventive Medicine 123 (2019) 84–90

Table 2
Parenting styles in childhood and BMI change in mid-life (The Midlife in the United States Study 1995/1996–2004/2005 questionnaire wave, N = 3929).
Model 1 Model 2 Model 3 Model 4

β (95% CI) β (95% CI) β (95% CI) β (95% CI)

Parenting styles (Authoritative as ref)

Authoritarian style 0.16 (0.05, 0.28)⁎⁎ 0.16 (0.05, 0.28)⁎⁎ 0.14 (0.03, 0.26)⁎ 0.12 (0.01, 0.23)⁎
Permissive style 0.06 (−0.06, 0.17) 0.06 (−0.05, 0.18) 0.07 (−0.05, 0.18) 0.06 (−0.06, 0.17)
Uninvolved style 0.11 (0.01, 0.20)⁎ 0.11 (0.01, 0.20)⁎ 0.10 (−0.00, 0.20) 0.08 (−0.02, 0.18)
Age (standardized) −0.21 (−0.24, −0.18)⁎⁎⁎ −0.22 (−0.25, −0.19)⁎⁎⁎ −0.22 (−0.25, −0.19)⁎⁎⁎ −0.21 (−0.25, −0.17)⁎⁎⁎
Male (female as ref) −0.09 (−0.15, −0.02)⁎⁎ −0.09 (−0.15, −0.02)⁎⁎ −0.09 (−0.16, −0.02)⁎⁎ −0.07 (−0.13, −0.002)⁎
Race (White as ref)
Black 0.14 (−0.06, 0.33) 0.12 (−0.08, 0.32) 0.11 (−0.09, 0.31) −0.10 (−0.10, 0.30)
Others 0.08 (−0.15, 0.31) 0.09 (−0.14, 0.31) 0.08 (−0.15, 0.31) 0.05 (−0.17, 0.28)
Childhood characteristics
Parental education (< high school as ref)
High school 0.01 (−0.09, 0.10) 0.02 (−0.08, 0.12) 0.02 (−0.07, 0.12)
Some college −0.01 (−0.11, 0.09) 0.00 (−0.10, 0.11) 0.02 (−0.08, 0.13)
College degree or higher −0.14 (−0.24, −0.04)⁎⁎ −0.11 (−0.22, −0.01)⁎ −0.08 (−0.18, 0.03)
Parental abuse score (standardized) 0.04 (0.00, 0.07)⁎ 0.03 (−0.01, 0.06)
Lived with biological parents (no as ref) 0.02 (−0.06, 0.10) 0.04 (−0.05, 0.12)
Childhood residential area (rural as ref)
Small town −0.03 (−0.12, 0.06) −0.03 (−0.12, 0.06)
Medium-sized town −0.06 (−0.18, 0.05) −0.07 (−0.19, 0.05)
Suburbs −0.06 (−0.16, 0.05) −0.07 (−0.17, 0.04)
City −0.07 (−0.17, 0.03) −0.08 (−0.18, 0.03)
Moved around −0.15 (−0.35, 0.05) −0.15 (−0.35, 0.05)
Childhood residential stability (no as ref) −0.03 (−0.12, 0.05) −0.04 (−0.12, 0.05)
Family religiousness (not at all as ref)
Religion not very important −0.16 (−0.35, 0.03) −0.17 (−0.36, 0.02)
Religion somewhat important −0.12 (−0.29, 0.05) −0.12 (−0.29, 0.05)
Religion very important −0.12 (−0.28, 0.05) −0.11 (−0.27, 0.05)
Adulthood characteristics
Marital status (married as ref)
Divorced/separated 0.22 (0.12, 0.33)⁎⁎⁎
Widowed 0.17 (−0.01, 0.36)
Never married 0.13 (0.00, 0.25)⁎
Education attainment (< high school as ref)
High school 0.05 (−0.10, 0.20)
Some college −0.04 (−0.20, 0.12)
College degree or higher −0.08 (−0.12, −0.05)
Household income (bottom quartile as ref)
Second quartile 0.06 (−0.04, 0.15)
Third quartile 0.10 (0.00, 0.21)⁎
Top quartile 0.11 (0.01, 0.22)⁎
Major depression (no as ref) 0.16 (−0.28, −0.04)⁎⁎
Any chronic health condition (no as ref) −0.01 (−0.11, 0.09)

Note: Generalized estimating equations with normal distribution and identity link were used in all models to estimate the mean change in BMI (standardized score,
mean = 0, standard deviation = 1) by parenting styles, adjusting for clustering by sibling status.
⁎
p < .05.
⁎⁎
p < .01.
⁎⁎⁎
p < .001.

Table 3
Parental warmth in childhood and BMI change in mid-life, stratified by levels of parental control (The Midlife in the United States Study 1995/1996–2004/2005
questionnaire wave, N = 3929).
Full sample Stratified by parental control

Bottom tertile Middle tertile Top tertile

β (95% CI) β (95% CI) β (95% CI) β (95% CI)

Model 1 −0.05 (−0.08, −0.02)⁎⁎ −0.06 (−0.11, 0.00) −0.01 (−0.07, 0.04) −0.08 (−0.14, −0.02)⁎⁎
Model 2 −0.05 (−0.08, −0.01)⁎⁎ −0.05 (−0.11, 0.01) −0.01 (−0.07, 0.05) −0.08 (−0.14, −0.02)⁎
Model 3 −0.04 (−0.07, −0.003)⁎ −0.03 (−0.09, 0.04) −0.01 (−0.07, 0.05) −0.08 (−0.15, −0.01)⁎

Note: Generalized estimating equations with normal distribution and identity link were used in all models to estimate the mean change in BMI (standardized score)
by change in parental warmth (standardized score, mean = 0, standard deviation = 1), adjusting for clustering by sibling status.
Model 1 adjusted for participant age, sex and race.
Model 2 additionally adjusted for childhood socioeconomic status (assessed by parental education).
Model 3 additionally adjusted for other childhood family environment factors (including parental abuse, whether lived with both biological parents, childhood
residential area, childhood residential stability, and family religiousness).
⁎
p < .05.
⁎⁎
p < .01.

88
Y. Chen, et al. Preventive Medicine 123 (2019) 84–90

parental control with increased risk of childhood obesity (Gartstein agencies had no role in the data collection, analysis, or interpretation;
et al., 2018; Larsen et al., 2015; Rhee et al., 2016). Low parental nor were they involved in the writing or submission of this publication.
warmth may result in higher emotional distress and low self-efficacy, We thank the staff of the clinical research centers at Georgetown
while high parental control may lead to lack of capacity for reasoning University, University of Wisconsin-Madison, and University of
and self-regulation. Each of these sequelae could increase risk of California, Los Angeles for their support in conducting this study. CTF
adopting unhealthy weight-related behaviors to cope with distress received a postdoctoral fellowship from the Fonds de Recherche du
(Larsen et al., 2015; Topham et al., 2011). Somewhat unexpectedly, Québec-Santé.
evidence suggesting the uninvolved versus authoritative parental style
was associated with higher offspring BMI increase was modest in this Appendix A. Supplementary data
study. This might be due in part to participant characteristics specific to
this sample, who generally reported high levels of parental warmth and Supplementary data to this article can be found online at https://
control. The limited variation might attenuate our ability to detect ef- doi.org/10.1016/j.ypmed.2019.03.001.
fects of an uninvolved style, if any.
This study has several limitations. First, parenting style and other References
childhood characteristics were retrospectively reported in mid-life,
which may be subject to recall bias. However, there has been evidence Aalto-Setala, T., Haarasilta, L., Marttunen, M., Tuulio-Henriksson, A., Poikolainen, K.,
suggesting that concerns about retrospectively reported childhood ex- Aro, H., Lonnqvist, J., 2002. Major depressive episode among young adults: CIDI-SF
versus SCAN consensus diagnoses. Psychol. Med. 32, 1309–1314.
periences may not be as significant as previously thought (Patten et al., Alvarez, E.C., Kawachi, I., Romani, J.R., 2016. Family social capital and health – a sys-
2015). For instance, in prior work, childhood maltreatment experiences tematic review and redirection. Sociology of Health & Illness 39, 5–29.
as recalled in adulthood showed high concordance with official records Anderson, S., Leventhal, T., Dupéré, V., 2014. Residential mobility and the family con-
text: a developmental approach. J. Appl. Dev. Psychol. 35, 70–78.
such as data from Child Protective Services and medical records, sug- Bornstein, M.H., Putnick, D.L., Lansford, J.E., Al-Hassan, S.M., Bacchini, D., Bombi, A.S.,
gesting reasonable validity (Pinto et al., 2014; Shin et al., 2013). In Chang, L., Deater-Deckard, K., Di Giunta, L., et al., 2017. ‘Mixed blessings’: parental
addition, the longitudinal study design and sensitivity analyses ad- religiousness, parenting, and child adjustment in global perspective. J. Child Psychol.
Psychiatry 58, 880–892.
justing for adulthood characteristics provide reassurance that the ob-
Brim, O.G., Ryff, C.D., Kessler, R.C., 2004. How Healthy Are We?: A National Study of
served associations are not entirely due to bias. Second, although this Well-being at Midlife. University of Chicago Press, Chicago.
study adjusted for a wide range of childhood characteristics, residual Britto, P.R., Lye, S.J., Proulx, K., Yousafzai, A.K., Matthews, S.G., Vaivada, T., Perez-
Escamilla, R., Rao, N., Ip, P., et al., 2017. Nurturing care: promoting early childhood
confounding is possible by other factors for which information was not
development. Lancet 389, 91–102.
available such as childhood health status and parental mental health. Bures, R.M., 2003. Childhood residential stability and health at midlife. Am. J. Public
Due to the lack of temporal separation in the assessments of childhood Health 93, 1144–1148.
parenting style and adulthood health-related characteristics, we did not Chen, Y., Kubzansky, L.D., VanderWeele, T.J., 2018. Parental warmth and flourishing in
mid-life. Soc. Sci. Med. 220, 65–72.
perform a causal mediation analysis (VanderWeele, 2015), and as a Chu, J.T., Farruggia, S.P., Sanders, M.R., Ralph, A., 2012. Towards a public health ap-
result it remains unclear whether parenting styles may have a direct proach to parenting programmes for parents of adolescents. J Public Health (Oxf) 34
effect on offspring adulthood weight, or only indirectly through shaping (Suppl. 1), i41–i47.
Committee on Accelerating Progress in Obesity Prevention, Food and Nutrition Board,
children's later life experiences. Finally, MIDUS participants did not Institute of Medicine, 2012. in: Glickman, D., Parker, L., Sim, L.J., Del Valle Cook, H.,
comprise a nationally representative sample (e.g., predominantly Miller, E.A. (Eds.), Accelerating Progress in Obesity Prevention: Solving the Weight of
white), thus results of this study may not be generalizable to other the Nation. National Academies Press. DC, Washington.
Danese, A., Tan, M., 2014. Childhood maltreatment and obesity: systematic review and
populations. These limitations were, however, balanced by strengthens meta-analysis. Mol. Psychiatry 19, 544–554.
including the prospective study design, the large sample size with a 9- Dienberg Love, G., Seeman, T.E., Weinstein, M., Ryff, C.D., 2010. Bioindicators in the
year follow-up, and the rigorous confounding control in this study. MIDUS national study: protocol, measures, sample, and comparative context. Journal
of aging and health 22, 1059–1080.
The World Health Organization has called for parenting programs at Enns, M.W., Cox, B.J., Clara, I., 2002. Parental bonding and adult psychopathology: re-
the population level (World Health Organization, 2007, 2009), yet the sults from the US National Comorbidity Survey. Psychol. Med. 32, 997–1008.
progress on implementing large-scale parenting programs has been Fuemmeler, B.F., Yang, C., Costanzo, P., Hoyle, R.H., Siegler, I.C., Williams, R.B., Ostbye,
T., 2012. Parenting styles and body mass index trajectories from adolescence to
relatively slow (Chu et al., 2012). Such programs seek to improve
adulthood. Health Psychol. 31, 441–449.
general parenting styles (e.g., reducing coercive discipline practice) and Gartstein, M.A., Seamon, E., Thompson, S.F., Lengua, L.J., 2018. Parenting matters:
specific parenting practices (e.g., improving skills in communicating moderation of biological and community risk for obesity. J. Appl. Dev. Psychol. 56,
about health), and have been linked to multiple better health and well- 21–34.
Gerards, S.M., Dagnelie, P.C., Jansen, M.W., van der Goot, L.O., de Vries, N.K., Sanders,
being outcomes in children including reduced risk of obesity (Gerards M.R., Kremers, S.P., 2012. Lifestyle triple P: a parenting intervention for childhood
et al., 2012; Hubbs-Tait et al., 2016; Yap et al., 2016). This study fur- obesity. BMC Public Health 12, 267.
ther suggests that the beneficial effects of such programs, if any, on Goeke-Morey, M.C., Cummings, E.M., 2017. Religiosity and parenting: recent directions
in process-oriented research. Curr. Opin. Psychol. 15, 7–12.
children's body weight may persist into mid-life, and helps strengthen Holmes, J., 2014. John Bowlby and Attachment Theory, 2nd ed. Routledge, London, UK.
the case for implementing parenting programs at the population level. Holt-Lunstad, J., Robles, T.F., Sbarra, D.A., 2017. Advancing social connection as a public
Further research on parenting styles and the broader family environ- health priority in the United States. Am. Psychol. 72, 517.
Hubbs-Tait, L., Kimble, A., Hingle, M., Novotny, R., Fiese, B., 2016. Systematic review of
ment can help inform more targeted interventions, and identify re- child obesity prevention and treatment trials addressing parenting. FASEB J. 30,
sources within the family that may be targeted for improving obesity 1155.6–55.6.
prevention and control. Jones, A., 2015. Residential mobility and trajectories of adiposity among adolescents in
urban and non-urban neighborhoods. J. Urban Health 92, 265–278.
Kessler, R.C., Andrews, G., Mroczek, D., Ustun, B., Wittchen, H.-U., 1998. The World
Conflict of interest disclosures Health Organization Composite International Diagnostic Interview short-form (CIDI-
SF). Int. J. Methods Psychiatr. Res. 7, 171–185.
Kubzansky, L.D., Winning, A., Kawachi, I., 2014. Affective states and health. In: Berkman,
None.
L.F., Kawachi, I., Glymour, M.M. (Eds.), Social Epidemiology: New Perspectives on
Social Determinants of Global Population Health. Oxford University Press, New York,
Acknowledgements pp. 320–364.
Kwon, S., Janz, K.F., Letuchy, E.M., Burns, T.L., Levy, S.M., 2015. Active lifestyle in
childhood and adolescence prevents obesity development in young adulthood.
The original Midlife in the United States study was supported by the Obesity 23, 2462–2469.
John D. and Catherine T. MacArthur Foundation Research Network on Landis, R., 2005. Standardized regression coefficient. In: Everitt, B., Howell, D. (Eds.),
Successful Midlife Development. Follow-up data collection was sup- Encyclopedia of Statistics in Behavioral Science. John Wiley & Sons Ltd, Chichester,
pp. 1892.
ported by the National Institute on Aging (P01-AG020166). Funding

89
Y. Chen, et al. Preventive Medicine 123 (2019) 84–90

Larsen, J.K., Hermans, R.C., Sleddens, E.F., Engels, R.C., Fisher, J.O., Kremers, S.P., 2015. adolescent smoking intentions: parents have more say, but their influence wanes as
How parental dietary behavior and food parenting practices affect children's dietary kids get older. PLoS One 9, e101275.
behavior. Interacting sources of influence? Appetite 89, 246–257. Shaw, B.A., Krause, N., Chatters, L.M., Connell, C.M., Ingersoll-Dayton, B., 2004.
Maccoby, E., Martin, J., 1983. Socialization in the context of the family: parent–child Emotional support from parents early in life, aging, and health. Psychol. Aging 19,
interaction. In: Mussen, P., Hetherington, E. (Eds.), Handbook of Child Psychology, 4–12.
4th ed. Wiley, New York, pp. 1–101. Shin, S.H., Miller, D.P., Teicher, M.H., 2013. Exposure to childhood neglect and physical
Miller, G.E., Lachman, M.E., Chen, E., Gruenewald, T.L., Karlamangla, A.S., Seeman, T.E., abuse and developmental trajectories of heavy episodic drinking from early adoles-
2011. Pathways to resilience: maternal nurturance as a buffer against the effects of cence into young adulthood. Drug Alcohol Depend. 127, 31–38.
childhood poverty on metabolic syndrome at midlife. Psychol. Sci. 22, 1591–1599. Simmonds, M., Llewellyn, A., Owen, C.G., Woolacott, N., 2016. Predicting adult obesity
Moran, K.M., Turiano, N.A., Gentzler, A.L., 2018. Parental warmth during childhood from childhood obesity: a systematic review and meta-analysis. Obes. Rev. 17,
predicts coping and well-being in adulthood. J. Fam. Psychol. 32, 610–621. 95–107.
NCD Risk Factor Collaboration, 2017. Worldwide trends in body-mass index, under- Sleddens, E.F., Gerards, S.M., Thijs, C., de Vries, N.K., Kremers, S.P., 2011. General
weight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 po- parenting, childhood overweight and obesity-inducing behaviors: a review. Int. J.
pulation-based measurement studies in 128.9 million children, adolescents, and Pediatr. Obes. 6, e12–e27.
adults. Lancet 390, 2627–2642. Sokol, R.L., Qin, B., Poti, J.M., 2017. Parenting styles and body mass index: a systematic
Patten, S.B., Wilkes, T.C.R., Williams, J.V.A., Lavorato, D.H., el-Guebaly, N., review of prospective studies among children. Obes. Rev. 18, 281–292.
Schopflocher, D., Wild, C., Colman, I., Bulloch, A.G.M., 2015. Retrospective and Sterne, J.A., White, I.R., Carlin, J.B., Spratt, M., Royston, P., Kenward, M.G., Wood, A.M.,
prospectively assessed childhood adversity in association with major depression, al- Carpenter, J.R., 2009. Multiple imputation for missing data in epidemiological and
cohol consumption and painful conditions. Epidemiology and Psychiatric Sciences clinical research: potential and pitfalls. BMJ 338, b2393.
24, 158–165. Straus, M.A., 1979. Measuring intrafamily conflict and violence: the conflict tactics (CT)
Pinquart, M., 2016. Associations of parenting styles and dimensions with academic scales. J. Marriage Fam. 41, 75–88.
achievement in children and adolescents: a meta-analysis. Educ. Psychol. Rev. 28, Topham, G.L., Hubbs-Tait, L., Rutledge, J.M., Page, M.C., Kennedy, T.S., Shriver, L.H.,
475–493. Harrist, A.W., 2011. Parenting styles, parental response to child emotion, and family
Pinquart, M., 2017a. Associations of parenting dimensions and styles with externalizing emotional responsiveness are related to child emotional eating. Appetite 56,
problems of children and adolescents: An updated meta-analysis. Dev. Psychol. 53, 261–264.
873–932. VanderWeele, T., 2015. Explanation in Causal Inference: Methods for Mediation and
Pinquart, M., 2017b. Associations of parenting dimensions and styles with internalizing Interaction. Oxford University Press, New York, NY.
symptoms in children and adolescents: a meta-analysis. Marriage Fam. Rev. 53, Vollmer, R.L., Mobley, A.R., 2013. Parenting styles, feeding styles, and their influence on
613–640. child obesogenic behaviors and body weight. A review. Appetite 71, 232–241.
Pinto, R., Correia, L., Maia, Â., 2014. Assessing the reliability of retrospective reports of Watts, A., Berge, J.M., Loth, K., Larson, N., Neumark-Sztainer, D., 2018. The transmission
adverse childhood experiences among adolescents with documented childhood of family food and mealtime practices from adolescence to adulthood: longitudinal
maltreatment. J. Fam. Violence 29, 431–438. findings from project EAT-IV. J. Nutr. Educ. Behav. 50, 141–47 e1.
Radler, B.T., 2014. The midlife in the United States (MIDUS) series: a national long- Widom, C.S., Raphael, K.G., DuMont, K.A., 2004. The case for prospective longitudinal
itudinal study of health and well-being. Open Health Data 2. studies in child maltreatment research: commentary on Dube, Williamson,
Rhee, K.E., Jelalian, E., Boutelle, K., Dickstein, S., Seifer, R., Wing, R., 2016. Warm Thompson, Felitti, and Anda (2004). Child Abuse Negl. 28, 715–722.
parenting associated with decreasing or stable child BMI during treatment. Childhood World Health Organization, 2007. Helping Parents in Developing Countries Improve
Obesity 12, 94–102. adolescents' Health. World Health Organization, Geneva.
Rossi, A.S., 2001. Caring and Doing for Others: Social Responsibility in the Domains of World Health Organization, 2009. Preventing Violence Through the Development of Safe,
Family, Work, and Community. University of Chicago Press. Stable and Nurturing Relationships Between Children and Their Parents and
Rothrauff, T.C., Cooney, T.M., An, J.S., 2009. Remembered parenting styles and adjust- Caregivers. Series of Briefings on Violence Prevention: The Evidence. In: World
ment in middle and late adulthood. J. Gerontol. B Psychol. Sci. Soc. Sci. 64, 137–146. Health Organization. Geneva, Switzerland.
Savla, J.T., Roberto, K.A., Jaramillo-Sierra, A.L., Gambrel, L.E., Karimi, H., Butner, L.M., Yap, M.B.H., Morgan, A.J., Cairns, K., Jorm, A.F., Hetrick, S.E., Merry, S., 2016. Parents in
2013. Childhood abuse affects emotional closeness with family in mid- and later life. prevention: a meta-analysis of randomized controlled trials of parenting interventions
Child Abuse Negl. 37, 388–399. to prevent internalizing problems in children from birth to age 18. Clin. Psychol. Rev.
Scalici, F., Schulz, P.J., 2014. Influence of perceived parent and peer endorsement on 50, 138–158.

90