HISPANIC FERTILITY, RELIGION AND RELIGIOUSNESS IN THE U.S

Introduction

About one-sixth of the U.S. population is now of Hispanic origin. This population is growing rapidly, fueled both by immigration and higher fertility. The total fertility rate for Hispanics was estimated for 2006 at 3.0, significantly higher than that for either non-Hispanic whites (1.9) or for non-Hispanic blacks (2.1) (CDC, NCHS, 2007). The rates for Hispanics are higher at every age, especially for teenagers, for whom it is three times higher than for non-Hispanic whites. Within the Hispanic population, the highest fertility is for those of Mexican origin and for those born abroad. According to a recent population projection (Passel and Cohn 2008), the Hispanic population in the United States is projected to triple in numbers by 2050 and increase from 14 to 29 percent of the total population.

The primary focus of interest here is on the role that religion and religiousness play in the higher fertility of Hispanics. Several sources of data have been used: the 2002 and the 1995 National Surveys of Family Growth¹ (the two surveys show a very similar picture), the four General Social Surveys between 2000 and 2006, and the 2006 survey of Hispanics by the Pew Forum on Religion and Public Life and the Pew Hispanic Center. The age range for women is 18–44.

The literature on religion and fertility proposes several pathways by which religion may influence fertility. Many authors use Goldscheider’s typology of approaches to understanding these mechanisms (Kertzer 2006; Lehrer 2008; McQuillian 2004; Mosher, Johnson, and Horn 1986). First, the “characteristics approach” attempts to explain observed correlations between religion and fertility by controlling for socio-economic characteristics; this approach tests whether associated differences in fertility across religious groups can be explained by differences in characteristics such as income and education. The other two approaches recognize correlations between socio-economic characteristics and fertility, but argue that these relationships may not account for observed differences in fertility. For example, scholars have found that higher fertility among Mexican-American women, compared to non-Hispanic white women, is only partly accounted for by education and age (Bean and Tienda 1987:230).

The “minority group status” approach claims that unexplained group differences in fertility may be explained by the social status of different groups: members of minority communities – in the absence or pronatalist rules and norms – may attempt to overcome barriers to economic and social integration by reducing fertility (Goldscheider and Uhlenberg 1969; Mosher et al. 1986: 375–76; McQuillan 2004:26–7). The third approach, called the “particularized theology” approach, seeks explanations for these unexplained differences in fertility in the content of religious teachings, both specific rules about practices (e.g., contraception, abortion), and more general teachings about values and norms (e.g., gender roles, importance of marriage).

Recently, scholars have elaborated upon Goldscheider’s framework, particularly his focus on religious teachings and norms: McQuillian (2004) highlights the institutional and social context of religious teachings, arguing that institutions that can transmit and enforce the relevant norms are necessary for religious teachings to affect fertility, and that their effects will vary with the intensity of attachment to religious identity. Hayford and Morgan (2008) argue that strength of religious identity (religiousness or religiosity) should be a main focus of research on the relationship between religion and fertility, as it is more relevant to fertility in the U.S. today than are specific norms or institutional enforcement. Voas also emphasizes the importance of religiousness for population studies, noting that it may be more important than religious affiliation for some outcomes of interest (2007:1167).

The relationship between fertility and religiousness has been examined in several studies. Variables commonly used to measure religiousness include self-reports of participation in religious activities (e.g., churchgoing, prayer) and strength of religious beliefs. One early study showed that daily religious activity was associated with higher fertility among conservative Protestants who attend church less than weekly (Marcum 1981). More recent studies have also found significant associations between religiousness and fertility: One finds that importance of religious beliefs is a significant predictor of children ever born, although frequency of attending religious services is not. Controlling for religious denomination does not change these associations, and the only significant difference by denomination appears for non-Christians (Zhang 2008). Hayford and Morgan (2008) find that respondents who say that religion is “very” or “somewhat” important in their everyday lives have significantly higher intended fertility than those who report no religion and those who say it is “not important.” These relationships may vary over time, however: as practicing Catholics in Spain became less numerous than non-practicing Catholics between 1985 and 1999, the difference in family size between the two groups went from non-significant to significant (Adsera 2006).

Empirical findings from other studies suggest that the relationship between religion and fertility may vary by ethnicity. While one study found fertility of Catholic and non-Catholic American women to be converging in the post-baby boom era among white, married women (Westoff and Jones 1979), a subsequent study found that including Spanish-speaking (non-English-speaking) women in the sample produced larger differences between Catholic and non-Catholic respondents, although they also found a trend toward convergence (Mosher and Hendershot 1984a). A later study found no difference in fertility between Catholics and non-Catholics among non-Hispanics, but a large difference between Catholic and non-Catholic Hispanics (Mosher, Johnson and Horn 1986). Another analysis, which excluded Hispanic women, found a negative relationship between education and fertility for white Catholic women, and a positive relationship for black Catholic women (Mosher and Hendershot 1984b). Evidence on the relationships among religiousness, ethnicity, and fertility is scarce: Zhang (2008) found that Hispanic ethnicity was a significant predictor of children ever born, controlling for denomination and religiousness, but no interactions were tested. Nevertheless, these findings indicate that a closer examination of ethnic differences and the relationships between religiousness and fertility is warranted.

Religion and Religiousness

Most Hispanics (66%) report their religion as Catholic, 21% as Protestant, and 11% report no religious affiliation (Table 1). The influx of Hispanics in this country has evidently had a major rejuvenating impact on the Catholic Church (Reiff 2006). The Church has apparently adjusted to the fact that many Hispanics are expressing a different form of Christianity, increasingly associated with Pentecostal and charismatic movements (Pew 2006). About 22% of Hispanics identify themselves as “born again” or “evangelical or charismatic,” with 31% of non-Hispanics in this category.

Table 1.

Mean number of children ever born for Hispanic and non-Hispanic women 18 – 44, by religion

	Mean Children Ever Born		Total Number Expected		Percent Distribution
	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic
All women	1.75	1.35	2.60	2.23	100	100
Catholic	1.73	1.37	2.62	2.26	66.3	20.2
Born again Christian	1.89	2.05	2.84	2.61	6.1	1.5
Charismatic or Evangelical	1.49	1.12	2.42	2.35	2.2	0.5
Other Catholic	1.72	1.31	2.60	2.21	58.0	18.2
Protestant	1.81	1.45	2.66	2.33	21.3	58.0
Born again Christian	2.03	1.52	2.72	2.61	10.6	27.3
Charismatic or Evangelical	1.39	1.43	2.75	2.35	3.0	1.8
Other Protestant	1.70	1.38	2.55	2.21	7.7	28.8
Other religion	1.37	1.06	2.46	2.12	1.4	6.7
No religion	1.84	1.03	2.48	1.83	11.0	15.1
Number of women					1,438	5,367

Source: NSFG 2002

Latinos tend to be more religious than non-Hispanics (Figure 1 and Table 2) with 38% reporting church attendance at least weekly compared with 31% of non-Hispanics; 61% of Hispanics say that religion is very important in their lives compared with 42% of non-Hispanics. Very similar differences are reported in the Pew 2006 survey. One striking difference between Hispanics and non-Hispanics measured in the Pew survey is the greater Hispanic belief in God as an active force in their lives. Some 52% of Christian Hispanics believe that “Jesus will return to earth in my lifetime” in contrast to 34% of Christian non-Hispanics (among Catholics, the contrast is even greater: 51% among Hispanics and 22% among non-Hispanics). Belief that the bible is the literal word of God also shows a pronounced difference, with 50% of all Hispanics in accord, compared with 35% of all non-Hispanics.

Figure 1.

Differences in religiousness.

Table 2.

Mean number of children ever born and mean total number expected for Catholic Hispanics and non-Hispanics by measures of religiousness.

	Mean Number of Children Ever Born		Mean Total Number Expected		Percent Distribution
	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic
Frequency of Church Attendance
At least weekly	1.86	1.70	2.74	2.54	37.7	32.6
1–3 times a month	1.79	1.49	2.64	2.38	21.2	19.7
Less than monthly	1.60	1.13	2.56	2.02	23.7	36.4
Never	1.55	1.03	2.41	2.01	17.4	11.3
				Number	950	1101
Importance of Religion
Very important	1.85	1.65	2.71	2.44	60.5	44.9
Rather important	1.55	1.21	2.50	2.20	35.9	45.8
Not important	1.40	0.84	2.29	1.68	3.7	9.4
				Number	954	1100
Religiousness Index*
Very religious	1.87	1.73	2.73	2.53	43.9	35.9
Religious	1.76	1.36	2.65	2.27	31.6	25.3
Not religious	1.42	1.05	2.38	2.00	24.5	38.7
				Number	932	1099

^*A combination of attendance and importance items. See text for details.

Source: NSFG 2002

The Pew report concludes that: “Overall, the findings thus suggest that Hispanic Catholics practice a distinctive form of Catholicism, one that incorporates many of the beliefs and behaviors most commonly associated with Pentecostal or renewalist Christianity, while at the same time upholding the main features of traditional Catholic teaching.” (p.30)

The Pew report covers religious experience of Hispanics in great detail but unfortunately includes no information on fertility. In contrast, the 2002 National Survey of Family Growth (NSFG) includes extensive coverage of reproductive behavior but little on religion and religiousness.

Religion and Fertility

The number of children ever born is greater among Hispanics than non-Hispanics in every denominational category except among “born again” Catholics, which is a small group (Table 1). The total expected number of children is higher among Hispanics in each denomination without exception. Overall, there is no difference in fertility between Catholics and Protestants but Hispanic fertility is higher than non-Hispanic fertility in virtually every religious category. The ethnic differences are statistically significant for all of the major religion categories but not for most of the subdivisions of the fundamentalist groups where the number of observations is small. In effect, holding religion constant, including women with no religion (interestingly, among whom the difference is the largest) does not erase the fertility difference between Hispanic and non-Hispanic women. The analysis of the General Social Survey data for 2000–2006 shows the same basic result (not shown here).

Religiousness and Fertility

In order to refine the comparisons, most of the analysis is confined to Catholic women, who are the largest subcategory of Hispanics. Among both Hispanic and particularly among non-Hispanic women, religiousness as measured by the frequency of church attendance and by the perceived importance of religion in daily life is directly associated with the number of children ever born and with the total number of children expected (Table 2). Again, the greater fertility of Hispanics appears at each category of religiousness at statistically significant levels but it is relevant that Hispanics are more religious especially in connection with the perceived importance of religion measure. A three-category summary index of religiousness was constructed in which women who both attend church more than once a month and who feel that religion is very important to them are at the high end of the scale while women who attend less frequently and who do not feel that religion is very important are at the low end of the scale. The two other mixed combinations are in the middle. The two components have a n overall correlation of .65. It is noteworthy that all of the comparisons in Table 2 show a greater range of fertility differences across the religious categories for non-Hispanic Catholics.

It is interesting that the ethnic difference in fertility is minimal at the most religious end of the scale but increases as religiousness declines. Education is a major part of the explanation. If we confine the comparisons to women who have attended college (one-third of Hispanic women and two-thirds of non-Hispanic women), the ethnic difference vanishes while a positive association between religiousness and fertility is the same for both ethnic groups (Table 3). The only ethnic difference in fertility that remains is among the less religious and less educated women. Although higher education does not eliminate the influence of religiousness on fertility, it does diminish most of the ethnic effect.

Table 3.

The effect of education on the ethnic difference in number of children ever born among Catholic women 18 – 44, by religiousness

	Some College		No College
	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic
Very religious	1.55	1.53	2.02	2.11
Religious	1.38	1.32	1.94	1.67
Not religious	0.80	0.86	1.83	1.40
Number	340	731	615	370

These associations are examined simultaneously along with whether the woman is foreign-born with logistic regression (Table 4). The dependent variable is whether the woman has had or expects to have two or more children. This cutting point was chosen because of the demographic significance of two births for population replacement. Comparing Hispanic and non-Hispanic women among Catholics shows that the index of religiousness has a significant predictive value for fertility, but the effect is somewhat greater for non-Hispanic women. If the woman is foreign-born rather than native-born, the odds of Hispanic women having two or more children are greater. (The mean number of children ever born is 1.89 for the foreign-born Hispanics and 1.53 for the native-born). Higher education has a significant negative impact on fertility among both Hispanic and non-Hispanic women.

Table 4.

Odds ratios of having two or more children and of expecting a total of more than two children by Hispanic origin for Catholic women 18 – 44.

Covariate	Two or More CEB		More than Two Expected Total
	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic
Religiousness	1.28	1.64	(1.10)	1.34
Foreign-born	1.37	(1.12)	(1.15)	(1.07)
More than high school	0.46	0.65	0.41	0.82
Number of women	948	1097	948	1097
Chi squared	56	53	49	17
R2	0.04	0.04	0.04	0.03

( ) not significant at the .05 level

Source: NSFG 2002

Analysis of the 2002–2006 General Social Survey data shows essentially the same results as the preceding analysis based on the 2002 NSFG. Hispanic fertility is higher than that of non-Hispanics even among Catholics. Women of Mexican origin have the highest fertility. The number of children ever born increases both with the frequency of church attendance and with the frequency of prayer. A multivariate analysis shows that religiousness retains its association with fertility in the presence of numerous controls.

Religiousness or Economic Insecurity?

In studying the influence of religiousness on childbearing, it is of interest to determine whether the observed direct association is a function of economic insecurity. The general notion is that economic insecurity can lead to a greater reliance on religious values and thus to higher fertility or, conversely, that women who feel secure economically rely less on religion and have fewer children. This general question assumes greater significance in the context of Hispanic fertility since estimates are that 37 percent of Hispanic women are classified in the poverty category compared with 16 percent of non-Hispanic women.

Is there a relationship between religiousness and poverty? The short answer is yes, but it is not very strong: about two-thirds of poor women respond that religion is very important in their daily life, compared with half of the non-poor. This applies both to Hispanic and non-Hispanic women.

Is there an association between poverty and fertility? The answer is clearly yes and the association is quite strong. Poverty and relative income (in six categories) was cross-tabulated with fertility for Hispanics and non-Hispanics for both women 18–44 and 35–44 (not shown). The latter group is analyzed separately because these women are nearer the end of their childbearing years, so the figures come closer to completed childbearing. In both age categories, fertility increases with poverty for both groups although Hispanic fertility remains higher at most income levels. There is some ambiguity in the direction of causation since poor people have more children but people with more children are poorer. Since the definition of poverty takes family size into account, some of the association is built into the measure. However, a substitution of total family income shows the same association with fertility as the poverty measure.

The key question is whether the association between religiousness and fertility is a function of their joint association with income and whether the same structure of association obtains for both Hispanics and non-Hispanics. A detailed tabulation showing the mean number of children ever born (not shown here) indicates that at each of the six intervals on the economic scale those women who feel that religion is important have higher fertility as measured both by the number of children ever born and the total number of children expected. Nonetheless, there is progressively higher fertility as poverty increases both for those who feel that religion is important and those who think that it is unimportant.

Another approach is through multivariate analysis which examines the simultaneous associations of religiousness and poverty with, as in Table 4, fertility dichotomized as having had less than two or two or more births and total expected fertility divided into women who expect two or more children and those who expect fewer. This analysis, which includes the index of religiousness and income as well as education and nativity (Table 5), shows that both religiousness and income significantly affect both measures of fertility. To rephrase this, the influence of religiousness on fertility is not eliminated in the presence of poverty or income. The measure of income in this analysis is dichotomized into the bottom and top three categories, but contrasting women at the extreme poverty level with all others does not alter the general results. Education consistently shows the familiar negative relationship with fertility, except for total births expected among non-Hispanics.

Table 5.

Odds ratios of having had two or more children ever born or of expecting a total of more than two children, for Catholic Hispanic and non-Hispanic women 18 – 44.

	Two or More CEB		More than Two Expected
Covariate	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic
Religiousness	1.26	1.64	(1.08)	1.33
Foreign-born	(1.15)	(1.06)	(1.01)	(1.01)
More than high school	0.57	0.72	0.48	(0.93)
Income	0.31	0.71	0.41	0.68
Number of women	948	1097	948	1097
Chi squared	96	60	71	25
R2	0.07	0.04	0.05	0.02
Full Model
	Hispanic	Non-Hispanic	Hispanic	Non-Hispanic
Religiousness	(0.95)	1.42	(1.08)	1.35
Foreign-born	(0.76)	0.62	(0.98)	(0.92)
More than high school	0.34	0.68	0.46	(0.93)
Above 300% of poverty	0.16	0.39	0.40	0.69
Ever married	3.88	5.73	1.52	1.88
White - other race	(0.95)	0.62	(0.95)	(0.87)
Age	2.28	1.74	1.23	(0.94)
Age squared	0.99	0.99	0.99	(0.83)
Number of women	948	1097	948	1097
Chi squared	393	429	89	43
R2	0.30	0.30	0.07	0.03

( ) not significant at the .05 level

Source: NSFG 2002

A fuller model is presented in the lower half of Table 5 with marital status, race and detailed age included. This eliminates the significant co-variation of religiousness with parity among Hispanics but not among non-Hispanics. Income is a strong covariate among both ethnic groups. Being foreign-born is now irrelevant (except, anomalously, among non-Hispanics with only 5 percent in this category). Not surprisingly, being married has a strong effect on fertility for both groups. The difference between whites and non-whites is not very relevant here partly because they are a small fraction of the Hispanic Catholics and have not been differentiated in the analysis.

Summary Multivariate Analysis

The detailed analyses after Table 1 have focused on Catholics. In Table 6, we return to all Hispanic and non-Hispanic women to summarize the overall associations of fertility with religion and religiousness along with the array of covariates. Among Hispanics, neither religious affiliation nor religiousness is significantly correlated with fertility when considered with the other covariates. Among non-Hispanics however, both Catholic and Protestant women and more religious women in general, have higher fertility. In both ethnic categories, education and income have negative effects on fertility and marriage shows very strong positive associations. Lower fertility is related to being white rather than nonwhite only among non-Hispanics.

Table 6.

Odds ratios of having had two or more children or of expecting a total of more than two children, for all Hispanic and non-Hispanic women 18 – 44.

Covariates	Two or more CEB		More than Two Expected
	Hispanic	non-Hispanic	Hispanic	non-Hispanic
No religion	1.00	1.00	1.00	1.00
Catholic	(0.87)	1.46	(1.32)	1.44
Protestant	(1.19)	1.34	(1.43)	1.24
Other	(0.55)	(1.08)	(1.14)	(1.07)
Religiousness	(1.01)	1.20	(1.06)	1.29
More than high school	0.38	0.49	0.48	(0.90)
Above 300% poverty	0.20	0.33	0.41	0.48
Ever married	3.35	4.64	1.60	1.98
White	(1.00)	0.58	(0.93)	0.79
Foreign born	(0.89)	0.54	(0.96)	(0.90)
Age	2.17	1.91	1.17	(1.00)
Age squared	0.99	0.99	0.98	(1.00)
Number of women	1428	5455	1428	5455
Chi squared	553	1852	127	385
R2	0.28	0.25	0.06	0.06

In sum, religion (Catholic or Protestant) and religiousness seem to influence fertility only among non-Hispanics.

Age at First Birth

Another measure of reproductive behavior that is important for both total fertility a nd for life chances in general is the age at which women begin childbearing. Hispanic women clearly begin at an earlier age with 42 percent having their first birth before age 22 compared with 19 percent of non-Hispanics. Religiousness, however, appears to have only a slight effect of encouraging earlier childbearing.

Summary and Discussion

This paper is the third part of a larger study of the effects on fertility of religion and religiousness. The first paper (Frejka and Westoff 2007) focused on the differences in religiousness between American and European women and their implications for the higher U.S. fertility. The second paper (Westoff and Frejka) looked at the levels of Muslim fertility in Europe and the extent to which differences in religion and religiousness explain the higher fertility of Muslims. The present effort is an extension of this same question of how much of the higher fertility of Hispanics in the U.S. can be explained by their Catholicism and the possibility of their greater religiousness.

The considerable influx of Hispanics in the U.S. has had a major impact on the Catholic Church in this country which has been recently commented on in the popular press. The analysis in this paper has shown that indeed Hispanic women (mostly of Mexican origin) are more religious than non-Hispanics especially in terms of the perceived importance of religion in their personal lives. It is also clear that religiousness is associated with higher fertility among both Hispanics and non-Hispanics, but that Hispanic fertility is higher regardless of religion or religiousness. However, higher education dramatically reduces this ethnic difference. Fertility is also negatively associated with income in both groups but has more of an effect among Hispanics because of their greater concentration at the poverty level.

In sum, American Hispanics are somewhat more religious than non-Hispanics but this difference explains only some of the higher fertility of Hispanics. The number of children ever born to Catholic Hispanic women is 28 percent higher than the fertility of comparable non-Hispanics. If Hispanic women attached the same (lesser) importance to religion, this difference would still be greater, by about 22%, among Hispanic women. More of the fertility difference is associated with the higher rate of poverty among Hispanics and their higher proportion of unintended births. For example, if all Hispanic women had the same income-poverty distribution as all non-Hispanic women, their fertility would be only 6 percent higher than that of non-Hispanics rather than the observed 30 percent higher.