Trivers and Willard (1973) hypothesized that mothers who give birth to sons when their condition is good and daughters when their condition is bad have an advantage in terms of fitness. This hypothesis is called the Trivers-Willard Hypothesis (TWH). Clutton-Brock, Albon, and Guinness (1984) and Clutton-Brock (2019) found that maternal rank in herds of red deer is passed on to offspring, which correlates with the reproductive success of males, not females, and that high-ranking mothers produce more male offspring than low-ranking mothers. Two meta-analyses by Cameron (2004) and Sheldon and West (2004) confirmed that the TWH is valid in most mammals except humans.

However, studies validating the TWH in humans have had inconsistent results. For example, in the United States, parents with higher educational attainment have a low infant mortality rate for sons (Abernethy & Yip, 1990); on the contrary, this trend was not found in Sweden (Low, 1991). There is also no consistency in studies focusing on the diverse investment patterns of human parents. For example, a study that found that wealthier parents spent more on backpacks for boys than backpacks for girls (Song, 2018) supported the TWH, whereas a report that found that the interaction of socioeconomic status (income and education) with child sex was significant in the opposite direction of the TW effect for three of the 24 variables such as money spent on education and child supervision (Freese & Powell, 1999) denied the existence of the TWH in humans.

These inconsistencies in humans suggest that in order to apply the TWH to humans, more attention needs to be paid to some of the underlying assumptions and the consequences it predicts. In other words, we must examine the theoretical validity of the independent and dependent variables, that is, parental condition and the outcome of parental investment, and the population sample to be tested. We propose some problems and solutions for the selection of independent and dependent variables and samples needed to validate the TWH in humans.

First, the independent variable, the indicator of parental condition, must satisfy all the several assumptions that the TWH contains. According to Trivers and Willard (1973), the TWH includes the following two assumptions. Assumption 1: There is a positive correlation between the maternal condition during the investment and the offspring's condition at the end of the investment. Assumption 2: Good conditions promote greater reproductive success in males than in females; hence the variance of reproductive success is greater in males than in females. Socioeconomic status, which has been employed as an independent variable in many of the studies that have attempted to validate the TWH in humans in the past, may not meet these assumptions, even though many studies on mammals except humans have used maternal rank in the group as an independent variable.

With respect to assumption 1, studies in non-human mammals suggest that the maternal condition during pregnancy can promote or prevent fetal development and birth through some physiological pathways (Cameron, 2004). However, among humans in modern societies with abundant resources, such as in the United States, there is also relatively little country-wide variation in individual health status (Luo, Ding, Gao, Sun, & Zhao, 2017). In fact, the gap between rich and poor is widening, while the population remains fairly obese due to the high rate of junk food consumption (Drewnowski, 2009). Thus, the mother's socioeconomic status is unlikely to affect the success of development and birth of the offspring physiologically in such a society.

Furthermore, with regard to assumption 2, the point cannot be ignored that in modern societies, the correlation between reproductive success and socioeconomic status has already been lost (Borgerhoff Mulder, 1998). According to this report, men of higher socioeconomic status have more children than men of lower socioeconomic status. This phenomenon has also often been explained in terms of life history strategies (e.g., Belsky, Steinberg, & Draper, 1991). The complex social statistics that have developed uniquely in humans emphasize the differences between humans and other mammals.

To address these issues, the independent variable as an indicator of parental condition should be limited to the maternal physiological condition during pregnancy. In humans, it is known that the physiological state of the mother during pregnancy affects the birth and development of the fetus (corresponding to assumption 1). For example, older maternal age (Tough et al., 2002), younger age (Slap & Schwartz, 1989), thinness, and hypertension (De Bernabé et al., 2004) during pregnancy are risk factors for low birth weight and preterm delivery. The physiological condition of the human mother during pregnancy also affects the lifetime disease resistance of the offspring (Gluckman & Hanson, 2006) and is related to the offspring's reproductive life span (Dupont et al., 2012). Furthermore, an increase in reproductive life span would increase the variance in male reproductive success (corresponding to assumption 2) because men, given that they do not conceive, give birth, and lactate, have a much high upper limit on reproductive success than women in terms of the cost in time spent on reproduction. The physiological condition of the mother during pregnancy not only affects the condition of the offspring at birth but also affect the reproductive success of the offspring, that is, the number of grandchildren, and cause subsequent increase or decrease in the mother's inclusive fitness value.

Second, with respect to the dependent variable, that is, the outcome of parental investment, it is necessary to restrict the timing of the observation of the sex ratio. The sex ratio, expressed as the ratio of the number of men to women, is classified into three categories depending on the timing of the observation. The three sex ratios are: primary sex ratio, which is the sex ratio at fertilization; secondary sex ratio, which is the sex ratio at birth; and tertiary sex ratio, which is the sex ratio at reproductive age. Of these, the currently observable secondary and tertiary sex ratios have been the subject of TWH verification in humans.

The conscious allocation of resources by parents after birth, which has been employed as the dependent variable in many of the studies that have attempted to test the TWH in humans in the past, can be interpreted as an indicator reflected in the tertiary sex ratio. However, the tertiary sex ratio uncovers several problems as a measure of parental condition for validating the TWH. First, it is necessary to note that this postnatal allocation of resources needs to be directed to the most cost-effective offspring to increase the parental fitness value, regardless of the fitness value of the offspring (Keller, Nesse, & Hofferth, 2001). Besides, logic that contradicts the TWH applies in some situations. For example, if a son with high education appears more attractive than a daughter, then parents should spend more money on their son's education than on their daughter's, even if the parents are not in a favorable condition. It has also been noted that adjustments to tertiary sex ratios as a result of infanticide may introduce influential factors other than the TWH (Dickemann, 1979; Voland, Siegelkow, & Engel, 1991). Therefore, a secondary sex ratio should be employed for the dependent variable.

However, simply observing the secondary sex ratio is also problematic. In modern society, this means that the secondary sex ratio appears to be less likely to reflect the TW effect. The TW effect is caused by the physical vulnerability of the male fetus (e.g., Wells, 2000). On the other hand, some believe that the TW effect is not detectable in modern societies, where resources are more abundant than in our ancestral environment (Keller et al., 2001). Fetuses who could not be born in ancestral environments are more likely to survive in modern society.

This problem could be solved by using birth weight as a surrogate indicator rather than using the secondary sex ratio. That is, either an increase in birth weight (leading to fetal birth) or a decrease in birth weight (leading to fetal death) should precede the appearance of a sex ratio imbalance. Furthermore, the fact that low birth weight leads to low reproductive success does not conflict with the TWH assumption. For example, women who were born with a low birth weight are associated with low birth weight of offspring, high frequency of stillbirths, and early infant mortality (Lumey & Stein, 1997). In addition, men born with low birth weight are less likely to marry successfully in their lifetime (Phillips et al., 2001). Given these considerations, birth weight is likely to be a more valid outcome variable than sex ratio in a modern society where nutrition deficiency is a rare cause of fetal death.

Finally, regarding sample selection, twin pregnancies may be more suitable than singleton pregnancies for validating the TWH using birth weight. This is because a simple examination of birth weight is problematic due to the difficulty in confirming the TW effect with it. Usually, male fetuses tend to be heavier than female fetuses, and fetuses in a good condition weigh more than those in a poor condition. Thus, there is the possibility that a male fetus will remain heavier than a female fetus by the simple effect of sex, even under poor conditions. In addition, due to the simple effect of the condition, a fetus in a good condition may weigh more than a fetus in a poor condition, even if the female fetus in the poor condition receives investments. Indeed, several studies have used birth weight as an indicator, but these studies found no evidence of a TW effect (e.g., Gaulin & Robbins, 1991; Keller et al., 2001).

To solve this problem, twin birth weights can be used for two types of validation. First, in opposite-sex paired twins, the male to female weight ratio can confirm the allocation of the investment. Usually, the weight of twins who achieve delivery is strongly correlated with the pair. If the twins have exactly the same weight, the weight ratio indicates 1. A weight ratio value >1 indicates that the male newborn is heavier. According to the TWH, the weight ratio should decrease because female fetuses may be prioritized in poor conditions.

A second validation using twins is to compare the simple birth weights of same-sex and opposite-sex twins. If the nutritional supply from the mother is adjusted to promote the development of the offspring of one sex, same-sex twins may not enjoy the same advantages as opposite-sex twins. Same-sex twins share the sex-specific nutritional benefits derived from their mothers equally (Corsello & Piro, 2010). In contrast, in the case of opposite-sex twins, only one fetus would gain more maternal condition-linked nutritional benefits. In this case, the fetus in the mother's utero receives nourishment depending on the sex. In other words, on the basis of the TWH, a same- or opposite-sex twin may have an advantage or disadvantage depending on the combination of the maternal condition and the sexes of the fetuses concerned.

With regard to this second validation, it is worth noting that twin pregnancies themselves are risk factors for stillbirths, premature births, and low birth weight; it has been reported that twins have inferior reproductive success throughout life compared to singletons (Wyshak & White, 1969). In other words, the assumption of twin pregnancies gives priority to female fetuses. Thus, it should be difficult to ascertain the relative male priority unless the effects of other maternal conditions, such as low risk maternal age for low birth weight, outweigh the effects of twin pregnancies. Given that twin pregnancies increase the risk of low birth weight more than eight fold (Taffel, 1992), whereas younger or older maternal age can only increase the risk of low birth weight by two-fold at most (citation), the effect of maternal age should not outweigh the effect of twin pregnancy. That is to say, based on TWH projections, the combination of conditions that would receive the most investment is probably “female fetus × high risk maternal age for low birth weight × opposite-sex twins.”

In light of the above discussion, the purpose of this study was to examine the TW effect in contemporary Indian society. In response to the above two-step approach, the following hypotheses were derived. If a TW effect exists: 1) the weight ratio of male to female twins born to mothers of high risk age for low birth weight should be smaller than that in mothers of low risk age for low birth weight. 2) The birth weight of female newborns from opposite-sex twins born to mothers of high risk age must be higher than the birth weight of female newborns born from same-sex twins.