Parental Income and the Sexual Behavior of Their Adult Children: A Trivers–Willard Perspective

Abstract

Parental income is negatively and linearly related to the digit ratio (2D:4D; a proxy for prenatal sex steroids) of their children. Children of parents with high income are thought to be exposed to higher prenatal testosterone and develop lower 2D:4D. It is further hypothesized that 2D:4D relates to sexual orientation, although it is unclear whether the association is linear or curvilinear. Here, we consider patterns of parental income and its association with the sexual behavior of their adult children in a large online study (the BBC internet study). There were curvilinear relationships with parental income in male and female children. The highest frequencies of homosexuality and bisexuality were found in the lowest income group (bottom 25% of the population), the lowest frequencies in the income group representing the upper 50% of the population, and intermediate values in the other groups (low 50% and top 25% of the population). Parental income showed a U-shaped association with scores for same-sex attraction and an inverted U-shaped association with opposite-sex attraction. Thus, for the first time, we show that same-sex attraction is related to parental income. The curvilinear relationship between parental income and sexual behavior in their adult children may result from an association between very high fetal estrogen or testosterone and attraction to partners of the same sex. Among non-heterosexuals, and in both sexes, very high fetal estrogen may be associated with femme or submissive sexual roles, and very high fetal testosterone with butch and assertive sexual roles.

Introduction

Parental income is related to the relative lengths of the 2nd and 4th digits (i.e., digit ratio or 2D:4D, a putative measure of prenatal sex steroids; Manning et al., 1998; Manning, 2002) in their children. Low parental income is associated with high 2D:4D (indicating low testosterone and high estrogen in the fetus) and high parental income is linked to low 2D:4D (indicating high testosterone and low estrogen in the fetus) in their children (Manning et al., 2022). The relationship between parental income and their children's 2D:4D opens up the possibility that disparities in parental income influence patterns of other traits, previously thought to be unrelated to income but for which there is evidence of links with 2D:4D. Sexual orientation is one such trait and much research on the topic has been conducted on same-sex attraction (i.e., male androphilia or male sexual attraction to men and female gynephilia or female sexual attraction to women). Here, we consider, for the first time, the relationships between parental income and the sexual orientation of their adult children. In addition, we discuss our findings in light of currently unresolved issues concerning the relationships between sexual orientation and 2D:4D.

The effect of parental income on the 2D:4D of their children may be viewed through the lens of the “Trivers–Willard” hypothesis (Trivers & Willard, 1973; see also Manning et al., 2022). This model assumes that sons from high-resource mothers (good condition) have higher reproductive success than daughters from high-resource mothers, while daughters from low-resource mothers (poor condition) will be more reproductively successful than sons from low-resource mothers. In such a model, parental income is a proxy for the maternal resource, which influences maternal condition. There is evidence that income is positively related to health across populations (Babones, 2008) and within populations (Pickett & Wilkinson, 2015). Therefore, parents with high incomes are expected to invest in sons, while those with low incomes will invest in daughters. Focussing on mothers, there is no evidence that maternal sex steroids are adjusted for the sex of the fetus. If this is the case, high-condition mothers with high income will masculinize their children in utero, while low-condition mothers with low income will feminize their children. Thus, mothers from high-parental-income households are expected to have low 2D:4D children (boys and girls), while mothers from low-parental-income households will have high 2D:4D children (boys and girls: Manning et al., 2022).

The suggestion that 2D:4D is negatively correlated with relative concentrations of prenatal testosterone and estrogen (Manning et al., 1998) was followed by two reports of links between 2D:4D and sexual orientation (Robinson & Manning, 2000, men only; Williams et al., 2000, men and women). A straightforward model of 2D:4D and sexual orientation would be to assume a linear relationship, such that male androphilia is linked to high (feminized) 2D:4D and female gynephilia is associated with low (masculinized) 2D:4D. Surprisingly, Robinson and Manning (2000) reported that male androphilia is associated with low (masculinized) not high 2D:4D. Williams et al. (2000) reported no association between male androphilia and 2D:4D and a significant (and theory-consistent) relationship between female gynephilia and low (masculinized) 2D:4D.

Mixed reports have subsequently been presented for male androphilia. For example, national differences in mean 2D:4D between homosexual and heterosexual men have been noted with lower 2D:4D in the former compared to the latter in some UK reports and a reverse tendency in US studies (Manning & Robinson, 2003, McFadden et al., 2005). A meta-analysis found that homosexual and heterosexual men did not differ significantly in their 2D:4D (Grimbos et al., 2010). However, in contrast, the two largest studies (Lippa, 2003; Manning et al., 2007) reported a higher 2D:4D in the former compared to the latter. Furthermore, recent studies have reported higher 2D:4D in androphilic compared to gynephilic men (Siegmann et al., 2020; Skorska et al., 2021; Xu et al., 2019). The evidence for female gynephilia and 2D:4D has been more consistent in that several studies have replicated the theory-consistent findings of Williams et al. (2000). The meta-analysis of Grimbos et al. (2010) confirmed this low 2D:4D pattern in gynophilic females compared to androphilic women. The finding was robust to the inclusion of the two largest studies (Lippa, 2003; Manning et al., 2007), which, in contrast, reported no difference in 2D:4D between gynephilic and androphilic women. Two more recent studies did not replicate the female sexual orientation difference in 2D:4D (Siegmann et al., 2020; Skorska et al., 2021).

The mixed results obtained in studies of 2D:4D and male and female sexual orientation are unlikely to be explained by the effects of measurement methodology (e.g., direct versus indirect measurement of 2D:4D; Manning et al., 2005) or by ethnic variation in 2D:4D (e.g., Manning, 2002). Hence, the assumption of a linear relationship, such that male androphilia is linked to high 2D:4D and female gynephilia is associated with low 2D:4D, may be incorrect. Male androphiles, for example, may be feminized in their physical and behavioral traits, but some are androgenized (Skorska & Bogaert, 2017). Therefore, there could be different etiologies in the phenotypes of both male androphilia and female gynephilia. That is, the former phenotype may be linked to very low and very high 2D:4D, which could map onto insertive and receptive sexual behaviors among homosexual men (Swift-Gallant et al., 2021, see review by VanderLaan et al., 2022). The latter phenotype may also be associated with low and high 2D:4D, which could be linked to butch and femme phenotypes, respectively (reviewed by VanderLaan et al., 2022).

The current study examines the relationship between parental income and the sexual behavior of their adult children in a large online survey (the BBC internet study; Reimers, 2007). We assumed that parental income would be negatively correlated with the 2D:4D of their male and female children. Therefore, we predicted the following: (i) if 2D:4D is related to sexual orientation, there will be differences in the frequencies of heterosexuals, homosexuals, and bisexuals and scores for attraction to same-sex and opposite-sex individuals across parental income, and (ii) if 2D:4D is related to sexual orientation in a linear fashion then, for men, the highest frequency of homosexuality and bisexuality and the highest same-sex attraction scores will be in adult children of parents with the lowest incomes. For women, the highest frequency of homosexuality and bisexuality and the highest same-sex attraction scores will be found in adult children of parents with the highest incomes, and (iii) if 2D:4D is related to sexual orientation in a curvilinear fashion, such that low and high 2D:4D is associated with non-heterosexuality, then the highest frequency of homosexuality and bisexuality and the highest same-sex attraction scores will be in adult male and female children of both lower and higher parental incomes.

Methods

The BBC internet study was a multiethnic and multinational survey, hosted by the BBC Science and Nature website in July 2005 (Reimers, 2007). It comprised around 200 questions and included information on demographics, personality, sexual behavior, and physical characteristics, such as digit length from which 2D:4D could be calculated. A sample of n = 255,116 participants completed all study tasks. The study was conducted in accordance with the Declaration of Helsinki on Research Involving Human Subjects. Participation was anonymous and voluntary. Informed consent was requested (online) from all participants. All participants had the right to comply or refuse participation. Further details about the ethical and methodological considerations in the research design of the study are provided in Reimers (2007).

Participants reported their ethnicity (Asian/Asian British, Black/Black British, Black other, Chinese, Middle/Near Eastern, Mixed Ethnic, White), age (integer 0–99 years), gender (male or female), and where they lived (the United Kingdom, then 240 other countries). The predominant ethnicity was White (reported by 84.1% of participants), and the most commonly represented nationalities were the United Kingdom (46.9%), and the United States (27.7%), with 13 other nations represented by n > 1000 participants.

We were concerned here with responses to three items. Parental Income: the item was phrased “What best describes your parents’ income [while growing up]?,” with response options ‘much lower than others (bottom 25% of the population)’ (=income group I), ‘slightly lower than others (low 50% of the population)’ (=income group II), ‘slightly higher than others (upper 50% of the population)’ (=income group III), and ‘much higher than others (top 25% of the population)’ (=income group IV).” It is to be noted that the phrasing of this item meant that it is a measure of relative (not absolute) parental income and as such may apply across different nations. Sexual Orientation: “What is your sexual orientation?” with response options heterosexual (straight), homosexual (gay/lesbian), bisexual). ¹ Sexual Attraction Towards Others: “How sexually attracted are you to …?,” response options, men (seven-point scale, 1 = not at all to 7 = very), women (seven-point scale, 1 = not at all to 7 = very).

Participants self-measured 2D and 4D of their right and left hands (Manning et al., 1998). A diagram of the hand was provided, and they were instructed to measure their fingers on the ventral side of the digit from the fingertip to the most proximal crease with a conventional ruler. Measurements were reported to the nearest millimeter using dropdown menus. The 2D:4D was calculated by dividing the 2D by 4D digit lengths. Analyses were restricted to participants 18 years and older. As in earlier reports, the tails of the 2D:4D distributions were removed by considering right and left 2D:4D within the range of ≥0.80 to ≤1.20 (Manning et al., 2007, 2022).

Results

Parental Income and Sexual Orientation of Their Children

Male Participants

A total of 116,904 men answered the items on their sexual orientation and the income of their parents. There were 105,548 (90.3%) heterosexuals, 6216 (5.3%) homosexuals, and 5140 (4.4%) bisexuals. Table 1 reports male participants by sexual orientation and parental income group.

Table 1.

Counts and Frequencies of Male and Female Participants’ Sexual Orientation from Parents of Income Groups I–IV (I = Bottom 25% of the Population, II = Low 50% of the Population; 3 = Upper 50% of the Population, IV = Top 25% of the Population).

	Parental Income
	Group I	Group II	Group III	Group IV
Heterosexual
Males	10,686 (88.8%)	36,778 (89.8%)	49,332 (90.8%)	8762 (90.6%)
Females	8021 (87.6%)	31,991 (89.9%)	41,905 (91.2%)	4705 (89.6%)
Homosexual
Males	731 (6.1%)	2306 (5.6%)	2669 (4.9%)	510 (5.3%)
Females	306 (3.3%)	1068 (3.0%)	1176 (2.6%)	154 (2.9%)
Bisexual
Males	615 (5.1%)	1865 (4.5%)	2270 (4.2%)	399 (4.1%)
Females	834 (9.1%)	2522 (7.1%)	2895 (6.3%)	395 (7.5%)
Total
Males	12,032	40,940	54,261	9671
Females	9161	45,581	45,976	5254

Parental income was significantly associated with the sexual orientation of their adult children (χ²₍₆₎ = 66.02, p < .0001). Percentages of heterosexuality, homosexuality, and bisexuality across the parental income groups are shown in Figure 1(a). Percent heterosexuality was lowest and homosexuality and bisexuality highest in the children of income group I parents. Children's heterosexuality increased linearly across the parental income groups and was most frequent among the children of income group III. The frequency of homosexuality decreased linearly with increases in parental income and was at its lowest among the children of parental income group III. Income group IV showed intermediate levels of heterosexuality and the percentage of homosexuality was higher than that of income group III. Thus, the overall impression was of a J-shaped curve for the relationship between parental income and homosexuality in their male children.

Figure 1.

Frequencies of male (a) and female (b) sexual orientation of adult children plotted against parental income groups (I = bottom 25% of the population, II = low 50% of the population; 3 = upper 50% of the population, IV = top 25% of the population).

Female Participants

A total of 95,972 women reported their sexual orientation and the income of their parents. Heterosexuality was reported by 86,622 (90.3%) women and (in contrast to men) homosexuality was less frequent (n = 2704, 2.8%) than bisexuality (n = 6646, 6.9%). Table 1 reports the counts and frequencies of female participants by sexual orientation and parental income group.

There was an association between the income of parents and the sexual orientation of their adult children (χ²₍₆₎ = 129.63, p < .0001). The relationship between the frequency of children's heterosexuality across parental income groups was similar to that for men, but with a more marked downturn from income group III to income group IV. The associations between parental income and homosexuality and bisexuality were U-shaped (Figure 1(b)).

Parental Income and Their Children's Reported Sexual Attraction to Men and Women

Reported scores for sexual attraction toward others mapped onto the sexual orientation categories. Thus separate ANOVA's for men and women with sexual attraction scores as the dependent variable and sexual orientation as the independent variable gave the following: (i) men attracted to women, heterosexuals n = 105,674, mean 6.69 (SD .76); homosexuals n = 6164, mean 1.98 (1.25); bisexuals n = 5174, mean 5.77 (1.41) (F_2,117009 = 95,770.74, p < .0001), (ii) men attracted to men, heterosexuals n = 105,166, mean 1.45 (SD .89); homosexuals n = 6261, mean 6.68 (.93); bisexuals n = 5168, mean 4.36 (1.71) (F_2,116592 = 110,361.57, p < .0001), (iii) women attracted to men, heterosexuals n = 86,787, mean 6.49 (SD .97); homosexuals n = 2705, mean 2.20 (1.35); bisexuals n = 6704, mean 5.71 (1.43) (F_2,96193 = 26,327.74, p < .0001), (iv) women attracted to women, heterosexuals n = 84,982, mean 2.04 (SD 1.24); homosexuals n = 2736, mean 6.45 (1.15); bisexuals n = 6710, mean 5.13 (1.34) (F_2,94425 = 34,388.51, p < .0001).

Male Participants

A sample of 116,160 men reported their sexual attraction toward women and 115,751 men indicated their attraction toward men. Mean scores (SE) per parental income group are given in Table 2.

Table 2.

Male and Female Participants’ Attraction Scores (Mean and SE) to Same- and Opposite-Sex Individuals by Parental Income Groups (I = Bottom 25% of the Population, II = Low 50% of the Population; 3 = Upper 50% of the Population, IV = Top 25% of the Population).

	Parental Income
	Group I	Group II	Group III	Group IV
Males
Attracted to women	6.357 (.013)	6.374 (.007)	6.421 (.006)	6.402 (.013)
Attracted to men	1.898 (.016)	1.870 (.008)	1.844 (.007)	1.893 (.016)
Females
Attracted to men	6.212 (.014)	6.293 (.007)	6.341 (.006)	6.325 (.016)
Attracted to women	2.499 (.019)	2.386 (.009)	2.465 (.008)	2.435 (.023)

Concerning attraction scores toward women, there were differences across parental income groups (F_3,116156 = 13.39, p < .0001). The lowest scores for attraction to women were reported by children of parental income group I and the highest by children of parental income group III. Post-hoc tests (Fisher's PLSD) showed relative attraction scores to women in parental income groups to be I < III (p < .0001), I < IV (p < .05), and II < III (p < .0001). The overall shape of the curve approximated that of the frequency of heterosexuality plotted on parental income groups (Figure 2(a)). Concerning attraction scores toward men, there were differences between the parental income groups (F_3,115747 = 5.78, p < .001). Post-hoc tests showed the lowest scores in children from the parental income group III, with I > III (p < .001), II > III (p < .05), and III < IV (p < .01). The relationship of attraction to men plotted on parental income groups was J-shaped and was similar to that of the frequency of offspring homosexuality plotted on parental income groups (Figures 1(a) and 2(a)).

Figure 2.

Attraction scores (mean ± SE) of male (a) and female (b) participants toward women and men plotted against parental income groups (I = bottom 25% of the population, II = low 50% of the population; 3 = upper 50% of the population, IV = top 25% of the population).

Female Participants

A sample of 95,320 women reported their sexual attraction toward men and 93,576 women reported their attraction toward women. Mean scores (SE) for attraction per parental income group are given in Table 2. Concerning attraction scores towards men, there were differences between parental income groups (F_3,95316 = 31.58, p < .0001). The lowest scores for attraction to men were reported by children of parental income group I and the highest by children of parental income group III. Post-hoc tests showed relative attraction scores to men in parental income groups to be II > I (p < .0001), III > I (p < .0001), IV > I (p < .0001), and III > II (p < .0001). The inverted U-shape of the curve approximated that of the frequency of heterosexuality plotted against parental income groups (Figures 1(b) and 2(b)). Concerning attraction scores toward women, there were significant differences between parental income groups (F_3,93572 = 18.33, p < .0001). Post-hoc tests showed the lowest scores in children from parental income group III, with I > II (p < .0001), I > III (p < .0001), I > IV (p < .05), IV > II (p < .05), and IV > III (p < .01). The relationship of attraction to women plotted on parental income groups was J-shaped and was similar to that of the frequency of offspring homosexuality plotted on parental income groups (Figures 1(b) and 2(b)).

Ethnic and National Variation

There are ethnic and national differences in mean 2D:4D (Manning, 2002), and there may be national differences in mean 2D:4D of male homosexuals (Manning & Robinson, 2003). Therefore, we consider the largest ethnic and national groups in the BBC internet study, that is, White participants resident in the UK and the USA.

Male Participants

Concerning sexual orientation, there were 49,155 White men from the UK and 22,367 White men from the USA. Counts and frequencies of the sexual orientation categories across the four parental income groups, split by nation, are reported in Table 3.

Table 3.

Counts and Frequencies of UK and USA Male and Female Participants’ Sexual Orientation from Parents of Income Groups (I = Bottom 25% of the Population, II = Low 50% of the Population; 3 = Upper 50% of the Population, IV = Top 25% of the Population).

	Parental Income
	Group I		Group II		Group III		Group IV
	UK	USA	UK	USA	UK	USA	UK	USA
Heterosexual
Males	5193 (89.00%)	1767 (86.96%)	16,404 (90.24%)	6902 (87.76%)	19,668 (91.16%)	9317 (89.44%)	3240 (90.83%)	1827 (88.99%)
Females	3763 (89.89%)	1488 (83.17%)	13,719 (91.93%)	6283 (85.45%)	16,129 (92.45%)	7964 (88.16%)	1575 (92.43%)	1068 (84.49%)
Homosexual
Males	375 (6.43%)	134 (6.59%)	1083 (5.96%)	502 (6.38%)	1133 (5.25%)	586 (5.63%)	189 (5.05%)	141 (6.87%)
Females	118 (2.82%)	88 (4.92%)	392 (2.62%)	325 (4.42%)	426 (2.44%)	344 (3.82%)	37 (2.17%)	49 (3.88%)
Bisexual
Males	266 (4.56%)	131 (4.94%)	692 (3.81%)	461 (4.43%)	1133 (3.59%)	514 (4.03%)	138 (3.87%)	85 (4.14%)
Females	305 (7.29%)	213 (11.91%)	812 (5.44%)	745 (10.13%)	892 (5.11%)	705 (7.82%)	92 (5.40%)	147 (11.63%)
Total
Males	5834	2032	18,179	7865	21,575	10,417	3567	2053
Females	4186	1789	14,923	7353	17,447	9013	1704	1264

Both national samples showed significant relationships between parental income and sexual orientation of their male adult children (UK χ²₍₆₎ = 30.21, p < .0001; USA χ²₍₆₎ = 27.25, p < .0001). The frequencies of male heterosexuality across all parental income groups were higher in the UK sample than in the US sample. Male homosexuality and bisexuality were less frequent in the UK sample compared to the US sample. Both nations showed increases in heterosexuality from income groups I–III with a downturn in frequency for parental income group IV. There was a tendency for reductions in the frequency of children's homosexuality and bisexuality with increasing parental income, except for the US sample. In the latter, there was a marked increase in the frequency of homosexuality in parental income group IV.

Concerning sexual attraction scores for men, 48,919 male participants from the UK reported scores for attraction toward women and 48,961 toward men (Table 4). The US samples were 22,231 and 22,291, respectively. Figure 3(a) shows male mean (±SE) attraction scores toward women and Figure 3(b) toward men, plotted on parental income and split by nation.

Table 4.

UK and USA Male and Female Participants’ Attraction Scores (Mean and SE) to Same- and Opposite-Sex Individuals by Parental Income Groups (I = Bottom 25% of the Population, II = Low 50% of the Population; 3 = Upper 50% of the Population, IV = Top 25% of the Population).

	Parental Income
	Group I		Group II		Group III		Group IV
	UK	USA	UK	USA	UK	USA	UK	USA
Males
Attracted to women	6.33 (.019)	6.39 (.033)	6.35 (.010)	6.40 (.016)	6.39 (.009)	6.47 (.013)	6.40 (.022)	6.32 (.032)
Attracted to men	1.89 (.022)	1.95 (.039)	1.86 (.012)	1.96 (.019)	1.84 (.011)	1.89 (.016)	1.86 (.026)	2.01 (.037)
Females
Attracted to men	6.20 (.021)	6.21 (.034)	6.29 (.010)	6.29 (.015)	6.35 (.009)	6.34 (.013)	6.32 (.022)	6.31 (.035)
Attracted to women	2.26 (.026)	2.75 (.045)	2.23 (.013)	2.61 (.021)	2.27 (.012)	2.54 (.018)	2.28 (.026)	2.67 (.049)

Figure 3.

Attraction scores (mean ± SE) of UK and US male participants toward women (a) and men (b) plotted against parental income groups (I = bottom 25% of the population, II = low 50% of the population; 3 = upper 50% of the population, IV = top 25% of the population).

A two-factor ANOVA with dependent variable attraction scores towards women and independent variables parental income (groups I–IV) and nation (UK, USA) showed effects for income (F_3,71142 = 10.28, p < .0001), nation (F_1,71142 = 4.53, p < .05), and an interaction between parental income and nation (F_3,71142 = 3.87, p < .01). Post-hoc comparisons of scores showed income group III > I (p < .0001), III > II (p < .0001), and III < IV (p < .05). Scores were higher for the USA compared to the UK (mean difference −.05, p < .0001). Notably, for the UK the scores were IV > III but for the US they were III > IV. For attraction toward men: a two-factor ANOVA with dependent variable attraction scores and independent variables parental income (groups I–IV) and nation (UK, USA) showed effects for income (F_3,71244 = 4.65, p < .01) and nation (F_1,71244 = 30.03, p < .0001), but no interaction effect (F_3,71224 = 2.07, p = .10). Post-hoc comparisons showed income group I > III (p < .05), II > III (p < .05), and III > IV (p < .05). Same-sex attraction scores were higher for the US compared to the UK (mean difference −.079, p < .0001). Notably, among the US participants, there was a large increase in same-sex attraction from income groups III–IV.

Female Participants

Concerning sexual orientation, there were 38,260 White women from the UK and 19,419 from the USA. Counts and frequencies of the sexual orientation categories across the four parental income groups, split by nation, are reported in Table 3.

Both national samples showed relationships between parental income and sexual orientation of their adult female children (UK χ²₍₆₎ = 34.93, p < .0001; USA χ²₍₆₎ = 60.87, p < .0001). The frequencies of female heterosexuality across all parental income groups were higher in the UK sample than in the US sample. Female homosexuality and bisexuality were less frequent in the UK sample compared to the US sample. Both nations showed increases in heterosexuality from income groups I–III with a further small increase for income group IV for the UK and a US downturn in frequency for parental income group IV. There was a tendency for reductions in the frequency of children's homosexuality and bisexuality with increasing parental income, except for the US sample. In the latter, there was an increase in the frequency of homosexuality in parental income group IV.

Concerning sexual attraction scores for women, 38,034 UK participants reported scores for attraction toward men and 37,502 toward women (Table 4). The US numbers were 19,316 and 191,741, respectively. Figure 4(a) shows female mean (±SE) attraction scores toward men and Figure 4(b) toward women, plotted on parental income and split by nation.

Figure 4.

Attraction scores (mean ± SE) of UK and US female participants toward men (a) and women (b) plotted against parental income groups (I = bottom 25% of the population, II = low 50% of the population; 3 = upper 50% of the population, IV = top 25% of the population).

A two-factor ANOVA with dependent variable attraction scores toward men and independent variables parental income (groups I–IV) and nation (UK, USA) showed an effect for income (F_3,57344 = 18.17, p < .0001), but not for nation (F_1,57344 = 0.006, p = .94 and the interaction between parental income and nation (F_3,57344 = 0.16, p = .93. Post-hoc comparisons showed income III > I or II or IV (p < .0001), and III > II (p < .0001). The scores were not significantly different between the USA and the UK (mean difference −.001, p = .92). For attraction toward women: a two-factor ANOVA with dependent variable attraction scores and independent variables parental income (groups I–IV) and nation (UK, USA) showed effects for income (F_3,56670 = 13.01), nation (F_1,56670 = 297.54), both p < .0001 and the interaction between income and nation (F_3,56670 = 5.34, p < .001). Post-hoc comparisons showed income groups I > II (p < .0001), I > III (p < .0001), IV > II (p < .01), and IV > III (p < .01). Same-sex attraction scores were higher for the US compared to the UK (mean difference −.328, p < .0001). Overall, same-sex attraction showed a U-shaped distribution across parental income groups for both males and females in both the UK and US samples.

Discussion

The current study found that in male adult children, homosexuality was the most frequent non-heterosexual group and bisexuality was rare. For female adult children, bisexuality was more frequent than homosexuality. Parental income was related to the frequency of sexual orientation (heterosexuality, homosexuality, bisexuality) of their adult children. The association between parental income and sexual orientation of children was found in the total sample and when we considered the most numerous ethnic group (Whites) in the two largest national samples (UK and USA), thus confirming our prediction (i).

Parents with income that was “much lower than others” (bottom 25% of the population) had the highest frequencies of homosexuality and bisexuality for both male and female adult children. The lowest frequency of homosexuality was found in the male and female children of parents with an income “slightly higher than others” (upper 50% of the population). Parents with an income “much higher than others” (top 25% of the population) had male and female children with intermediate frequencies of homosexuality.

Concerning adult children's scores for sexual attraction to others, in men, same-sex attraction was highest among children of parents with an income corresponding to the bottom and top 25% of the population, and the lowest scores were found in children of parents with an income “slightly higher than others” (upper 50% of the population). For women, same-sex attraction was greatest for adult children of parents with an “income much lower than others” (bottom 25% of the population), lowest in children of parents with an income “slightly higher than others” (as for men) and intermediate for children of parents with an income representing the top 25% of the population. Splitting the sample for ethnicity (Whites only) and nation (UK and USA), UK scores for adult children's same-sex attraction were lower than those for the US with the former reporting the highest scores for the income group representing the bottom 25% of the population and the latter reporting very high scores for the highest income group (top 25% of the population). Overall, for both the frequency of sexual orientation groups and the same-sex attraction scores, we found no evidence for linear relationships. Concerning same-sex attraction scores, post hoc tests confirmed a U-shaped association such that the frequency of homosexuality/bisexuality and the magnitude of same-sex attraction scores were highest in the adult children from parents with the lowest (bottom 25% of the population) and to a lesser extent in parents with the highest income (top 25% of the population). The nadir of the attraction scores was observed in adult children from parents with an income “slightly higher than others” (upper 50% of the population). Therefore, our prediction (ii) was rejected and prediction (iii) was supported.

The curvilinear relationships between parental income and the sexual behavior of their adult children are, to our knowledge, novel. Our interpretation of these relationships is through the lens of the “Trivers–Willard” hypothesis (Trivers & Willard, 1973). We suggest that parental income is a marker of maternal condition (Babones, 2008; Pickett & Wilkinson, 2015). Male children have a greater variance in reproductive success than female children in human and nonhuman animals (Bateman, 1948; Trivers & Willard, 1973). Male children of mothers in poor condition (lowest parental income) will have below-average male reproductive success, but in contrast, daughters from mothers with the lowest parental income will have reproductive success closer to the female mean. Thus, mothers will be selected to favor their daughters depending on parental income by reducing their prenatal testosterone and increasing their prenatal estrogen. If mothers cannot distinguish between their male and female fetuses this will mean they expose their male offspring to a feminized fetal environment. Conversely, mothers in good condition (highest parental income) will androgenize their fetuses of both sexes. We have found high frequencies of homosexuality and bisexuality and same-sex attraction scores in adult children of mothers with the lowest and highest parental income. This suggests that both a high fetal estrogen-to-testosterone ratio and a high fetal testosterone-to-estrogen ratio are related to an increase in the probability of non-heterosexuality in adult children. We speculate that among the former, there may be a high proportion of “receptive” male homosexuals and “femme” female homosexuals. Among the latter, there may be a preponderance of “insertive” male homosexuals and “butch” female homosexuals.

In conclusion, the current study builds on the finding that parental income is negatively related to 2D:4D in both adult female and male children (Manning et al., 2022). Insofar as 2D:4D is a correlate of prenatal sex steroids and prenatal sex steroids influence sexual behavior, we predicted that parental income would be associated with children's sexual orientation and same-sex scores for sexual attraction to others. Our prediction was confirmed, so that compared to adult children of parents with the lowest income and those with an income “slightly higher,” the former was associated with higher frequencies of homosexuality and bisexuality and high attraction scores toward the same sex for both male and female adult children. Moreover, a comparison between children from “slightly higher” income families and those that had “much higher” income also showed an increase in homosexuality, bisexuality, and same-sex attraction scores in the latter. We suggest that very high fetal estrogen and testosterone are both associated with nonheterosexual behavior in adult male and female children. That is, prenatal sex steroids are related to sexual orientation in a U-shaped fashion. The relationship between 2D:4D and sexual orientation should now be reexamined for curvilinear effects.