The association between religiosity and pregnancy acceptability among Latino/a young adults: Does generational status matter?

Abstract

The US Latino/a population is considered to be at high risk for unintended pregnancy; some research indicates that Latino/a parents are more likely to express happiness about an unintended pregnancy than other racial/ethnic groups. Associations between pregnancy attitudes and factors such as religiosity and nativity have also been documented in the Latino/a population, but existing research is sparse, dated and primarily focused on women of Mexican heritage. This study sought to expand this literature by examining the relationship between religiosity and pregnancy acceptability and assessing effect modification by generational status and gender in a national sample of young Latino/a cisgender women and men of various ancestral backgrounds currently in relationships. In multivariable logistic regression models, there was a positive association between importance of religion and pregnancy acceptability for both men and women; being highly or moderately religious was associated with elevated odds of finding a pregnancy acceptable. Effect modification by generational status was significant for women, but not men. Results suggest that religiosity, gender, and generational status have differential influences on and relationships to pregnancy orientations for Latina women and men and should be considered in the design and delivery of family planning care for Latino/a clients.

Introduction

Despite recent declines, the USA still has higher rates of pregnancies classified as unintended compared to other industrialized nations, with 45% of pregnancies described as either mistimed or unwanted, with the highest rates among women who are younger, cohabitating, lower income, and of colour (Sedgh, Singh and Hussain 2014; Guttmacher Institute 2016). What the literature calls ‘unintended pregnancy’ is often seen as a primary indicator of sexual and reproductive health (Guttmacher Institute 2016; U.S. Department of Health and Human Services, Office of Disease Prevention and Health Promotion). Although the association between unintended pregnancy and health outcomes in the literature is mixed, some studies link it to adverse health outcomes, including increased risk for physical and mental health problems, low birthweight, lower rates of breastfeeding and insufficient prenatal care (Gipson, Koenig and Hindin 2008). Research has consistently demonstrated higher rates of unintended pregnancy among people of colour in the USA, particularly in women from one of the fastest growing (and currently largest) ethnic minority groups in the nation—the Latino/a¹ population (Guttmacher Institute 2016; Velasco-Mondragon et al. 2016; Flores 2017).

In addition to structural issues such as socioeconomic status, limited research has indicated that higher rates of unintended pregnancy among Latino/a individuals may be associated with factors such as nativity (place of birth) and religiosity (the importance of religion in one’s life) (Kim, Dagher and Chen 2016; Amaro 1988; Wilson 2008). In one of the few studies that examined unintended pregnancy among Latina mothers, Kim, Dagher, and Chen (2016) found that, compared to White mothers, being foreign-born contributed more than any other factor to Latina respondents’ higher risk for unintended pregnancy. In the same study, regardless of race/ethnicity, mothers who identified as being religious had lower odds of unintended pregnancy compared to those who were not religious (Kim, Dagher and Chen 2016). Other research with Mexican Americans revealed that highly religious women have significantly smaller family sizes and that women more oriented towards Mexican cultural values preferred more children compared to those classified as ‘bicultural’ (that is, equally oriented toward Mexican and American cultural values) (Amaro 1988; Cuellar, Harris and Jasso 1980). A few studies have also found a positive relationship between religiosity and likelihood of happiness about an unintended pregnancy among Mexican American women (Amaro 1988; Wilson 2008; Hartnett 2012; Chandra et al. 2005). Taken together, these results indicate that religiosity and nativity may be important factors for Latino/a individuals’ pregnancy intentions. However, the existing literature is dated and limited; to our knowledge, no prior studies examined associations between religiosity or nativity and pregnancy intentions among Latino/a men, and few include Latino/a individuals with non-Mexican ancestry (Amaro 1988; Wilson 2008; Hartnett 2012; Chandra et al. 2005).

There is some research that suggests that nativity may interact with religiosity to predict pregnancy attitudes (Wilson 2008; Hartnett 2012; Sabagh and Lopez 1980). For instance, an analysis of data collected in 1973 indicated that highly religious Mexican American women who were raised in the USA have more children than Mexican women with low religiosity, while no such relationship exists among women raised in Mexico (Sabagh and Lopez 1980). Harnett (2012) found that foreign-born Latina women who were highly religious were significantly more likely to be happy about an unintended pregnancy, but this relationship was not present for Latina women born in the USA. Analysing national survey data from 1995, Wilson (2008) showed that Mexican women who immigrated to the USA were less likely to have an unintended pregnancy but were more likely to be happy about a pregnancy than women born in the USA. This study also found that the association between happiness about a pregnancy and nativity was mediated by religiosity, such that a significant relationship was found between being born in the USA and happiness about a pregnancy; however, after adding religiosity to the model, the association between nativity and happiness about a pregnancy was no longer significant (Wilson 2008). This body of literature is also limited in scope and somewhat dated, with no known research conducted on this topic in over a decade. While the literature does address nativity, none of the above studies measured generational status (the generational distance from US immigration). Generational status is frequently used as an indicator of cultural shift or cultural identity, and some studies have found associations between generational status and sexual health or attitudes in the Latino/a population (Deutsch and Crockett 2016; Smith 2015; Foulkes et al. 2005). Little is known about whether the relationship between religiosity and pregnancy intention may be moderated by generational status.

Another challenge is that existing research typically uses measures of pregnancy intention that may insufficiently explain individuals’ complex attitudes and feelings toward pregnancy, especially among low-income people of colour for whom the concept of planning for pregnancy may not resonate (Jones, Frohwirth and Blades 2016; Borrero et al. 2015; Gomez et al. 2018). The difference between pregnancy intention and attitudes or emotions about a pregnancy is an additional complication. For instance, research has indicated that some people indicate happiness about a pregnancy, even when it occurred at a time when it was not wanted and therefore classified as unintended by conventional standards; thus, conventional measures of pregnancy intention may neglect more nuanced aspects of pregnancy, such as affect, and may even lead to inappropriate conclusions such as believing that if a pregnancy is classified as mistimed or unwanted, it must also be a disappointment, a negative event or a source of sadness or distress (Hartnett 2012; Chandra et al. 2005; Aiken, Dillaway and Mevs-Korff 2015; Aiken et al. 2016; Gomez et al. In press). Furthermore, research has found the phenomenon of positive emotions (such as happiness) about unintended pregnancy to be most prevalent among Latina women compared to women of other races and ethnicities (Hartnett 2012; Chandra et al. 2005). Measures of happiness or interpretation of such measures about a pregnancy also may fail to account for normative mores and structural limitations, such as financial feasibility of raising children and community/social support for childbearing and childrearing (Aiken, Dillaway and Mevs-Korff 2015). Pregnancy acceptability captures not only whether a pregnancy is wanted but how it would fit (or not fit) into individuals’ complex lives, taking into account possible social, relational, financial and cultural barriers, and has been proposed as an alternative to the traditional conceptualisation of pregnancy intention (Aiken, Dillaway and Mevs-Korff 2015; Aiken et al. 2016; Gomez et al. 2018). Aiken and colleagues (2016) argue that the construct of pregnancy acceptability adds the layers of financial security, relationship support, personal belief, and ‘internalized social and cultural norms pertaining to childbearing’ to other more simplistic measures of pregnancy attitudes. Other research suggests that young people may assess the acceptability of pregnancy by evaluating their lives holistically, including their relationship quality, desire for children in general, knowledge of parenting and perceived readiness based on meeting traditional adulthood milestones (Gomez et al. 2018).

As the Latino/a population in the USA grows, the healthcare system must address enduring inequities in patient care experienced by Latino/a patients, who typically rate their family planning care as less helpful than white patients, report more pressure to limit family size or consider sterilisation, and have lower satisfaction with the family planning care they receive (Dehlendorf et al. 2010; Borrero et al. 2009; Becker and Tsui 2008; Gomez and Wapman 2017). There is evidence to suggest that providers treat patients differently depending on their race/ethnicity; for example, providers’ implicit biases impact provision of healthcare for Latino/a patients, as they are more likely to misunderstand their Latino/a patients’ beliefs and desires than their white patients’ beliefs and desires (Street and Haidet 2011; Hall et al. 2015). Specifically, health providers’ perceptions of their patients’ health beliefs are more likely to differ from the patients’ actual beliefs (such as the meaning of health concerns or preferences for interaction with providers) when patients are Latino/a (Street and Haidet 2011). Researchers have suggested that providers’ stereotypes and/or misunderstandings about Latino/a attitudes toward childbearing may contribute to the disparities experienced by Latino/a individuals when seeking family planning care (e.g., lower patient satisfaction, pressure to use contraception or limit family size) (Velasco-Mondragon et al. 2016; Amaro 1988; Dehlendorf et al. 2010; Shedlin et al. 2013). For example, some studies document that, when interacting with Latino/a clients, healthcare providers may be deploying common historical stereotypes that portray Mexican American women as highly religious and submissive, resistant to contraception and more accepting of large family sizes (Amaro 1988; Andrade 1982).

In order to address healthcare inequities experienced by Latino/as in family planning setting, research is needed to illuminate how religiosity and generational status may be associated with pregnancy intentions. The current study aims to address current gaps in the literature by considering the effects of religiosity on pregnancy acceptability among a diverse sample of Latino/a men and women from various regions of ancestry, and by assessing whether such a relationship is moderated by generational status.

Materials and methods

Data and sample

We conducted a national, longitudinal survey of Latino/a young adults using the existing online panel of YouGov, a market and political research firm. YouGov used proximity matching to collect a subsample of possible panel participants for the survey, ensuring that the sample was as representative of the US Latino/a population as possible by matching it to a sampling frame of US citizens from the 2012 American Community Survey, 2012 Current Population Survey, and 2007 Pew U.S. Religious Landscape Study (YouGov 2017; Rivers 2006). Selected panel members completed a screening survey assessing inclusion criteria: identifying as Latino/a or Hispanic, identifying as a cisgender woman or man, being 18–34 years old, being in a committed and heterosexual relationship, and not being currently pregnant or actively trying to become pregnant. The study protocol was approved by the Committee for Protection of Human Subjects at the University of California, Berkeley. This analysis utilised the baseline survey data (n=1442), collected in June-July 2017.

In the current analysis, respondents who reported using female or male sterilisation (n=63) were excluded, as were participants who were born in the USA and missing data on generational status owing to not having any parental figures (n=29), or missing data on religiosity (n=1). The final, unweighted sample size was 1,349.

Measures

Independent variable

Religiosity was assessed using the question, ‘How important is religion in your life?’ with four response options: not at all important, not too important, somewhat important, very important. Response options were collapsed into three categories: not at all important (low religiosity), somewhat or not too important (moderate religiosity), and very important (high religiosity). The middle two categories were combined in this analysis both because they were conceptually similar and doing so allowed for increased power to assess for interaction by generational status. In sensitivity analyses, maintaining the original four categories did not produce significantly different results (results not shown).

Outcome variable

Pregnancy acceptability was assessed using the question, ‘How okay would it be if you got pregnant/got a partner pregnant right now?’ Response options included: not at all okay, a little okay, somewhat okay, and very okay. Because the current study aimed to better understand the differences between respondents who would find a pregnancy completely unacceptable versus those who would find one at all acceptable, we utilised a binary outcome variable that captured ‘any acceptability’ (very, somewhat, or a little okay) and ‘no acceptability’ (not at all okay).

Moderating variable

A generational status variable was created using three questions: (1) ‘Were you born in the United States?’; (2) ‘Was [the woman who mostly raised you when you were growing up] born in the United States?’; and (3) ‘Was [the man who mostly raised you when you were growing up] born in the United States?’ (Prior to questions 2 and 3, respondents identified the parental figures that mostly raised them, e.g., their biological mother or father, stepparent, grandparent.) Respondents who were not born in the USA were categorised as first generation. Those born in the U.S. and who had at least one parental figure born outside of the USA were categorised as second generation. Finally, respondents born in the USA and whose parental figures were also born in the USA were categorised as third or more generation (US Census Bureau n.d.).

Covariates

Region of ancestry was assessed using an item that read, ‘What is your background?’ Response options included a list of Latino/a/Hispanic nationalities. We classified responses into six categories: Mexican/Chicano/a, Puerto Rican/Cuban/Dominican, Central American, South American, European, and multiple ancestry regions.

Covariates included participant age (in years), number of children (0, 1, 2, 3 or more), annual household income (Less than $20,000, $20,000–$39,999, $40,000–59,999, $60,000–99,999, and $100,000 or above), current employment status (full-time, part-time, not working, full-time caregiver, student), and relationship status (having a main partner, boyfriend, or girlfriend; being in a serious relationship with someone or engaged; or being married).

Analytic approach

Bivariate analyses were conducted using chi-square tests and t-tests to assess for associations between each covariate and the exposure and outcome, stratified by gender. Multivariable logistic regression models were estimated to investigate the relationship between religiosity and pregnancy acceptability, including effect modification by generational status. Stata’s lincom command was used to obtain point estimates and 95% confidence intervals (CI) for the linear combination of regression coefficients in analyses with interaction terms. Wald tests were used to assess the significance of the addition of each interaction term to the model.

We expected that associations between religiosity and pregnancy acceptability might differ by gender, therefore, all results are presented for both the overall sample and stratified by gender. Additionally, because we also anticipated that the relationship between religiosity and generational status might vary by gender, we tested a three-way interaction between gender, generational status, and religiosity.

All analyses were completed in Stata statistical analysis software (version 14), using the svy commands to account for complex survey design (StataCorp 2015). We applied sampling weights generated by YouGov, which used propensity score matching using sociodemographic variables, including age, gender, education, and relationship status, to create weights to match the subsample to the sampling frame of US citizens from the 2012 American Community Survey, 2012 Current Population Survey, and 2007 Pew U.S. Religious Landscape Study (YouGov 2017; Rivers 2006).

Results

Descriptive statistics are presented in Table 1. The mean respondent age was 26.22 years (SE=0.17). Most respondents were married (38.73%) or in serious or engaged relationships (33.92%), had children (51.81%), and worked full-time (44.23%). About half the sample (51.48%) identified as Mexican or Chicano/a. Most respondents were second (40.21%) or third or more (35.57%) generation. Regarding the importance of religion, about one-quarter (25.65%) of respondents said that religion was very important to them, while half (52.59%) said somewhat or not too important, and about one-fifth (21.77%) said not at all important. Just over half the sample (55.79%) said that a pregnancy would not be acceptable right now (58.99% of women; 50.87% of men; F=20.53, p=0.02; results not shown).

Table 1.

Demographic characteristics of study sample

	Total sample (unweighted n=1,349)		Women (unweighted n=619)		Men (unweighted n=729)
	Unweighted N	Weighted %	Unweighted N	Weighted %	Unweighted N	Weighted %
Age (in years)
Mean (SE)	26.22 (0.17)		26.04 (0.23)		26.50 (0.22)
Number of children
0	785	48.19	309	41.12	476	59.08
1	254	20.19	122	20.50	132	19.71
2	179	17.56	112	20.98	67	12.29
3 or more	130	14.06	76	17.40	54	8.92
Annual household income
Below $20,000	215	23.31	129	28.40	86	15.57
$20,000–39,999	356	34.23	190	37.80	166	29.11
$40,000–59,999	261	18.68	108	16.77	153	21.42
$60,0000–99,999	265	15.38	75	10.39	190	22.53
$100,000 or above	158	8.58	49	6.63	109	11.37
Employment status
Full-time employed	706	44.23	225	31.93	481	62.83
Part-time employed	216	15.57	112	15.89	104	15.09
Not working	109	9.75	53	9.44	56	10.21
Caregiver	141	19.28	136	31.21	5	1.24
Student	150	11.17	75	11.53	75	10.63
Relationship status
Main partner	443	27.35	155	20.27	288	38.25
Serious/engaged	430	33.92	218	36.62	212	29.75
Married	474	38.73	246	43.10	228	32.00
Generational status
1st generation	320	24.23	189	28.46	131	17.71
2nd generation	554	40.21	235	38.38	319	43.01
3rd or higher generation	474	35.57	195	33.16	279	39.28
Region of ancestry
Mexican/Chicano	645	51.48	279	50.98	366	52.27
Puerto Rican/Cuban/Dominican	242	19.14	106	18.83	136	19.62
European	44	2.58	26	3.08	18	1.79
South American	166	8.62	90	9.40	76	7.40
Central American	66	5.23	34	5.78	32	4.38
Multiple	175	12.95	82	11.94	93	14.54
Importance of religion
Very important	345	25.65	168	27.27	177	23.14
Somewhat/not very important	683	52.59	329	53.73	354	50.82
Not at all important	320	21.77	122	19.00	198	26.03
Pregnancy acceptability
Not at all okay	746	55.79	383	58.99	363	50.87
A little/somewhat/very okay	602	44.21	236	41.01	366	49.13

Age, number of children, relationship status, and employment status were significantly associated with both the exposure and outcome variables in bivariate analyses (Tables 2 and 3). Region of ancestry and income were not significantly associated with the exposure and outcome variables but yielded a greater than 10% change in effect size when removed from the adjusted logistic regression model with all potential covariates (results not shown). All variables were considered confounders and were therefore included in the multivariable logistic multiple regression analysis.

Table 2.

Bivariate associations between religiosity and covariates

	Women (unweighted n= 619)					Men (unweighted n= 729)
	Religiosity					Religiosity
	High	Moderate	Low			High	Moderate	Low
				Test statistic	P-value				Test statistic	P-value
Age (in years)				F=3.86	0.0213				F=1.08	0.3397
Mean (SE)	26.67 (0.37)	25.46 (0.33)	26.79 (0.54)			26.58 (0.44)	26.74 (0.32)	25.96 (0.43)
Number of children				F= 1.33	0.2383				F= 3.48	0.0021
0	62 (33.89)	174 (43.56)	73 (44.58)			101 (52.21)	212 (53.56)	163 (75.95)
1	40 (20.54)	59 (20.06)	23 (21.71)			38 (20.08)	78 (23.41)	16 (12.16)
2	45 (28.43)	49 (16.57)	18 (22.78)			20 (16.03)	33 (12.21)	14 (9.11)
3 or more	21 (17.14)	47 (19.81)	8 (10.93)			18 (11.68)	31 (10.81)	5 (2.78)
Annual household income				F= 1.39	0.1960				F= 1.48	0.1611
Below $20,000	31 (31.83)	78 (30.14)	20 (19.46)			13 (11.92)	50 (17.62)	23 (14.77)
$20,000–39.999	47 (33.61)	103 (41.36)	40 (33.84)			45 (34.35)	80 (27.62)	41 (27.36)
$40,000–59,999	38 (20.26)	47 (12.81)	23 (22.70)			39 (18.99)	82 (23.93)	32 (18.58)
$60,000–99,999	22 (9.84)	38 (9.56)	15 (13.29)			36 (18.23)	90 (21.27)	64 (28.97)
$100,000 or above	10 (4.46)	25 (6.13)	14 (10.71)			38 (16.52)	42 (9.56)	29 (10.33)
Employment status				F= 3.29	0.0010				F= 1.09	0.3695
Full-time employed	48 (25.15)	122 (31.02)	55 (44.49)			122 (66.35)	222 (60.52)	137 (64.19)
Part-time employed	29 (10.40)	65 (18.81)	18 (15.40)			23 (12.89)	56 (15.83)	25 (15.59)
Not working	18 (12.96)	25 (7.26)	10 (10.67)			13 (8.65)	33 (13.25)	10 (5.72)
Caregiver	56 (44.39)	65 (30.71)	15 (13.44)			1 (0.66)	2 (1.00)	2 (2.22)
Student	13 (7.11)	44 (12.21)	18 (16.00)			16 (11.47)	37 (9.40)	22 (12.28)
Relationship status				F= 3.81	0.0044				F= 4.69	0.0009
Main partner	33 (16.68)	91 (21.40)	31 (22.24)			58 (29.10)	139 (37.85)	91 (47.20)
Serious/engaged	40 (24.24)	132 (41.70)	46 (40.05)			41 (24.46)	110 (31.87)	61 (30.32)
Married	95 (59.09)	106 (36.90)	45 (37.71)			78 (46.44)	105 (30.28)	45 (22.48)
Generational status				F= 0.97	0.4217				F= 1.93	0.1034
1st generation	58 (33.33)	102 (28.97)	29 (20.02)				41 (23.43)	67 (17.92)	23 (12.20)
2nd generation	58 (33.78)	130 (39.08)	47 (43.04)				71 (39.32)	161 (45.58)	87 (41.28)
3rd or higher generation	52 (32.89)	97 (31.95)	46 (36.94)				65 (37.25)	126 (36.50)	88 (46.51)
Region of ancestry				F= 0.87	0.5533				F= 0.68	0.7385
Mexican/Chicano	82 (53.86)	138 (47.65)	59 (56.25)			81 (47.74)	189 (55.65)	96 (49.70)
Puerto Rican/Cuban/Dominican	24 (14.70)	61 (21.03)	21 (18.54)			36 (21.09)	59 (17.82)	41 (21.87)
European	5 (2.22)	13 (2.40)	8 (6.24)			6 (3.08)	9 (1.47)	3 (1.24)
South American	22 (10.03)	57 (10.39)	11 (5.66)			24 (7.99)	33 (6.65)	19 (8.37)
Central American	10 (5.32)	19 (6.46)	5 (4.52)			8 (6.59)	13 (3.43)	11 (4.23)
Multiple	25 (13.87)	40 (12.07)	17 (8.79)			22 (13.51)	47 (14.98)	24 (14.60)
Pregnancy acceptability				F= 3.40	0.0335				F= 9.32	0.0001
Not at all okay	93 (55.98)	200 (55.53)	90 (73.09)			66 (40.15)	164 (47.81)	133 (66.35)
A little/somewhat/very okay	75 (44.02)	129 (44.47)	32 (26.91)			111 (59.85)	190 (52.19)	65 (33.65)

Table 3.

Bivariate associations between pregnancy acceptability and covariates

	Women (unweighted n= 619)				Men (unweighted n= 729)
	Pregnancy not acceptable	Pregnancy acceptable			Pregnancy not acceptable	Pregnancy acceptable
			Test statistic	P-value			Test statistic	P-value
Age (in years)			F=20.53	0.0000			F=22.42	0.0000
Mean (SD)	25.18 (0.29)	27.29 (0.36)			25.48 (0.33)	27.56 (0.29)
Number of children			F=7.49	0.0001			F= 5.82	0.0006
0	222 (50.57)	87 (27.52)			268 (67.73)	208 (50.12)
1	51 (13.65)	71 (30.36)			43 (13.68)	89 (25.96)
2	62 (19.03)	50 (23.79)			34 (12.59)	33 (11.97)
3 or more	48 (16.75)	28 (18.33)			18 (6.00)	36 (11.95)
Annual household income			F= 1.81	0.1300			F= 0.35	0.8418
Below $20,000	90 (31.62)	39 (23.77)			47 (16.87)	39 (14.26)
$20,000–39,999	115 (35.04)	75 (41.77)			88 (27.91)	78 (30.31)
$40,000–59,999	62 (15.26)	46 (18.95)			72 (22.49)	81 (20.34)
$60,000–99,999	41 (9.52)	34 (11.65)			88 (21.26)	102 (23.81)
$100,000 or above	28 (8.56)	21 (3.86)			51 (11.47)	58 (11.27)
Employment status			F= 4.71	0.0009			F= 1.81	0.1270
Full-time employed	128 (31.44)	97 (32.63)			223 (58.88)	258 (66.98)
Part-time employed	78 (19.17)	34 (11.15)			58 (17.38)	46 (12.68)
Not working	39 (11.29)	14 (6.77)			30 (10.18)	26 (10.25)
Caregiver	69 (23.61)	67 (42.19)			1 (0.57)	4 (1.94)
Student	56 (14.49)	19 (7.25)			48 (13.00)	27 (8.15)
Relationship status			F= 9.30	0.0001			F= 9.33	0.0001
Main partner	122 (27.19)	33 (10.31)			163 (43.63)	125 (32.70)
Serious/engaged	138 (36.51)	80 (36.79)			121 (33.37)	91 (26.01)
Married	123 (36.30)	123 (52.90)			78 (23.01)	150 (41.30)
Generational status			F=0.10	0.9068			F=0.84	0.4305
1st generation	125 (28.11)	64 (28.97)				64 (17.85)	67 (17.56)
2nd generation	140 (39.34)	95 (37.00)				162 (40.36)	157 (45.76)
3rd or higher generation	118 (32.55)	77 (34.03)				137 (41.79)	142 (36.68)
Region of ancestry			F=1.64	0.1475			F= 1.90	0.0948
Mexican/Chicano	163 (47.41)	116 (56.11)			165 (45.58)	201 (59.17)
Puerto Rican/Cuban/Dominican	62 (17.19)	44 (21.19)			72 (21.40)	64 (17.79)
European	18 (3.58)	8 (2.35)			10 (1.91)	8 (1.67)
South American	64 (11.25)	26 (6.74)			42 (8.66)	34 (6.10)
Central American	21 (5.96)	13 (5.51)			16 (5.13)	16 (3.60)
Multiple	54 (14.62)	28 (8.09)			53 (17.32)	40 (11.67)
Importance of religion			F=3.40	0.0335			F=9.32	0.0001
Very important	93 (25.88)	75 (29.28)			66 (18.27)	111 (28.19)
Somewhat/not very important	200 (50.58)	129 (58.26)			164 (47.77)	190 (53.98)
Not at all important	90 (23.54)	32 (12.47)			133 (33.96)	65 (17.83)

An unadjusted logistic regression model was estimated to assess the relationship between the religiosity and pregnancy acceptability, overall and stratified by gender (Table 4). The unadjusted logistic regression analysis demonstrated a statistically significant and positive relationship between religiosity and pregnancy acceptability, such that in the overall sample, being either very religious or moderately religious was associated with about twice the odds of finding a pregnancy acceptable compared to being not at all religious (high religiosity OR 2.29; 95% CI 1.49, 3.53; moderate religiosity OR 2.09; 95% CI 1.43, 3.06). Results were similar among women and men. In adjusted analyses, the positive relationship between religiosity and pregnancy acceptability remained significant, such that having high or moderate religiosity was associated with finding a pregnancy acceptable compared to low religiosity (high religiosity OR 2.68; 95% CI 1.67, 4.33; moderate religiosity OR 2.74; 95% CI 1.87, 4.22). Results were similar in models stratified by gender, with moderately religious women having a greater magnitude for the odds of finding pregnancy acceptable (OR 3.98; 95% CI 1.88, 8.41) than moderately religious men (OR 2.22, 95% CI 1.38, 3.56), with the reference group of low religiosity.

Table 4.

Odds ratios of pregnancy acceptability by level of religiosity in crude and adjusted logistic regression models, overall and stratified by gender

Crude association between religiosity and pregnancy acceptability
OR [95% CI]	Overall	Men	Women
High religiosity	2.29 [1.49–3.53]**	2.94 [1.75–4.94]**	2.14 [1.09–4.18]*
Moderate religiosity	2.09 [1.43–3.06]**	2.15 [1.38–3.35]**	2.17 [1.18–4.00]*
Adjusted association between religiosity and pregnancy acceptability
	Overall□	Men●	Women●
High religiosity	2.68 [1.67–4.33]**	3.17 [1.83–5.50]**	2.87 [1.28–6.43]**
Moderate religiosity	2.74 [1.78–4.22]**	2.22 [1.38–3.56]**	3.98 [1.88–8.41]**

Notes:

Reference group for all models is low religiosity

^□Model adjusts for gender, number of children, employment, age, relationship status, region of ancestry, and generational status

^●Model adjusts for number of children, employment, age, relationship status, region of ancestry, and generational status

^**p<0.01;

^*p<0.05

Multivariable logistic regression models assessing effect modification between religiosity and gender and between religiosity and generational status were then estimated. The Wald test for effect modification by gender did not meet the criteria of p<0.10 (F=1.96, p=0.3755), suggesting that the relationship between religiosity and pregnancy acceptability was not moderated by gender. The Wald test for effect modification by generational status was statistically significant (F=2.09, p=0.0803), indicating that generational status moderated the relationship between religiosity and pregnancy acceptability. Specifically, the fully adjusted model with the interaction term for religiosity and generation status indicated that the positive relationship between religiosity and pregnancy acceptability was mostly driven by the third generation or higher respondents (high religiosity OR 5.99, 95% CI 2.84, 12.63; moderate religiosity OR 4.58, 95% CI 2.34, 8.93; Table 5). With the exception of moderately religious first-generation respondents, the odds of finding a pregnancy acceptable among high and moderately religious respondents in the first and second generation were not statistically significant.

Table 5.

Adjusted association between interaction of religiosity and generational status, and pregnancy acceptability, overall and among women

OR [95% CI]	Overall□	Women●
1st Generation
High religiosity	2.32 [0.70–7.67]	3.02 [0.48–19.10]
Moderate religiosity	3.26 [1.07–9.95]*	3.71 [0.65–21.30]
2nd Generation
High religiosity	1.50 [0.75–3.02]	0.91 [0.29–2.88]
Moderate religiosity	1.66 [0.90–3.09]	1.51 [0.53–4.29]
3rd Generation +
High religiosity	5.99 [2.84–12.63]**	11.03 [2.68–45.37]**
Moderate religiosity	4.58 [2.34–8.93]**	15.37 [4.10–57.53]**

Notes:

Reference group for all models is low religiosity.

Adjusted odds ratios are linear combinations of the main effect for each level of religiosity and the interaction term for religiosity and generational status.

^□Model adjusts for gender, number of children, employment, age, relationship status, and region of ancestry

^●Model adjusts for number of children, employment, age, relationship status, and region of ancestry. Stratified model results for men not shown, as there was no effect modification by generational status for men.

^**p<0.01;

^*p<0.05

As the three-way interaction between gender, generational status, and religiosity was statistically significant (F=2.01, p=0.0907), we then stratified the regression model by gender to determine whether the generational status by religiosity interaction effect varied for women versus men (Table 5). There was no effect modification by generational status for men (F=0.93, p=0.4465), but there was for women (F=2.27, p=0.0594). Among first- and second-generation women, being highly religious or moderately religious was not significantly associated with pregnancy acceptability. However, among third generation and above women, highly religious women had 11.03 (95% CI 2.68, 45.37) times the odds of finding a pregnancy acceptable, and moderately religious woman had 15.37 (95% CI 4.10, 57.53) times the odds of finding a pregnancy acceptable.

Discussion

In a US national sample of Latino/a young adults, we found a significant, positive association between religiosity and pregnancy acceptability even when accounting for other variables. Latino/a young adults who indicate that religion is important in their lives have higher odds of finding a pregnancy acceptable as compared to those for whom religion is not considered important. Among women, the association between religiosity and generational status differed by generational status. Specifically, the positive association between religiosity and generational status observed among women overall appeared to be primarily driven by third generation or above women, since religiosity was not associated with pregnancy acceptability among first and second-generation women in the gender-stratified model. These findings are in line with previous studies finding that religiosity and nativity may impact attitudes about pregnancy (Wilson 2008; Hartnett 2012; Sabagh and Lopez 1980). Among men, however, this association was not moderated by generational status. That is, regardless of which generation men are from, having high or moderate religiosity is associated with higher odds of pregnancy acceptability.

Although previous research has investigated the relationship between religiosity and pregnancy intention and/or attitudes among Mexican American women (Amaro 1988; Wilson 2008; Hartnett 2012; Chandra et al. 2005; Sabagh and Lopez 1980), this study is the first to assess the relationship between religiosity and pregnancy acceptability among men and women of diverse Latino/a ancestries. Although other studies have collected related data, the most relevant studies have relied on older data focusing only on Mexican American women and have used more traditional measures of pregnancy intention rather than measuring pregnancy acceptability (Amaro 1988; Wilson 2008; Hartnett 2012; Chandra et al. 2005; Sabagh and Lopez 1980). Therefore, it is difficult to infer why religiosity seems to be a salient factor for third generation and above women when considering pregnancy acceptability but less so for first and second-generation women in this diverse sample. The exiting literature does suggest a few reasons for the variation in pregnancy acceptability by generational status and gender. For example, it is possible that first and second-generation women are more attuned to factors that are pragmatic (such as financial concerns or lack of access to healthcare), as opposed to emotion- or value-oriented (such as religiosity), when considering pregnancy. Immigration is associated with many structural barriers, including language barriers, access to sufficient health insurance and healthcare, access to reliable transportation and adequate housing, employment barriers and financial constraints (Velasco-Mondragon et al. 2016; Dehlendorf et al. 2010; Rocca et al. 2010; Sullivan and Rehm 2005; Beck et al. 2017). It is possible that these factors carry the most importance when considering pregnancy among first and second-generation women. A qualitative study on pregnancy attitudes among Latino/a immigrant women by Sable and colleagues (2009) found that, contrary to the researchers’ expectations, pragmatic concerns (such as financially supporting and feeding children) were more salient than religion in their sample. In this study, participants suggested that such pragmatic concerns may not bother men as much when considering pregnancy, since in Latino/a cultures, the responsibility to care for children often falls on women’s shoulders (Sable et al. 2009).

Alternatively, it is possible that for first-generation immigrants, an unmeasured ‘traditional’ cultural Latino/a value (such as marianismo or fatalismo) is more important than religiosity when considering pregnancy acceptability (Bornstein 2002; Sabogal et al. 1987; Hartnett 2014; Lara-Cinisomo, Clark and Wood 2018; Valdez et al. 2011; Maternowska, Withers and Brindis 2014). For example, the value of familism includes elements such as the responsibility to reproduce as a central goal in life and the recognition of family as an extension of oneself (Bornstein 2002; Sabogal et al. 1987; Velasco-Mondragon et al. 2016; Maternowska, Withers and Brindis 2014). Although strong amongst most first-generation Latino/a immigrants, research has demonstrated that values such as familism tend to diminish in importance among second generation Latino/a Americans and vary by gender (Sabogal et al. 1987; Lorenzo-Blanco et al. 2012). As such cultural values were not measured in this study, it is unknown to what extent they may be associated with religiosity in the study population.

These findings should be interpreted with consideration of study limitations. First, the measure of religiosity lacked adequate nuance to unpack what ‘importance of religion’ means to this population. It is possible that the meaning of this question differs between first, second, and third generation women. Future research should employ mixed methods to better understand the meaning of religion in Latino/a young people’s lives as it relates to pregnancy. Second, though the sample size of the study is robust for regression analyses, cell sizes were somewhat small when assessing three-way interaction between religion, gender, and generational status. Therefore, some confidence intervals are wide and point estimates should be interpreted with caution. Third, because the data were collected using an online market research panel, the generalisability of the sample may be limited. Although efforts are made to construct the most representative sample as possible and YouGov is considered to provide more representative and accurate national surveys than the other leading online nonprobability panels (Kennedy et al. 2016; Rivers 2016), it is possible that the opt-in, computer-based nature of the panel led to differences between study participants and the US Latino/a population and may have excluded more immigrants who could not complete the survey in English.

The study also exhibits several strengths. First and foremost, there is no recent published research on this topic, and the study is novel in its inclusion of men’s perspectives on pregnancy acceptability across a range of relationship types. Additionally, the study population includes Latino/a young people from a variety of regions of ancestry and generations. Research on pregnancy intentions among the Latino/a population is scarce, and the few studies that do focus on this group mostly focus exclusively on Mexicans and are limited to first and second generations (Kim, Dagher and Chen 2016; Amaro 1988; Wilson 2008; Cuellar, Harris and Jasso 1980; Hartnett 2012; Chandra et al. 2005; Sabagh and Lopez 1980). Second, although some confidence intervals in the adjusted models with the interactions stratified by gender are wide, the sample did have sufficient power to detect two- and three-way interactions, which is somewhat rare when studying Latino/a populations. Lastly, the use of pregnancy acceptability as the outcome variable is an important advance in research on pregnancy intention; this construct may allow for a more nuanced view of the Latino/a population’s feelings about pregnancy in the context of social and cultural norms.

More research is needed to replicate and better understand the observed association between religiosity and pregnancy acceptability and why this seems to differ by generational status and gender. Future research should use mixed methods to focus specifically on better understanding the role of religion and how importance of religion is practised and enacted in the lives of Latino/a young adults from different generations. Qualitative analysis may also help illuminate the mechanisms through which generational status impacts on individuals’ pregnancy planning and attitudes by exploring how cultural rituals and values, immigration experiences and/or pragmatic concerns associated with immigration and/or other structural realities affect young peoples’ pregnancy attitudes and expectations. Larger sample sizes of both men and women from a variety of regions of ancestry are also needed to further investigate differences between various subgroups and generations of the US Latino/a population.

Implications

The association between religiosity and pregnancy acceptability among the Latino/a population has implications for the healthcare system. As the proportion of Latino/a users within the healthcare system has grown, disparities in access and care remain, including within sexual, reproductive, and maternal healthcare (Velasco-Mondragon et al. 2016; Chandra et al. 2005; Dehlendorf et al. 2010). Immigrants face complex challenges in navigating the health care system, including language barriers, access to health insurance, restricted access to contraception, and cultural shifts in the values of family and parenting (Dehlendorf et al. 2010; Rocca et al. 2010). There is also evidence that providers perceive and treat Latino/a family planning patients differently (Dehlendorf et al. 2010), perhaps in part as a result of stereotypes that Mexican American women are ‘devoutly Catholic and subservient to men—super-mothers who value and accept unplanned and continuous reproduction’ (Amaro 1988, 7). Compared to White women, Latina women rate their family planning appointments less positively and report more pressure to limit family size (Becker and Tsui 2008; Downing, LaVeist and Bullock 2007; Dehlendorf et al. 2010). Researchers have suggested that in order to improve services for Latino/a patients, healthcare providers must better understand and account for how social and cultural factors may impact family planning strategies and attitudes about pregnancy (Dehlendorf et al. 2010; Maternowska, Withers and Brindis 2014; Shaffer 2002; van Dijk et al. 2013). This study provides a window into how religiosity, generational status and gender might matter for Latino/as.

Particularly relevant is our finding that religiosity is significantly associated with pregnancy acceptability in this population; providers should consider how to better address family planning needs in this context. Although there is limited research exploring the needs of religious patients, some findings have indicated that religious hospital inpatients desire more conversations with their healthcare providers about their religious beliefs, and that patients who were able to discuss their spiritual concerns indicated greater satisfaction with the care they received (Williams et al. 2011). This indicates that family planning providers may also want to invite Latino/a patients to discuss their religious and spiritual concerns when considering pregnancy planning. Some research has also demonstrated positive results integrating faith-based organisations into evidence-based family planning education and services in low-income countries (Barot 2013). Such efforts may be beneficial for US Latino/a populations as well, in the light of our findings that religion is an important factor when considering pregnancy acceptability.