Introduction
Childhood mental health problems place a strain on both the child and their family (Farrell & Barrett, Reference Farrell and Barrett2007; Houtrow & Okumura, Reference Houtrow and Okumura2011). Parental factors such as socioeconomic status, parenting style, and parental mental health are important predictors of mental health outcomes in children (Bøe et al., Reference Bøe, Sivertsen, Heiervang, Goodman, Lundervold and Hysing2014; Leinonen, Solantaus, & Punamäki, Reference Leinonen, Solantaus and Punamäki2003; Manning & Gregoire, Reference Manning and Gregoire2009; Melchior & van der Waerden, Reference Melchior and van der Waerden2016). There is also some evidence for parental religiosity playing a role in offspring mental health, but the limited research that exists has found inconsistent associations (Bartkowski, Xu, & Levin, Reference Bartkowski, Xu and Levin2008; Schottenbauer, Spernak, & Hellstrom, Reference Schottenbauer, Spernak and Hellstrom2007; Svob et al., Reference Svob, Wickramaratne, Reich, Zhao, Talati, Gameroff and Weissman2018). We define religiosity as a combination of beliefs, behaviours and rituals related to a higher or divine power (adapted from Koenig, Reference Koenig2009). Some studies that examined the relationship between parental religiosity and child mental health have found evidence for associations between parent religiosity and child internalising (Schottenbauer et al., Reference Schottenbauer, Spernak and Hellstrom2007), such as depression and anxiety. Others have found associations with externalising problems (Bartkowski et al., Reference Bartkowski, Xu and Levin2008) such as hyperactivity, conduct disorder and oppositional defiant disorder (ODD). One study has also found an association with lower suicidal ideation (Svob et al., Reference Svob, Wickramaratne, Reich, Zhao, Talati, Gameroff and Weissman2018). However, other studies have found a mixture of positive or no relationship (van der Jagt-Jelsma et al., Reference van der Jagt-Jelsma, de Vries-Schot, de Jong, Hartman, Verhulst, Klip and Buitelaar2015, Reference van der Jagt-Jelsma, de Vries-Schot, Scheepers, van Deurzen, Klip and Buitelaar2017), or indirect relationships with mental health, through parenting factors (Kim-Spoon, Longo, & McCullough, Reference Kim-Spoon, Longo and McCullough2012). Overall, there is stronger evidence for an association between parental religiosity and internalising symptoms than externalising symptoms, but previous research has only focussed on one or two mental health outcomes per study. Our intention is to simultaneously explore a range of both internalising and externalising outcomes, which may provide evidence for differential relationships between specific mental health outcomes and parental religious belief.
Some inter-study variability may be attributed to small sample sizes (e.g. Kim, McCullough, & Cicchetti, Reference Kim, McCullough and Cicchetti2009; Kim-Spoon et al., Reference Kim-Spoon, Longo and McCullough2012; Svob et al., Reference Svob, Wickramaratne, Reich, Zhao, Talati, Gameroff and Weissman2018; van der Jagt-Jelsma et al., Reference van der Jagt-Jelsma, de Vries-Schot, Scheepers, van Deurzen, Klip and Buitelaar2017; Varon & Riley, Reference Varon and Riley1999 all have less than 600 participants) and inadequate adjustment for confounders (e.g. Kim-Spoon et al., Reference Kim-Spoon, Longo and McCullough2012; Schottenbauer et al., Reference Schottenbauer, Spernak and Hellstrom2007; van der Jagt-Jelsma et al., Reference van der Jagt-Jelsma, de Vries-Schot, Scheepers, van Deurzen, Klip and Buitelaar2017, which use no confounds, only parenting variables, and marital status and socioeconomic status, respectively). These studies are also limited due to study design – of all the studies mentioned that examine parental religiosity and offspring mental health, only three use a longitudinal cohort study design, (the rest being cross-sectional) and none of these examined parental belief before the birth of the child.
Parental mental health is a potentially important confounder of the relationship between parental religiosity and child mental health. A large body of evidence indicates that parental mental health is a strong and consistent predictor of offspring mental health outcomes (Bould et al., Reference Bould, Koupil, Dalman, DeStavola, Lewis and Magnusson2015; Dean et al., Reference Dean, Stevens, Mortensen, Murray, Walsh and Pedersen2010; Jacobs, Miller, Wickramaratne, Gameroff, & Weissman, Reference Jacobs, Miller, Wickramaratne, Gameroff and Weissman2012; Jacobs, Talati, Wickramaratne, & Warner, Reference Jacobs, Talati, Wickramaratne and Warner2015), and an individual's religiosity is consistently, positively related to their own mental health in US samples (Braam & Koenig, Reference Braam and Koenig2019; Koenig, Reference Koenig2009). Parental socioeconomic position (SEP) is also a plausible confounder of the relationship between parental religiosity and offspring mental health. Higher parental SEP is associated with better mental and physical health in offspring (Cohen, Yoon, & Johnstone, Reference Cohen, Yoon and Johnstone2009; Lemstra et al., Reference Lemstra, Neudorf, D'Arcy, Kunst, Warren and Bennett2008; Vukojević et al., Reference Vukojević, Zovko, Talić, Tanović, Rešić, Vrdoljak and Splavski2017). There is also evidence for a relationship (albeit an inconsistent one) between socioeconomic variables and religiosity (Brandt & Henry, Reference Brandt and Henry2012; Heaton, Reference Heaton2013; Horowitz & Garber, Reference Horowitz and Garber2003; Mueller & Johnson, Reference Mueller and Johnson1975; Schieman, Reference Schieman2010; Schwadel, Reference Schwadel2015; Storm, Reference Storm2017; Thompson, Thomas, & Head, Reference Thompson, Thomas and Head2012), which was also identified in Avon Longitudinal Study of Parents and Children (ALSPAC) (Halstead, Heron, & Joinson, Reference Halstead, Heron and Joinson2022; Major-Smith et al., Reference Major-Smith, Halstead, Morgan, Tohidinik, Iles-Caven, Golding and Northstone2023). Adverse childhood experiences (ACEs) are also associated with subsequent religious struggles (e.g. feelings of abandonment by God) such as after the death of a loved one or life-threatening events (McCormick, Carroll, Sims, & Currier, Reference McCormick, Carroll, Sims and Currier2017), but also a desire to connect to a higher power (Santoro, Suchday, Benkhoukha, Ramanayake, & Kapur, Reference Santoro, Suchday, Benkhoukha, Ramanayake and Kapur2016), and a parent's ACEs are associated with offspring mental health outcomes (Schickedanz, Halfon, Sastry, & Chung, Reference Schickedanz, Halfon, Sastry and Chung2018). Finally, greater parental age is simultaneously related to greater religiosity (Schwadel, Reference Schwadel2011), and better child mental health (albeit inconsistently) (Zondervan-Zwijnenburg et al., Reference Zondervan-Zwijnenburg, Veldkamp, Neumann, Barzeva, Nelemans, Beijsterveldt and Boomsma2020), and should be included as a confounder.
Furthermore, previous research has been dominated by US samples. There is evidence for differences in the relationship between religious belief and mental health in the US compared to other countries, such as the UK (United Kingdom), Korea, Spain, The Netherlands, Slovenia, Estonia, Portugal, and Chile (King et al., Reference King, Marston, McManus, Brugha, Meltzer and Bebbington2013; Leurent et al., Reference Leurent, Nazareth, Bellón-Saameño, Geerlings, Maaroos, Saldivia and King2013; Lewis, Maltby, & Day, Reference Lewis, Maltby and Day2005; Park, Hong, Park, & Cho, Reference Park, Hong, Park and Cho2012). Epidemiological and psychological studies that examine parental religious belief and offspring mental health have also almost exclusively used single item measures of religiosity – attendance at a place of worship, or the importance of religion in their lives. While commonly used in the religious belief literature, church attendance functions poorly when differentiating Atheists and Agnostics, which are both unlikely to attend church, barring weddings, funerals etc. Additionally, the importance of church attendance may also differ between religious denominations. Consequently, use of these items may be artificially constraining the variety of distinct kinds of religious (non)belief and conceal their true relationships to outcome variables. Furthermore, given the nature of religiosity, which consists of a range of beliefs and practices, it is likely to be a multidimensional construct, and unlikely to be comprehensively measured with a single dimension or item.
Religiosity also relates to a variety of childhood psychosocial outcomes, such as higher self-worth (Top, Chadwick, & McClendon, Reference Top, Chadwick and McClendon2003), academic achievement/scholastic competence (Jeynes, Reference Jeynes2003; McKune & Hoffmann, Reference McKune and Hoffmann2009), and lower antisocial behaviours (Adamczyk, Reference Adamczyk2012; Laird, Marks, & Marrero, Reference Laird, Marks and Marrero2011; Munir & Malik, Reference Munir and Malik2020). However, the role of parental religiosity has not been extensively examined (Abar, Carter, & Winsler, Reference Abar, Carter and Winsler2009; Bartkowski et al., Reference Bartkowski, Xu and Levin2008; Regnerus, Reference Regnerus2003), with a mixture of positive, negative, and no associations.
The present study, based on data from a large birth cohort, the ALSPAC, examines the prospective relationship between maternal religiosity and a range of child mental health outcomes (parent-reported) and psychosocial outcomes (child-reported) at age 7–8 years. This study addresses limitations of previous studies by using latent classes that describe patterns of maternal religious belief (Halstead et al., Reference Halstead, Heron and Joinson2022) measured before the birth of the child. In the present study the latent classes describe qualitatively distinct patterns of religious belief and distinguish between highly religious, moderately religious, Agnostic and Atheist. The analysis adjusts for a range of confounders including parental mental health, parental adverse childhood events, demographic variables, and SEP indicators.
We have the following research objectives:
(1) Examine the risk ratios of each religious latent class for the following mental health symptoms – Attention deficit hyperactivity disorder (ADHD), conduct disorder, depression, general anxiety, obsessive compulsive disorder (OCD), social and specific phobias, oppostional defiant disorder (ODD), and separation anxiety using logistic regression in relation to a reference class taken from a latent class analysis of maternal religious beliefs.
(2) Examine the risk ratios of each religious latent class for the following psychosocial outcomes – antisocial behaviour, low scholastic competence, low self-worth, victim of overt bullying, overt bully, relational bully, victim of relational bullying, and unhappiness with friends, using logistic regression in relation to a reference class taken from a latent class analysis of maternal religious beliefs.
Methods
Participants
The ALSPAC was established to understand how genetic and environmental characteristics influence health and development in parents and children. All pregnant women resident in a defined area in the Southwest of England, with an expected date of delivery between 1 April 1991 and 3 December 1992 were invited to take part in the study. The initial number of pregnancies enrolled is 14 541. Of these initial pregnancies, there was a total of 14 676 foetuses, resulting in 14 062 live births and 13 988 children who were alive at 1 year of age. These parents and children have been followed over the last 30 years and have completed a variety of questionnaires concerning their demographics, physiological and genetic data, life events, physical, and psychological characteristics. For more information, see Boyd et al., Reference Boyd, Golding, Macleod, Lawlor, Fraser, Henderson and Davey Smith2013; Fraser et al., Reference Fraser, Macdonald-Wallis, Tilling, Boyd, Golding, Davey Smith and Lawlor2013. Please note that the study website contains details of all the data that is available through a fully searchable data dictionary and variable search tool (http://www.bristol.ac.uk/alspac/researchers/our-data/). Overall, our sample was 97.4% White at baseline, 64.4% identified as Church of England, 15.3% as No religion, 8.3% as Roman Catholic, 7.8% as Other Christian and the remainder as a non-Christian religious group (Fig. 1).
Sample flowchart showing each stage of exclusion.
Exposures
Latent classes of maternal religiosity
Latent class analysis is a ‘person-centred’ statistical technique (Nylund-Gibson & Choi, Reference Nylund-Gibson and Choi2018) that uses a set of observed variables and the conditional probability of responding in a particular pattern to those variables to probabilistically assign participants to a mutually exclusive unobserved group (i.e. latent class).We used the maternal latent class membership variables derived by Halstead et al. (Reference Halstead, Heron and Joinson2022) as our indicators of religiosity in the mothers of the ALSPAC parent cohort measured in 1990, based upon the assumptions that they would be the primary caregiver, the classes are stable over time (Major-Smith et al., Reference Major-Smith, Halstead, Morgan, Tohidinik, Iles-Caven, Golding and Northstone2023) and that maternal and paternal religious latent classes are associated (see Halstead et al., Reference Halstead, Heron and Joinson2022). The latent classes are composed of a series of conditional probabilities, which are used to label and describe the classes. These classes provide a more nuanced alternative to variable centred approaches that measure religiosity using a single item by providing qualitatively distinct types of religious belief rather than a simple continuum. The questions used to derive the classes include belief in God, whether a person has asked for help from God, whether they would ask for God's help when in trouble, the duration of their faith, their church attendance, and whether they have received help from individuals from their own or another religion, which were measured at the antenatal timepoint. Our choice to use the latent classes generated at the antenatal (1990) timepoint were motivated by the larger sample size provided by this timepoint. This decision is also supported by the small amount of transition between classes over time (see Major-Smith et al., Reference Major-Smith, Halstead, Morgan, Tohidinik, Iles-Caven, Golding and Northstone2023 for a descriptive account of these transitions). The classes were named the Highly Religious, Moderately Religious, Agnostic, and Atheist and each represented approximately 14, 30, 38 and 18% of the sample, respectively. The Highly Religious class exhibits a consistent endorsement of religious beliefs, high level of church attendance, and high likelihood to obtain help and support from religious individuals. The Moderately Religious class shares many of the characteristics of the Highly Religious but were highly unlikely to visit church regularly. The Agnostic class expressed an uncertainty about the existence of God and were also uncertain about whether they would ask for help from God if they were in trouble or whether God had helped them previously. The Atheists possessed strong disbelief in the existence of God, and general disagreement with statements related to religious belief and practice. For details of the questions used to derive the latent classes and the conditional probabilities of each class, see Table S2 and S3 of the appendix, respectively.
Outcomes
Parent-reported child mental health outcomes
When their study child was aged 7 years (approximately 1997/8), mothers were asked to complete the Development and Wellbeing Assessment (DAWBA) (Goodman, Ford, Richards, Gatward, & Meltzer, Reference Goodman, Ford, Richards, Gatward and Meltzer2000) which includes questions about symptoms of common mental health disorders. This was based upon the rationale that age 7/8 are associated with mental health problems in adolescence and adulthood (Fergusson, John Horwood, & Ridder, Reference Fergusson, John Horwood and Ridder2005), and we are looking at the associations with maternal RSBB, and maternal factors have a stronger influence in childhood than in adolescence (Bhargava & Witherspoon, Reference Bhargava and Witherspoon2015; Lau, Faulkner, & Qian, Reference Lau, Faulkner and Qian2016; Powell, Son, File, & Froiland, Reference Powell, Son, File and Froiland2012). Later timepoints will be explored as part of this programme of research. We included symptoms of separation anxiety, phobias, social anxiety, OCD, generalised anxiety disorder, depression, ADHD, ODD and conduct disorder. A small number of children met DSM-IV criteria for psychiatric disorders in the ALSPAC cohort at this age (Joinson, Heron, Emond, & Butler, Reference Joinson, Heron, Emond and Butler2006). We therefore created binary variables to indicate the presence of any symptom that was severe, e.g., rated as ‘a lot more than others’ or ‘a great deal’ (See Table 1 of the appendix for details of items used). Only the complete case samples were used for analyses. The prevalence of mental health symptoms is provided in Table S4 of the appendix.
Distribution of confounders across the samples
Note. The samples in the table correspond to the samples in the flowchart.
Self-reported psychosocial outcomes
Child self-reported psychosocial outcomes were obtained from a ‘Focus Clinic’ attended by children when they were aged 8 years. The prevalence of psychosocial outcomes is provided Table S5 of the appendix. All reported Cronbach's α refer to the current paper, using the complete case samples.
Bullying (as the bully and victim)
We assessed peer victimisation through self-report using a modified version of the bullying and friendship interview schedule (Wolke, Woods, Stanford, & Schulz, Reference Wolke, Woods, Stanford and Schulz2001). The scale consists of measures of being an overt victim (α = 0.593), relational victim (α = 0.623), overt bully (α = 0.510), and relational bully (α = 0.447).
Scholastic competence and global self-worth
This measure used a 12-item version of Harter's Self Perception Profile for Children (Harter, Reference Harter1985) comprising global self-worth (α = 0. 651) and scholastic competence (α = 0. 691), rather than the full 36 item scale that contains the other subscales.
Unhappy with friends
A series of five questions taken from the Cambridge Hormones and Moods Project Friendship Questionnaire (Goodyer, Wright, & Altham, Reference Goodyer, Wright and Altham1990) to indicate unhappiness with friends (α = 0. 503).
Antisocial activities
This measure used 15 questions, including 11 from the Self-Reported Antisocial Behaviour for Young Children Questionnaire (Loeber, Stouthamer-Loeber, Van Kammen, & Farrington, Reference Loeber, Stouthamer-Loeber, Van Kammen, Farrington and Klein1989), three dummy questions and an additional example question, to measure antisocial activities (α = 0. 582).
Confounders
We chose confounders based upon empirical evidence of a relationship with the exposure and outcome variables. Confounders were assessed by maternal reports in the antenatal period and included maternal age at baseline, maternal mental health, and SEP indicators. We also adjusted for retrospective reports of maternal ACEs assessed in questionnaires completed during the antenatal period and when the study child was aged 2 years. Details of for the construction of the confounding variables are provided in Table S1 of the appendix.
Statistical analyses
We used logistic regression to calculate odds ratios and 95% confidence intervals for each mental health and psychosocial variable and their association with maternal religious latent class, both before and after adjustment for confounders. As the outcomes we are examining are rare (⪅10%) we interpreted the effect estimates as risk ratios, and we attributed changes seen to parameter estimates in the multivariable models as being due to confounding (Zammit, Allebeck, Andreasson, Lundberg, & Lewis, Reference Zammit, Allebeck, Andreasson, Lundberg and Lewis2002). Risk ratios were estimated in relation to the Agnostic class, as this was the largest class in the sample, and was characterised by the most ‘neutral’ beliefs (i.e. neither religious nor Atheist) (Johfre & Freese, Reference Johfre and Freese2021).Parameter estimates were then adjusted for confounders. The dataset was constructed in R studio (R Core Team, 2021) and all analyses were carried out in Mplus (Version 8.7), using a bias adjusted 3 step latent class analysis which incorporates uncertainty in latent class assignment (Heron, Croudace, Barker, & Tilling, Reference Heron, Croudace, Barker and Tilling2015; Vermunt, Reference Vermunt2010; Vermunt & Magidson, Reference Vermunt and Magidson2021).
Weighted estimates
To address potential bias due to missing data, we created weighting variables based on SEP variables as SEP is associated with attrition in the ALSPAC sample (Fernández-Sanlés et al., Reference Fernández-Sanlés, Smith, Clayton, Northstone, Carter, Millard and Lawlor2021; Howe, Tilling, Galobardes, & Lawlor, Reference Howe, Tilling, Galobardes and Lawlor2013). This was done by using inverse probability weighting, using SEP indicators with minimal missingness (less than 5%) including home ownership status, cigarette smoking, car ownership, and education, with any missingness recoded in these variables to be the modal response category. These were then used as exposures in a logistic regression model, with missingness at the 7-year and 8-year timepoints (when our outcome variables were measured) as the outcome variables. From these models, the weights were calculated and added to the main dataset, for use in the weighted analyses. The results of the weighted analyses are presented as the main results and the unweighted results are provided in Table S8 and S9 of the appendix. Additionally, full details of each adjusted model may be found in Table S6 and S7 of the appendix.
Results
Compared with the baseline sample, the restricted sample used in the complete case analysis comprised a higher proportion of participants of higher SEP (i.e. homeowners, non-manual social class, no major financial problems, and no financial hardship). The restricted sample also had a lower proportion of maternal depression, anxiety and adverse childhood events. See Table 1 for more details.
Association between maternal religiosity latent classes and parent-reported child mental health outcomes
Table 2 shows the results of the logistic regressions with parent-reported mental health variables as outcomes. In the unadjusted models, children of mothers in the Highly Religious class, compared with the Agnostic class have increased risk of ADHD, depression, OCD, and ODD. Compared with the Agnostic class, children of mothers in the Moderately Religious class have increased risk of depression, anxiety, and ODD. Compared with the Agnostic class, children of mothers in the Atheist class have increased risk of ADHD, and conduct disorder. The highest level of attenuation in the adjusted models was found for conduct disorder, with a 9% reduction in the risk ratio in the fully adjusted model. There was little evidence of confounding in the adjusted models for the other mental health symptoms.
Weighted risk ratios and 95% confidence intervals for the associations between maternal religious latent class and offspring mental health at age 7, with the Agnostic class as the reference class
Note. Model 1 is the unadjusted model, Model 5 adjusts for maternal age, SEP, ACE, and maternal mental health. p values are omnibus p values based on a Wald test with 3 df.
Association between maternal religiosity latent classes and self-reported psychosocial outcomes
Table 3 that compared with the Agnostic class, children of mothers in the Highly Religious class have increased risk of antisocial behaviour, and being the victim of relational bullying and decreased risk of being an overt bully. Compared with the Agnostic class, children of mothers in the Moderately Religious class have increased risk of self-reported antisocial behaviour and low scholastic competence. Compared with the Agnostic class, children of mothers in the Atheist class have increased risk of self-reported antisocial behaviour, but lower risk of being overt bullies and being unhappy with friends. The highest level of attenuation in the adjusted models was found for low scholastic competence, with an 8% increase in the risk ratio in the fully adjusted model. There was little evidence of confounding in the adjusted models for the other psychosocial outcomes.
Weighted risk ratios and 95% confidence intervals for the associations between maternal religious latent class and offspring psychosocial outcomes at age 8, with the Agnostic class as the reference class
Note. Model 1 is the unadjusted model, Model 5 adjusts for maternal age, SEP, ACE, and maternal mental health. p values are omnibus p values based on a Wald test with 3 df.
Discussion
The current study found evidence that maternal religious belief was associated with a range of mental health and psychosocial outcomes in their offspring at age 7–8 years. Compared with children of Agnostic mothers, the children of Highly religious and Moderately religious parents were at greater risk of internalising symptoms, and the children of Atheist parents were at greater risk of externalising symptoms. However, there was no clear pattern of results for psychosocial outcomes. These associations were independent of maternal age, SEP variables, ACE, and mental health. There were only a few instances of attenuation, with most of the results being robust to the inclusion of confounding variables.
Strengths and limitations
There are several strengths to this study. The use of data from a large prospective community based cohort, use of latent classes of belief, rather than relying upon single item measures which dominate the religiosity literature, availability of parent- and child-reported mental health and psychosocial outcomes based on validated questionnaires, and availability of data on a wide range of important confounders The use of latent classes provided insights that may have otherwise been lost by using items that fail to differentiate between Agnostic and Atheist individuals, such as through the use of items that ask participants to indicate the importance of religion in their lives or church attendance.
There are also limitations that should be considered when interpreting the findings. There was a large amount of attrition between the baseline sample, and the final complete case analysis which could lead to selection bias. Specifically, those with higher SEP (Howe et al., Reference Howe, Tilling, Galobardes and Lawlor2013) and religiosity (Morgan, Halstead, Northstone, & Major-Smith, Reference Morgan, Halstead, Northstone and Major-Smith2022) are more likely to participate in ALSPAC initially and continue to participate over time. Attrition is strongly related to low SEP in ALSPAC, and the present study attempted to mitigate this using weighting to account for the potential bias, in line with recommendations (Howe et al., Reference Howe, Tilling, Galobardes and Lawlor2013). Our study is also limited in its generalisability, given the very small numbers of Non-White, and Non-Christian religious individuals, which prevents us from generalising to these groups. Furthermore, due to the data on religious belief and mental health being collected in the 90s, there is the possibility of cohort effects, as well as the changes in prevalence and perception of religion over time. These factors may mean that our findings may not generalise to the present day. Finally, it is possible that the concordance/discordance between maternal and paternal religious classes may play a role in offspring mental health outcomes (van der Jagt-Jelsma et al., Reference van der Jagt-Jelsma, de Vries-Schot, de Jong, Verhulst, Ormel, Veenstra and Buitelaar2011). This possibility should be explored in future research.
Comparisons with previous research
Our findings are contrary to previous research which examined the relationship between parental religious belief and child mental health, which has found that greater parental (either maternal or paternal) religious belief is associated with better mental health outcomes in offspring (Svob et al., Reference Svob, Wickramaratne, Reich, Zhao, Talati, Gameroff and Weissman2018; Varon & Riley, Reference Varon and Riley1999). A possible reason for this difference is that the current study is based on a UK sample, compared to previous research which has predominantly been based on US samples (e.g. 75% of the religious belief and mental health research reviewed by Koenig, Reference Koenig2009 was conducted on US samples). There is evidence the UK differs in its relationship with religious belief and mental health compared to other countries, with some studies finding that increased religious or spiritual belief was associated with worse mental health outcomes such as depression, anxiety, and phobias (King et al., Reference King, Marston, McManus, Brugha, Meltzer and Bebbington2013; Leurent et al., Reference Leurent, Nazareth, Bellón-Saameño, Geerlings, Maaroos, Saldivia and King2013). Additionally, given that the US is highly religious compared to the UK, and most previous studies contain relatively small numbers of individuals that could be considered Atheist or Agnostic, previous studies may be less suited to capturing the differences between them and religious individuals.
The use of items that assume religious belief exists on a simple continuum from non-religious to highly religious may be obscuring the true nature of religion's relationship with mental health and psychosocial outcomes. The relationship between religious belief and mental health may be better explained by using qualitatively different types of belief in future research. Given previous literature's inclination to use items that ask about religious attendance or the importance of religion, and both Atheists and Agnostics are likely to respond the same way to these items (i.e. they are both unlikely to attend church), previous research may be failing to acknowledge the way these items function for different groups. Conflating Atheist and Agnostic groups appears to hide important differences in outcomes between the two.
Possible mechanisms explaining the current findings
The increased mother-reported externalising problems and psychosocial outcomes in the Highly Religious mothers (e.g. ADHD, ODD, antisocial behaviour) may be partially explained by highly religious mothers having higher expectations of their child's morals (Rhodes & Nam, Reference Rhodes and Nam1970; Smith, Reference Smith2003), an increased likelihood to perform parental monitoring activities (Guo, Reference Guo2018; Kim & Wilcox, Reference Kim and Wilcox2014) and to be more engaged in their child's life (Guo, Reference Guo2018). This possibly leads them to be more attentive to ADHD or ODD symptoms, as well as negative (e.g. antisocial) behaviours. However, children of religious parents also perceive their parents to be more controlling, which could in turn lead to more internalising and externalising problems (Bornstein et al., Reference Bornstein, Putnick, Lansford, Al-Hassan, Bacchini, Bombi and Alampay2017). Combined, this suggests that a degree of monitoring or control is healthy and may lead to more attentive parenting, but when it is perceived to be excessive, it can be a stressor to the child. In the context of the current study, the higher degree of control that religious mothers exert over their children may impact their levels of stress, leading to a greater risk of depression or anxiety (Rapee, Reference Rapee1997; Yap, Pilkington, Ryan, & Jorm, Reference Yap, Pilkington, Ryan and Jorm2014). The increased mother reported internalising symptoms may be explained by children of religious parents being religious themselves, which may incline them towards rumination, which leads to greater internalising symptoms (Saunders et al., Reference Saunders, Svob, Pan, Abraham, Posner, Weissman and Wickramaratne2021). Religious belief is associated with OCD (Abramowitz, Deacon, Woods, & Tolin, Reference Abramowitz, Deacon, Woods and Tolin2004; Himle, Chatters, Taylor, & Nguyen, Reference Himle, Chatters, Taylor and Nguyen2011; Sica, Novara, & Sanavio, Reference Sica, Novara and Sanavio2002), possibly through heightened pathogen disgust sensitivity (Olatunji et al., Reference Olatunji, Williams, Tolin, Abramowitz, Sawchuk, Lohr and Elwood2007), which is also associated with religiosity (Stewart, Adams, & Senior, Reference Stewart, Adams and Senior2020; Yu, Bali, Tsikandilakis, & Tong, Reference Yu, Bali, Tsikandilakis and Tong2022). Consequently, the behaviours associated with OCD symptoms may be taught to offspring during their upbringing (Waters & Barrett, Reference Waters and Barrett2000). There is little existing literature to explain the pattern of results in the Atheist class, or for the differences between the Moderately and Highly Religious classes. We may nevertheless speculate that the association between spirituality, depression, and other internalising disorders may be partially explained by an increased interior-orientation, observed for example in rumination (Saunders et al., Reference Saunders, Svob, Pan, Abraham, Posner, Weissman and Wickramaratne2021) and default-mode network connectivity (Svob, Wang, Weissman, Wickramaratne, & Posner, Reference Svob, Wang, Weissman, Wickramaratne and Posner2016) in those who are spiritual. Whereas Atheism may support a more external worldview and contribute to its greater likelihood to be associated with externalising disorders. Further research is needed to examine the replicability of our findings in non-US samples.
Conclusions
Contrary to existing literature, we found evidence that maternal religiosity is associated with a higher risk of internalising symptoms. Children of Atheist parents are at greater risk of externalising symptoms. Future research is needed to determine whether these relationships are causal and to identify the underlying mechanisms. For example, it is conceivable that the relationship may be mediated by parenting style/quality variables or moderated by partner religious class.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S003329172300079X.
Data availability
Please see the ALSPAC data management plan which describes the policy regarding data sharing (http://www.bristol.ac.uk/alspac/researchers/data-access/documents/alspac-data-management-plan.pdf), which is by a system of managed open access. Data used for this submission will be made available on request to the Executive (alspac-exec@bristol.ac.uk). The datasets presented in this article are linked to ALSPAC project number B4001, please quote this project number during your application.
The steps below highlight how to apply for access to the data included in this study and all other ALSPAC data:
(1) Please read the ALSPAC access policy (http://www.bristol.ac.uk/media-library/sites/alspac/documents/researchers/data-access/ALSPAC_Access_Policy.pdf) which describes the process of accessing the data and samples in detail, and outlines the costs associated with doing so.
(2) You may also find it useful to browse our fully searchable research proposals database (https://proposals.epi.bristol.ac.uk/?q=proposalSummaries), which lists all research projects that have been approved since April 2011.
(3) Please submit your research proposal (https://proposals.epi.bristol.ac.uk/) for consideration by the ALSPAC Executive Committee. You will receive a response within 10 working days to advise you whether your proposal has been approved.
Acknowledgements
We are extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses.
Financial support
The UK Medical Research Council and Wellcome Trust (Grant ref: 217 065/Z/19/Z) and the University of Bristol currently provide core support for ALSPAC. This publication is the work of the authors, and they will serve as guarantors for the contents of this paper. A comprehensive list of grants funding is available on the ALSPAC website (http://www.bristol.ac.uk/alspac/external/documents/grant-acknowledgements.pdf). This project was made possible through the support of grants from the John Templeton Foundation (ref no. 61356 and 61917). The opinions expressed in this publication are those of the author(s) and do not necessarily reflect the views of the John Templeton Foundation.
Conflict of interest
The authors declare none.
Ethical standards
Ethical approval for the study was obtained from the ALSPAC Ethics and Law Committee and the Local Research Ethics Committees. Informed consent for the use of data collected via questionnaires and clinics was obtained from participants following the recommendations of the ALSPAC Ethics and Law Committee at the time.
Open access
