Late Nineteenth and Early Twentieth Century Social Feminism and Women’s Suffrage: A Female–Male Net Nutrition Comparison using Differences- in-decompositions

Abstract

When other measures for economic welfare are scarce or unreliable, the body mass index (BMI) is a biological measure that reflects current net nutrition. This study uses a difference-in-decompositions framework to analyse how women’s BMIs varied with the advent of early twentieth century social feminism. Late nineteenth and early twentieth century US economic development improved the relative status of women relative to both men before and after the transition to social feminism. Twentieth century women’s BMIs were higher than nineteenth century women relative to men with the rise of social feminism. The primary source of female–male across-group variation was height and nativity, indicating that there was net nutritional progress for women relative to men associated with changing cumulative net nutrition. The primary source of female–male within-group variation was nativity and socioeconomic status, indicating that there was net nutritional progress relative to women born before the transition for women born after the rise of social feminism association with socioeconomic status.

JEL Codes: C1, C4, D1, I1, N3

Keywords

BMI Variation Economic Transitions Oaxaca Decompositions

Introduction

Resource allocation within the household reflects relative bargaining power between women and men. During the nineteenth and twentieth centuries, US wealth and income were allocated unevenly within the household, and males disproportionately received more resources within the household (Burnette, 2013, p. 306; Marques et al., 2018, p. 158; William et al., 2018, pp. 278–297). However, wealth and income are shared resources, which mask household gender inequality. Net nutrition is an alternative resource measure to wealth and income, and nutrition accrues to each family member individually and reflects only how resources accrue to each household member. Furthermore, the changing economic and social status of women during economic development is different from men because of biological, economic and intuitional conditions (Burnette, 2013, pp. 307–309).

Throughout the nineteenth and early twentieth centuries, the economic and social role of women in the USA changed considerably, and across the USA, women’s changing biological living standards occurred at uneven rates. Nevertheless, women’s roles within the household and the economy changed with economic and political change (Carson, 2021).Various factors combined to influence the material, economic and social roles between women and men that frequently cannot be measured with standard welfare measures. In traditional economies, greater strength is required to complete manual tasks, and men have more muscle mass per unit of tissue, consequently, are stronger than women (Marques et al., 2018, p. 151; Robb, 1994, p. 222). As a result, men historically had different labour market standing when occupations required greater physical strength. Because of childbirth and rearing, travel in traditional economies is also more difficult for women, and women are less mobile and are further excluded from paid-labour market participation. Furthermore, women during the nineteenth century were excluded from market opportunities in developing economies because of institutional constraints, such as religion and social stigmas that foreclosed women from skilled positions. Over time, much of these biological and institutional constraints have changed with economic development, technological innovations and labour market transitions, and women presently have greater economic and nutritional opportunities compared with their female and male historical counterparts.

Not all of the economic changes that women faced were due to technological change and labour market transitions (Burnette, 2013, pp. 309–311). Throughout history, the legal rights of women were tied to their male companions (Becker, 1981, pp. 31–39, 251). By the mid-nineteenth century, pressure increased for women’s legal and political rights to vote and have greater economic opportunity. The National American Women Suffrage Association (NAWSA) was organised in the 1890s; however, when women’s suffrage was denied by the Supreme Court’s 1874 Minor vs. Happersett decision, much of women’s suffrage was refocused toward individual states. Ironically, the first state victories for women’s suffrage were in Wyoming (1869) and Utah (1870), two territories along the Western frontier where the relative bargaining power of women—both within society and within the household—was greater than other regions within the USA. States with larger suffrage movements and competitive political systems were another reason women secured the right to vote in the West compared with the East and South (Teele, 2018). Two late nineteenth and early twentieth century events occurred to secure women’s right to vote. First, suffrage leaders broadened their political support by creating cross-class bridges between working and upper class women, which increased the likelihood of political support and required greater organisational diversity and increased ideological flexibility. Second, the US’ entry into the First World War also aided women’s suffrage, which provided suffrage leaders greater political leverage by questioning US efforts to fight for democracy abroad before it was domestically established (Buechler, 1990, p. 207). In 1920, the US Constitution’s Nineteenth Amendment later codified women’s right to vote, important progress toward women’s equality within the USA.

When monetary measures that reflect economic development are scarce or unreliable, various alternatives are developed to measure the relationship between individual welfare and economic development. A population’s average stature reflects the cumulative net nutrition between calories consumed and calories expended for work and to withstand the physical environment (Fogel et al., 1978, 1979). A population’s average body mass index (BMI) reflects current net nutrition available between the same variables (Fogel, 1994, pp. 375–383), and the BMI is used here to assess how women’s current net nutrition varied relative to men and relative to women born before and after the rise of social feminism.

It is against this backdrop that this study uses US’ historical women and men’s BMIs and difference-in-decompositions to illustrate how women’s relative current net nutrition varied with the rise of social feminism and women’s suffrage. Two paths of inquiry are considered. First, how did women and men’s BMIs vary across and within groups with the rise of social feminism and suffrage? Women’s BMIs increased relative to men and were higher after the transition. Second, what was the greatest source of BMI variation with the transition? The primary source of across-group variation was height and nativity. The greatest source of within-group variation was nativity and socioeconomic status. Subsequently, women’s net nutrition improved relative to men and relative to women observed before the rise of social feminism, and nativity was an important factor explaining variation in current nutrition.

Institutional Change and Biology

An institutional explanation is needed to account for how long-run biological variation changed within the household by gender. When direct measures for net nutrition are scarce or do not exist, BMI reflects current and cumulative net nutrition and resource allocation within the household (Carson, 2012a, b, 2014, 2021; Komlos & Brabec, 2010). If nutrition is restricted over prolonged periods, average stature is stunted, and body mass is wasted (Carson, 2020; Mamoboloa et al., 2005). However, if net nutrition disruptions are short-run events, the body can make up for previous net nutrition shortfalls if sufficient nutrition is restored at older ages, a process known as catch-up growth (Steckel, 1995). Nevertheless, catch-up growth requires that calorie privation be sufficiently prolonged to alter the body’s composition, and short-run disruptions—such as scarce calories and infectious disease—are less likely sources for permanent physical change (Carson, 2011).

Institutions affect resource allocation, and political agents facilitate social and economic influence that affect change (Bush, 1987, p. 1099; Hodgson, 2009; North, 1990), and their actions alter how economic interactions are conducted. For the most part, women place higher priority within the household on child welfare, and with economic development, men come to realise that allocating rights to women to care for children improves early childhood human capital formation (Kose et al., 2021). Elizabeth Cady Stanton, Susan B. Anthony and Carrie Chapman Catt were instrumental in changing US gender arrangements in labour markets that contribute to household resource allocation, net nutrition around the time of the rise of social feminism, and passage of the Nineteenth Amendment. In pre-industrial agricultural economies, child labour and productivity are low, and arrangements within the household allow husbands to control their wives’ property. However, with economic development, the benefit of allocating more childcare rights to women increases child human capital formation, and women gain more allocative rights within the household with economic development (Jennings, 1992, pp. 132–133; Moehling & Thomasson, 2020, p. 8).

Women and Men’s Body Mass and Health

The BMI is one measure that reflects health by gender before modern economic measures and medical research developed. There has been a modern increase in BMIs, and higher BMIs are associated with deleterious health outcomes (Pope, 2021; Waaler, 1984). However, late nineteenth and early twentieth century BMIs were in healthy ranges, indicating historically poor health was not related to high BMIs, and historical comparisons are less complicated than modern comparisons because BMIs were in lower, healthy ranges. Mortality risk is associated with body mass, and if historical mortality risk by gender is comparable with modern standards, mortality risk for women and men is minimised for a BMI around 25 (Costa, 1993). However, women’s relative mortality risk increases and is higher than men for BMIs lower and higher than 25 (Fogel, 1994, p. 376; Waaler, 1984). There is a relationship between early life conditions and later-life outcomes, and a novel explanation is a pre-natal adaptive response to in utero nutrition, where a child’s metabolism and growth trajectory are programmed early to match conditions in later life (Barker, 1992; Carson, 2016; Schneider, 2017, pp. 4–7). Carson (2018a) shows that women and men’s BMIs stagnated throughout the late nineteenth and early twentieth centuries (Carson, 2016), and individuals of African descent had greater BMIs than individual of European descent (Carson, 2009, 2012c, 2018). Women and men from the Southwest were taller and had lower BMIs compared with their counterparts from elsewhere within the USA (Carson, 2019a, pp. 32–33). Women’s BMIs did not vary by socioeconomic status, while male farmer and unskilled workers’ BMIs were higher than workers in other occupations (Carson, 2012c; Carson, 2018).

Women are shorter than men and measuring obesity with BMIs is difficult because women are shorter, which upwardly biases their BMIs and obesity classification (Himes, 2011, p. 40). However, because they have less protein in muscle tissue, women’s BMIs and obesity may be lower than men for the same tissue mass (McLannahan & Clifton, 2008, p. 42). There are other drawbacks of using BMIs as a measure for health. Burkhauser and Cawley (2008) demonstrate that BMIs are misleading because they do not distinguish between fat and fat-free mass. In general, BMI is less accurate among men than women at classifying obesity and when available, more accurate measures for classifying obesity should be used. However, none of the advanced BMI measuring techniques were available during the late nineteenth and early twentieth centuries.

Late Nineteenth and Early Twentieth Century Female and Male Body Mass Index Data

Evaluating late nineteenth century women and men’s BMI variation is difficult because institutions that randomly collected weight and height were yet to develop. Military and prison records are two common sources that collected late nineteenth and early twentieth century male weight and height data. All historical data reflect the purposes for which they are collected, and because the purpose of military records did not include women, prison records are the only late nineteenth and early twentieth century sources that exist to compare women and men’s BMI values. Moreover, because prison records reflect conditions among lower socioeconomic groups, prison records measure lower socioeconomic status female and male biological conditions (Bereczki et al., 2018, p. 190; Ellis, 2004; Sokoloff & Villaflor, 1982). Given the limited number of institutions that recorded women’s historical weight and height, it is unlikely that a comparable data set will emerge, making prison records a valuable source to compare women and men’s historical net nutrition during economic development.

Data to evaluate BMIs is part of an extensive collection project to collate and organise the weight and height of nineteenth century women and men (Carson, 2011, 2013, 2018). At the time of incarceration, prison enumerators recorded weight, height, gender, complexion, observation year, nativity, occupations, age and residence. Because weight and height measures had legal implications in case an inmate escaped and was recaptured, during this pre-photographic period, prison enumerators were careful when recording physical descriptions. Physical descriptions were also used to identify individuals within prisons. There are 4,592 women and 172,277 men used in this study, and women made up about 2.6% of the prison sample.

There are various measurement concerns that began early in stature and BMI studies. Among the first was whether individuals were measured with or without shoes, which may have influenced their terminal statures, therefore, BMIs. Fogel et al. (1978, p. 456) address this concern using a sample of Union Army recruits known to have been measured without shoes. These observations are then compared with an adult black recruit’s sample, and there is little difference between the two samples. There are no nineteenth century data that exist that measures women’s height without shoes, so a similar female comparison is not available. There is a parallel concern for weight that influenced weight measures because inmates may have been measured with or without clothes. Because females and males were incarcerated at the same time and because of gender social roles at the time, inmates were probably measured with clothes. Prison enumerators did not systematically record whether a woman was pregnant, so that, no information was recorded for pregnancy status.

There is a recent challenge to the established pattern known as the antebellum paradox, the result that average statures decreased with industrialisation and urbanisation during the nineteenth century’s second and third quarters (Bodenhorn et al., 2017). However, vigorous rebuttal calls these criticisms into question (Carson, 2019a, pp. 32–33; Komlos, 2019). The recent criticism also does not account for shorter urban compared with rural statures, and urban statures decreased when food consumption was separated from food production (Carson, 2008, p. 368; Komlos, 1987). The nineteenth century’s second quarter stature decrease is also observed across interdisciplinary studies. For example, Davidson et al. (2002, p. 268), and Steckel and Rose (2002, p. 575) show that statures were shorter in geographic areas with higher disease rates, and the disease explanation is an important part of the antebellum paradox (Haines et al., 2003, p. 406).

Occupations are among the best measures for historical socioeconomic status, and occupation classifications used here are skilled, unskilled and workers without listed occupations. For women, the prison records include skilled occupations, such as nurses and dressmakers. For men, there is greater skilled occupational diversity because male workers included bankers and also included occupations that restricted women from participation, such as the clergy. Examples of unskilled women include waitresses and cooks, while male unskilled workers also include day labourers and miners. Because women were not listed as farmers, male farmers are excluded from the analysis. A final occupational category is for women and men who did not report an occupation at the time of incarceration, who are classified here as ‘no listed occupations’.

Race is an important means to identify individuals within prisons and is inferred from a complexion variable. The complexion category is less gender-specific than occupations, and across prisons, women and men were recorded with the same complexion categories. For both women and men, individuals of African descent were classified as black, light black and various shades of mulatto. European inmates were recorded as white, light, medium and dark. This white European classification is also supported by foreign-born individuals from primarily white European populations who were recorded with the same white, light, medium and dark complexions. In census and prison records, until the 1930s, it was common to designate individuals of mixed African and European ancestry as ‘mulattos’. However, individuals recorded as mulattoes are referred to as ‘mixed-race’ in the results that follow.

Birth regions are classified into seven broad geographic regions. Individuals from Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island and Vermont are classified as from the Northeast. Birth in Delaware, Washington DC, Maryland, New Jersey, New York and Pennsylvania are classified as from the Middle Atlantic. Individuals from Illinois, Indiana, Michigan, Ohio and Wisconsin are classified as from the Great Lakes. Individuals from Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota and South Dakota are classified as from the Plains. Individuals from Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Virginia and West Virginia are classified as from the Southeast. Individuals from Arizona, California, Colorado, Idaho, Montana, Nevada, Oregon, Utah, Washington and Wyoming are classified as from the Far West (Carlino & Sill, 2001).

Birth cohorts for late nineteenth and early twentieth century women and men are classified into four groups: women and men born before and after 1900. Table 1 indicates that males were a larger portion of the prison population compared with females. Women were also more likely to be black and mixed race than men, and there is an increase in both the number and proportion of white women later in the sample (Carson, 2009, 2018). This change in female composition may be related to vagrancy laws that imprisoned idle individuals, which were applied to white women later in the sample (Brands, 2010, p. 156). Women were more likely to be incarcerated at younger ages, and older men were incarcerated before and after the transition to social feminism. Before and after the transition, international nativity is mixed for women; however, foreign-born males were likely to be incarcerated after 1900. Native-born results are mixed before and after the transition, as are women and men’s residence. Reflecting labour market, biological and institutional constraints, women were more likely to be listed without occupation before and after the rise of social feminism. Nevertheless, men were more likely to be listed as unskilled workers before 1900, and women were more likely to be enumerated as unskilled workers after the transition. Before and after the transition to social feminism, men were more likely than women to be listed as skilled workers. Women residing in the nineteenth century Northeast were more likely to be from the Northeast than nineteenth century men, a pattern that persistent into the twentieth century.

Table 1.

Women and Men’s Nineteenth and Twentieth Century Characteristic Distributions

	Female, Nineteenth Century		Female, Twentieth Century		Male, Nineteenth Century		Male, Twentieth Century
	N	%	N	%	N	%	N	%
Ethnic
Black	874	44.19	869	33.24	17,337	22.77	22,219	23.11
Mexican	44	2.22	41	1.57	2,779	3.65	3,846	4.00
Mixed race	387	19.57	737	28.19	15,833	20.80	10,298	10.71
White	673	34.02	967	36.99	40,178	52.78	59,787	62.18
Ages
Teens	609	30.79	433	16.56	12,674	16.65	11,725	12.19
20s	836	42.26	1,330	50.88	39,880	62.39	47,469	49.37
30s	311	15.72	568	21.73	14,495	19.04	22,299	23.19
40s	140	7.08	201	7.69	5,898	7.75	9,548	9.93
50s	61	3.08	62	2.37	2,449	3.22	3,831	3.98
60s	21	1.06	20	0.77	731	0.96	1,278	1.33
Native
International
Canada	7	0.35	25	.96	588	0.77	990	1.03
Europe	59	2.98	82	3.14	3,308	4.35	6,039	6.28
Great Britain	141	7.13	31	1.19	3,126	4.11	1,891	1.97
Latin America	28	1.42	56	2.14	2,385	3.13	4,265	4.44
National
Far West	1	0.05	74	2.83	838	1.10	3,002	3.12
Great Lakes	88	4.45	310	11.86	3,845	5.05	11,454	11.91
Middle Atlantic	422	21.33	132	5.05	14,901	19.57	9,036	9.40
Northeast	6	0.30	12	0.46	819	1.08	1,125	1.17
Plains	44	2.22	488	18.67	2,616	3.44	17,585	18.29
Southeast	797	40.29	879	33.63	30,749	40.39	25,553	26.58
Southwest	385	19.46	525	20.08	12,952	17.01	15,210	15.82
Residence
Arizona	5	0.25	19	0.73	782	1.03	3,250	3.38
Colorado			301	11.51			5,720	5.95
Idaho			12	0.46	50	0.07	629	0.65
Illinois	96	4.85	408	15.61	1,453	1.91	9,861	10.26
Kentucky	101	5.11	19	0.73	8,197	10.77	3,323	3.46
Missouri	19	0.96	469	17.94	2,086	2.74	17,114	17.80
Mississippi			34	1.30	24	0.03	1,674	1.74
Montana	2	0.10	83	3.18	1,211	1.59	7,822	8.14
Nebraska			112	4.28			7,364	7.66
New Mexico	31	1.57	22	0.84	973	1.28	2,031	2.11
Oregon	3	0.15			2,189	2.88
PA, East	129	6.52	88	3.37	4,866	6.39	4,095	4.26
PA, West	155	7.84	28	1.07	6,675	8.77	1,009	1.05
Philadelphia	377	19.06			6,671	8.76	2,025	2.11
Tennessee	576	29.12	453	17.33	16,493	21.67	11,746	12.22
Texas	484	24.47	566	21.65	24,457	32.13	18,487	19.23
Occupations
No Occupations	783	39.59	433	16.56	16,253	21.35	9,104	9.47
Skilled	181	9.15	228	8.72	17,531	23.03	33,307	34.64
Unskilled	1,014	51.26	1,953	74.71	42,343	55.62	53,739	55.89
Residence
Northeast	661	33.42	116	4.44	18,212	23.92	7,129	7.41
Plains	216	10.92	896	34.28	11,736	15.42	30,298	31.51
South	1,060	53.59	1,053	40.28	40,974	53.82	31,907	33.18
West	41	2.07	549	21.00	5,205	6.84	26,816	27.89
Total	1,978		2,614		76,127		96,150

Source: Arizona State Library, Archives and Public Records; Colorado State Archives; California State Archives; Idaho State Archives; Illinois State Archives; Kentucky Department for Libraries and Archives; Maryland State Archives; Missouri State Archives; William F. Winter Archives and History Building; Montana State Archives; Nebraska State Historical Society; New Mexico State Records and Archives; Ohio Archives Library; Oregon State Archives; Pennsylvania Historical and Museum Commission; Philadelphia City Archives; Tennessee State Library and Archives; Texas State Library and Archives Commission; Utah State Archives; Washington State Archives.

Econometric Model

In the quasi-experimental literature, difference-in-difference estimators measure treatment effects with only cross-sectional data by creating designs ‘as if treat- ment’ is randomly assigned (Card & Kruegar, 1993). An Oaxaca decomposition separates response variable differences into structural and compositional components (Blinder, 1973; Oaxaca, 1973; Oaxaca & Ransom, 1999).¹ A difference-in-decompositions estimator combines difference-in-difference estimators with Oaxaca decompositions to approximate differences between response variables around the time of an event. Rather than identifying sources of variation, difference-in-decompositions partition group variations into structural and compositional returns. If there is a measurable effect at the time of the event associated with returns to characteristics and only small composition differences, the event effects are more likely due to structural effects associated with the event. However, if there are small structural and large compositional differences, response variable differences are associated with compositional differences.

The Oaxaca decomposition is used to assess the before and after difference in response variables around the time of treatment. Let y_c and y_t be control and treatment response variables.

y_{c} = α_{c} + β_{c} X_{c}

(1)

and

y_{t} = α_{t} + β_{t} X_{t}

(2)

where α_c and α_t are control- and treatment-group autonomous components, respectively. β_t and β_c are structural returns to treatment and control characteristics, respectively. X_t and X_c are treatment and control characteristic matrices, respectively.

A decomposition partitions the response variable differences into structural and compositional effects around the time of the event. In the case of female and male BMI differences, the event is the 1900 rise of social feminism. To summarise the effect of the transition, BMIs are partitioned into female and male vectors before and after the rise of social feminism.

B M I_{f}^{n} = α_{f}^{n} + β_{f}^{n} X_{f}^{n}

(3)

B M I_{f}^{t} = α_{f}^{t} + β_{f}^{t} X_{f}^{t}

(4)

B M I_{m}^{n} = α_{m}^{n} + β_{m}^{n} X_{m}^{n}

(5)

B M I_{m}^{t} = α_{m}^{t} + β_{m}^{t} X_{m}^{t}

(6)

where $α_{f}^{n}$ and $α_{m}^{n}$ are the nineteenth century female and male autonomous BMI components, respectively. $α_{f}^{t}$ and $α_{m}^{t}$ are the twentieth century female and male autonomous components, respectively. $β_{f}^{n}$ and $β_{m}^{n}$ are the nineteenth century female and male returns associated with characteristics, respectively and $β_{f}^{t}$ and $β_{m}^{t}$ are the twentieth century female and male BMI returns associated with characteristics, respectively. $X_{f}^{n}$ and $X_{m}^{n}$ are nineteenth century average female and male characteristic matrices, respectively, while $X_{f}^{t}$ and $X_{m}^{t}$ are defined similarly for twentieth century females and males, respectively.

Across-group Difference-in-decompositions

There are two ways to decompose an event’s effect: across- and within-groups. The across-group difference-in-decompositions isolates the across-group effects before and after social feminism by partitioning the differences between women and men into structural and compositional effects. Because women have higher BMIs than men, they are assigned as the base category (Carson, 2009, 2018). Twentieth century female–male decompositions are calculated by taking the difference between twentieth century female and male models (Equation [4] minus Equation [6]).

B M I_{f}^{t} - B M I_{m}^{t} = α_{f}^{t} + β_{f}^{t} {\overset{─}{X}}_{f}^{t} - α_{m}^{t} - β_{m}^{t} X_{m}^{t}

(7)

The across-group difference-in-decompositions are constructed by first taking the female–male post- and pre-transition decompositions by adding appropriate counterfactuals (Carson, 2018, 2019b). Equation (8) is the twentieth century counterfactual for average male characteristics observed at twentieth century female returns to characteristics. Equation (9) is the twentieth century male BMI returns to characteristics observed at twentieth century female average characteristics.

β_{f}^{t} X_{m}^{t} - β_{f}^{t} X_{m}^{t} = 0

(8)

and

β_{m}^{t} X_{f}^{t} - β_{m}^{t} X_{f}^{t} = 0

(9)

Equation (10) is the twentieth century female–male decomposition for males observed at female coefficients and is obtained by adding Equation (8) to Equation (7) (Table 3, panel A). Equation 11 is the twentieth century female–male decomposition for females observed at male characteristics and is obtained by adding Equation (9) to Equation (7) (Table 3, panel A).

Table 2.

Women and Men’s Nineteenth and Twentieth Century Social Feminism BMI Regression Models by Demographic and Socioeconomic Status

	Female, Nineteenth Century	Female, Twentieth Century	Male, Nineteenth Century	Male, Twentieth Century
Intercept	39.76***	47.65***	34.05***	32.23***
Height (cm)	–0.104***	–0.136***	–0.066***	–0.053***
Complexion
White	Reference	Reference	Reference	Reference
Black	0.131	0.684***	1.21***	0.968***
Mixed race	0.156	0.382*	1.01***	0.726***
Mexican	–0.490	–0.383	0.140**	0.051
Ages
14	–3.35***	–3.07***	–3.46***	–3.05***
15	–2.98***	–2.31***	–2.92***	–2.33***
16	–1.80***	–.614**	–2.19***	–1.84***
17	–1.39***	–1.31***	–1.50***	–1.28***
18	–0.700**	–0.850**	–1.14***	–0.964***
19	–0.594*	–0.556**	–0.705***	–0.567***
20–29	Reference	Reference	Reference	Reference
30s	1.03***	1.52***	0.264***	0.361***
40s	0.783**	2.27***	0.416***	0.655***
50s	1.36***	1.99***	0.523***	0.696***
60s	0.249	2.86***	0.280***	0.662***
Nativity
National
Northeast	Reference	Reference	Reference	Reference
Middle Atlantic	0.626	–4.39***	–0.064	–0.151*
Great Lakes	–0.487	–2.95***	0.057	–0.014
Plains	0.318	–3.32***	–0.059	–0.333***
Southeast	–0.346	–3.57***	0.076	–0.355***
Southwest	–0.935	–3.65***	–0.285***	–0.374***
Far West	3.49	–3.45***	0.051	–0.207**
International
Canada	0.681	–3.17**	0.222*	–0.023
Europe	0.768	–2.12*	0.712***	0.795***
Britain	0.530	–4.83***	0.149*	–0.126
Latin America	–1.39	–3.58***	–0.657***	–0.451***
Residence
Northeast	–1.10***	1.09**	–0.322***	–0.486***
Plains	–0.252	–0.030	–0.343***	–0.726***
South	Reference	Reference	Reference	Reference
West	–0.466	–0.111	0.472***	–0.027
Occupation
Skilled	0.344	0.931***	0.013	–0.266***
Unskilled	0.674***	0.487**	0.208***	–0.196***
No occupation	Reference	Reference	Reference	Reference
N	1,978	2,614	76,127	96,150
R ²	0.1210	0.1288	0.1334	0.0943

Source: See Table 1.

Notes: ***Significant at 0.01, **significant at 0.05, and *significant at 0.10.

Table 3.

Women and Men’s Across-group Nineteenth and Twentieth Century BMI Difference-in-decompositions with the Rise of Social Feminism

Panel A	Equation (10)		Equation (11)
Post 1900	$(β_{f}^{P o s t} - β_{m}^{P o s t}) {\overset{─}{X}}_{m}$ Structure	$({\overset{─}{X}}_{f}^{P o s t} - {\overset{─}{X}}_{m}^{P o s t}) β_{f}^{P o s t}$ Composition	$(β_{f}^{P o s t} - β_{m}^{P o s t}) {\overset{─}{X}}_{f}$ Structure	$({\overset{─}{X}}_{f}^{P o s t} - {\overset{─}{X}}_{m}^{P o s t}) β_{m}^{P o s t}$ Composition
Levels
Sums	–0.442	1.04	–0.003	0.600
Total		0.597		0.597
Proportions
Intercept	25.84		25.84
Height	–23.73	2.12	–22.44	0.827
Complexions	–0.201	0.244	–0.332	0.375
Ages	0.884	–0.282	0.764	–0.162
Nativity	–5.39	–0.048	–5.36	–0.082
Residence	0.524	–0.043	0.487	–0.006
Occupations	1.34	–0.251	1.03	0.054
Sum	–0.741	1.74	–0.006	1.01
Total		1		1
Panel B	Equation (15)		Equation (16)
Pre 1900	$(β_{f}^{P o s t} - β_{m}^{P o s t}) {\overset{─}{X}}_{m}$	$({\overset{─}{X}}_{f}^{\Pr e} - {\overset{─}{X}}_{m}^{\Pr e}) β_{f}^{\Pr e}$	$(β_{f}^{\Pr e} - β_{m}^{\Pr e}) {\overset{─}{X}}_{f}^{}$	$({\overset{─}{X}}_{f}^{\Pr e} - {\overset{─}{X}}_{m}^{\Pr e}) β_{m}^{\Pr e}$
Levels
Sums	–0.998	0.680	–.934	.615
Total		–0.318		–.318
Proportions
Intercept	–17.93		–17.93
Height	20.34	–3.33	19.13	–2.12
Complexions	1.40	–0.104	2.07	–0.768
Ages	–0.754	0.766	–0.777	0.788
Nativity	0.390	0.070	0.450	0.010
Residence	0.742	0.223	0.846	0.118
Occupations	–0.105	0.242	–0.845	0.034
Sum	3.14	–2.14	2.93	–1.93
Total		1		1
Panel C	Equation (17)		Equation (18)
Difference-in-decompositions
Levels
Sums	0.556	0.359	0.930	–0.015
Total		0.915		0.915
Proportions
Intercept	43.77		43.77
Height	–44.08	5.45	–41.57	2.94
Complexions	–1.60	0.348	–2.40	1.14
Ages	1.64	–1.05	1.54	–0.950
Nativity	–5.78	–0.119	–5.80	–0.092
Residence	–0.218	–0.266	–0.359	–0.125
Occupations	2.39	–0.049	1.88	0.195
Sum	–3.88	3.88	–2.94	2.94

Source: See Tables 1 and 2.

B M I_{f}^{t} - B M I_{m}^{t} = (α_{f}^{t} - α_{m}^{t}) + (β_{f}^{t} - β_{m}^{t}) {\overset{─}{X}}_{m}^{t} + ({\overset{─}{X}}_{f}^{t} - {\overset{─}{X}}_{m}^{t}) β_{f}^{t}

(10)

B M I_{f}^{t} - B M I_{m}^{t} = (α_{f}^{t} - α_{m}^{t}) + (β_{f}^{t} - β_{m}^{t}) {\overset{─}{X}}_{f}^{t} + ({\overset{─}{X}}_{f}^{t} - {\overset{─}{X}}_{m}^{t}) β_{m}^{t}

(11)

Equation (12) is the nineteenth century female–male decomposition, calculated by taking the difference between nineteenth century female and male BMI models (Equation [3] minus Equation [5]).

B M I_{f}^{n} - B M I_{m}^{n} = α_{f}^{n} + β_{f}^{n} {\overset{─}{X}}_{f}^{n} - α_{m}^{n} - β_{m}^{n} X_{m}^{n}

(12)

Equation (13) is the counterfactual for nineteenth century female returns to characteristics observed at nineteenth century male characteristics. Equation (14) is the counterfactual for nineteenth century male return to characteristics observed at nineteenth century female characteristics.

β_{f}^{n} X_{m}^{n} - β_{f}^{n} X_{m}^{n} = 0

(13)

and

β_{m}^{n} X_{f}^{n} - β_{m}^{n} X_{f}^{n} = 0

(14)

Equation (15) is the nineteenth century female–male decomposition for males observed at female coefficients and is obtained by adding Equation (13) to Equation (12) (Table 3, panel B). Equation 16 is the nineteenth century female–male decomposition for females observed at male characteristics and is obtained by adding Equation (14) to Equation (12) (Table 3, panel B).

B M I_{f}^{n} - B M I_{m}^{n} = (α_{f}^{n} - α_{m}^{n}) + (β_{f}^{n} - β_{m}^{n}) {\overset{─}{X}}_{m}^{n} + ({\overset{─}{X}}_{f}^{n} - {\overset{─}{X}}_{m}^{n}) β_{f}^{n}

(15)

B M I_{f}^{n} - B M I_{m}^{n} = (α_{f}^{n} - α_{m}^{n}) + (β_{f}^{n} - β_{m}^{n}) {\overset{─}{X}}_{f}^{n} + ({\overset{─}{X}}_{f}^{n} - {\overset{─}{X}}_{m}^{n}) β_{m}^{n}

(16)

To derive the female–male across-group difference-in-decompositions, the second step is to take the female–male difference before and after the transition. Equation (17) is the female–male BMI difference-in-decompositions for female returns observed at male characteristics is Equation (10) minus Equation (15) (Table 3, panel C).

\begin{matrix} (B M I_{f}^{t} - B M I_{m}^{t}) - (B M I_{f}^{n} - B M I_{m}^{n}) = (α_{f}^{t} - α_{m}^{t}) - (α_{f}^{n} - α_{m}^{n}) + \\ [(β_{f}^{t} - β_{m}^{t}) {\overset{─}{X}}_{m}^{t} - (β_{f}^{n} - β_{m}^{n}) {\overset{─}{X}}_{m}^{n}] + \\ [({\overset{─}{X}}_{f}^{t} - {\overset{─}{X}}_{m}^{t}) β_{f}^{t} - ({\overset{─}{X}}_{f}^{n} - {\overset{─}{X}}_{m}^{n}) β_{f}^{n}] \end{matrix}

(17)

Equation (18) is the female–male difference-in-decompositions for male returns observed at female characteristics is Equation (11) minus Equation (16) (Table 3, panel C).

\begin{matrix} (B M I_{f}^{t} - B M I_{m}^{t}) - (B M I_{f}^{n} - B M I_{m}^{n}) = (α_{f}^{t} - α_{m}^{t}) - (α_{f}^{n} - α_{m}^{n}) + \\ [(β_{f}^{t} - β_{m}^{t}) {\overset{─}{X}}_{f}^{t} - (β_{f}^{n} - β_{m}^{n}) {\overset{─}{X}}_{f}^{n}] \\ + [({\overset{─}{X}}_{f}^{t} - {\overset{─}{X}}_{m}^{t}) β_{m}^{t} - ({\overset{─}{X}}_{f}^{n} - {\overset{─}{X}}_{m}^{n}) β_{m}^{n}] \end{matrix}

(18)

Within-group Difference-in-decompositions

The within-group difference-in-decompositions isolates structural and com- positional differences within female and male groups after and before the transition to social feminism. Women are assigned as the base structure (Carson, 2018b, 2019b). The within-group difference-in-decompositions are constructed by first taking the difference between twentieth and nineteenth century female and male BMIs and adding appropriate counterfactuals. Equation (19) is Equation (4) minus Equation (3). Equation (20) is Equation (6) minus (5).

B M I_{f}^{t} - B M I_{f}^{n} = α_{f}^{t} + β_{f}^{t} {\overset{─}{X}}_{m}^{n} - α_{f}^{n} - β_{f}^{n} {\overset{─}{X}}_{f}^{n}

(19)

B M I_{f}^{t} - B M I_{m}^{n} = α_{m}^{t} + β_{m}^{t} {\overset{─}{X}}_{m}^{t} - α_{m}^{n} - β_{m}^{n} X_{m}^{n}

(20)

Equation (21) is the counterfactual for nineteenth century female returns to BMI at average twentieth century female average characteristics. Equation (22) is the counterfactual for nineteenth century female returns counterfactual observed at twentieth century female average characteristics.

β_{f}^{n} X_{f}^{t} - β_{f}^{n} X_{f}^{t} = 0

(21)

β_{f}^{t} X_{f}^{n} - β_{f}^{t} X_{f}^{n} = 0

(22)

Equation (23) is women’s within-group decomposition for twentieth century women’s characteristics observed at nineteenth century average characteristics and is obtained by adding Equation (19) to Equation (21) (Table 4, panel A). Equation (24) is women’s within-group decomposition for nineteenth century characteristics observed at twentieth century returns to characteristics and is obtained by adding Equations (20) to Equation (22) (Table 4, panel A).

B M I_{f}^{t} - B M I_{f}^{n} = (α_{f}^{t} - α_{f}^{n}) + (β_{f}^{t} - β_{f}^{n}) {\overset{─}{X}}_{f}^{t} + ({\overset{─}{X}}_{f}^{t} - {\overset{─}{X}}_{f}^{n}) β_{f}^{n}

(23)

B M I_{f}^{t} - B M I_{f}^{n} = (α_{f}^{t} - α_{f}^{n}) + (β_{f}^{t} - β_{f}^{n}) {\overset{─}{X}}_{f}^{n} + ({\overset{─}{X}}_{f}^{t} - {\overset{─}{X}}_{f}^{n}) β_{f}^{t}

(24)

Equation (25) is the counterfactual for observing nineteenth century male BMI returns at twentieth century average male characteristics. Equation (26) is the counterfactual for observing twentieth century male BMI returns at nineteenth century average male characteristics.

β_{m}^{n} X_{m}^{t} - β_{m}^{n} X_{m}^{t} = 0

(25)

β_{m}^{t} X_{m}^{n} - β_{m}^{t} X_{m}^{n} = 0

(26)

Equation (27) is the difference between twentieth and nineteenth century male returns to characteristics observed at twentieth century average male characteristics, Equation (20) plus Equation (25) (Table 4, panel B). Equation (28) is the difference between twentieth century male returns to characteristics observed at nineteenth century average male characteristics, Equation (20) plus Equation (26) (Table 4, panel B).

Table 4.

Women and Men’s Within-Group Nineteenth and Twentieth Century BMI Difference-in-decompositions with the Rise of Social Feminism

Panel A	Equation (23)		Equation (24)
Females	$(β_{f}^{P o s t} - β_{f}^{\Pr e}) {\overset{─}{X}}_{f}^{P o s t}$ Structure	$({\overset{─}{X}}_{f}^{P o s t} - {\overset{─}{X}}_{f}^{P o s t}) β_{f}^{\Pr e}$ Composition	$(β_{f}^{P o s t} - β_{f}^{\Pr e}) {\overset{─}{X}}_{f}^{\Pr e}$ Structure	$({\overset{─}{X}}_{f}^{P o s t} - {\overset{─}{X}}_{f}^{\Pr e}) β_{f}^{P o s t}$ Composition
Levels
Sums	0.133	0.474	0.472	0.135
Total		0.607		0.607
Proportions
Intercept	13.01		13.01
Height	–8.51	–0.187	–8.45	–0.245
Complexions	0.411	0.004	0.480	–0.065
Ages	0.471	0.436	0.498	0.409
Nativity	–5.42	–0.013	–5.94	0.508
Residence	0.409	0.283	1.26	–0.567
Occupations	–0.146	0.258	–0.070	0.182
Sum	0.219	0.781	0.778	0.222
Total		1		1
Panel B	Equation (27)		Equation (28)
Males	$(β_{m}^{P o s t} - β_{m}^{\Pr e}) {\overset{─}{X}}_{m}^{P o s t}$	$({\overset{─}{X}}_{m}^{P o s t} - {\overset{─}{X}}_{m}^{\Pr e}) β_{m}^{\Pr e}$	$(β_{m}^{P o s t} - β_{m}^{\Pr e}) {\overset{─}{X}}_{m}^{\Pr e}$	$({\overset{─}{X}}_{m}^{P o s t} - {\overset{─}{X}}_{m}^{\Pr e}) β_{m}^{P o s t}$
Levels	–0.400	0.092	–0.285	–0.024
Sums		–0.309		–0.309
Total
Proportions
Intercept	5.90		5.90
Height	–7.19	0.042	–7.18	0.034
Complexions	0.291	0.316	0.381	0.226
Ages	–0.268	–0.339	–0.271	0.336
Nativity	0.639	0.006	0.730	–0.086
Residence	0.882	–0.315	0.429	0.137
Occupations	1.05	–0.007	0.937	0.102
Sum	1.30	–0.297	0.923	0.077
Total		1		1
Panel C	Equation (29)		Equation (30)
Difference-in-decompositions
Levels
Sums	0.533	0.382	0.757	0.159
Total		0.915		0.915
Proportions
Intercept	7.11		7.11
Height	–1.32	–0.229	–1.27	–0.278
Complexions	0.120	–0.312	0.099	–0.292
Ages	0.739	0.774	0.768	0.745
Nativity	–6.06	–0.019	–0.667	0.594
Residence	–0.473	0.599	0.829	–0.704
Occupations	–1.19	0.265	–1.01	0.080
Sum	–1.08	1.08	–0.145	0.145

Source: See Tables 1 and 2.

B M I_{m}^{t} - B M I_{m}^{n} = (α_{m}^{t} - α_{m}^{n}) + (β_{m}^{t} - β_{m}^{n}) {\overset{─}{X}}_{m}^{t} + ({\overset{─}{X}}_{m}^{t} - {\overset{─}{X}}_{m}^{n}) β_{m}^{n}

(27)

B M I_{m}^{t} - B M I_{m}^{n} = (α_{m}^{t} - α_{m}^{n}) + (β_{m}^{t} - β_{m}^{n}) {\overset{─}{X}}_{m}^{n} + ({\overset{─}{X}}_{m}^{t} - {\overset{─}{X}}_{m}^{n}) β_{m}^{t}

(28)

To derive the female–male within-group difference-in-decompositions, the second step is to take the difference between the female after-before decomposition measured at nineteenth century BMI returns to characteristics observed at twentieth century female characteristics. Equation (29) is the within-group difference-in-decompositions observed at twentieth century females at nineteenth century male characteristics, Equation (23) minus Equation (27) (Table 4, panel C).

\begin{matrix} (B M I_{f}^{t} - B M I_{f}^{n}) - (B M I_{m}^{t} - B M I_{m}^{n}) = (α_{f}^{t} - α_{f}^{n}) - (α_{m}^{t} - α_{m}^{n}) + \\ [(β_{f}^{t} - β_{f}^{n}) {\overset{─}{X}}_{f}^{t} - (β_{m}^{t} - β_{m}^{n}) {\overset{─}{X}}_{m}^{n}] \\ + [({\overset{─}{X}}_{f}^{t} - {\overset{─}{X}}_{f}^{n}) β_{f}^{n} - ({\overset{─}{X}}_{m}^{t} - {\overset{─}{X}}_{m}^{n}) β_{m}^{n}] \end{matrix}

(29)

Equation (30) is the within-group difference-in-decomposition for twentieth century returns to characteristics measured at nineteenth century characteristics is Equation (24) minus Equation (28) (Table 4, panel C).

\begin{matrix} (B M I_{f}^{t} - B M I_{f}^{n}) - (B M I_{m}^{t} - B M I_{m}^{n}) = (α_{f}^{t} - α_{f}^{n}) - (α_{m}^{t} - α_{m}^{n}) + \\ [(β_{f}^{t} - β_{f}^{n}) {\overset{─}{X}}_{f}^{t} - (β_{m}^{t} - β_{m}^{n}) {\overset{─}{X}}_{m}^{n}] \\ + [({\overset{─}{X}}_{f}^{t} - {\overset{─}{X}}_{f}^{n}) β_{f}^{t} - ({\overset{─}{X}}_{m}^{t} - {\overset{─}{X}}_{m}^{n}) β_{m}^{t}] \end{matrix}

(30)

Female and Male Body Mass Index Returns: A Difference-in-decompositions Approach

Regression Results

Isolating female and male BMI changes across and within genders demonstrates how relative net-nutrition was related to the rise of social feminism (Table 2). A common result when comparing women and men’s height is that men are biologically taller than women associated with sexual dimorphism (Marques et al., 2018, p. 151; Williams et al., 2018, p. 288), and men were taller than women before and after the rise of social feminism. Moreover, the inverse female relationship between BMI and height was nearly twice the magnitude for men, indicating women’s BMIs were particularly sensitive to height (Carson, 2018c; Komlos & Carson, 2017). As was common throughout the nineteenth and twentieth centuries, individuals with darker complexions had greater BMIs and weight than individuals with fairer complexions (Carson, 2009, 2012, 2015a, 2015b). Before and after the transition, BMI returns to age were similar for women and men, and the female regional BMI returns associated with northeastern nativity was largest in the nineteenth century. During the nineteenth century, the Far West was only in the early stages of economic development, and occupations were segregated along gender lines, where physical strength was required in agricultural occupations along the Far Western frontier (Marques et al., 2018, p. 142; Williams et al., 2018, p. 293). However, women in the northeast before and after the transition had higher BMIs than women elsewhere in the USA.

Across-group Difference-in-decompositions

Isolating female and male BMI differences across and within groups illustrates how current net nutrition was related to the transition to social feminism. Across-group decompositions are first calculated. In Table 3, panel A (Equations [10] and [11]) is the post-1900 female–male BMI compositions. Panel B (Equations [15] and [16]) is the pre-1900 female–male BMI decomposition. Panel C (Equations [17] and [18]) is the female–male across-group difference-in-decompositions with the transition to social feminism, and the sign and magnitude indicate how women and men’s across-group BMI variation changed with the transition to social feminism. For example, a positive intercept difference between twentieth century women and men indicates that women’s autonomous BMI returns were greater than men, whereas their nineteenth century autonomous BMI returns were lower than men.

Across-group, Post-transition Decomposition

In panel A of Table 3 positive-level intercept, 0.597 BMI units, indicates that women’s BMIs were greater than men after social feminism. However, sources of the across-group post-transition were important. In proportions, panel A’s positive intercept indicates that the greatest source of differences was associated with autonomous, unobserved factors in the intercept, and women’s net nutrition after the transition was greater than men. The greatest source of observable characteristics was height, and after the transition, men had greater returns to height than women. Men’s advantage between BMI and height was greater with the transition to social feminism; however, the difference in average height favoured women. Nevertheless, the male BMI returns to height offset women’s returns to average characteristics. Although not as large, male returns to nativity and returns to average nativity were greater after the transition to social feminism. Alternatively, women’s returns to occupations were greater after social feminism, when social and economic forces aligned to compensate women more than men in socioeconomic status. Returns to age and residence also favoured women, however, were offset slightly by greater returns to average age and residence characteristics. Although the differences vary, the majority of women’s post-transition BMIs advantage was associated with average characteristics and not returns to characteristics.

Across-group Pre-transition Decomposition

Before the transition, men had greater BMIs compared with women associated with level differences, and this difference is supported by high-autonomous proportional BMI returns. From proportions, pre-transition decomposition in panel B of Table 3 indicates that men before the transition to social feminism had an advantage in both the levels and proportional intercept, indicating biological disparity between men and women favoured men. Before the transition, women had greater BMI returns associated with height, complexions, residence and nativity. Men had greater BMI returns associated with age and occupations. Before the transition to social feminism, characteristic returns to women’s height favoured women; however, cumulative net nutrition offset a small part of the returns to women’s height. Before the transition to social feminism, women also had greater returns to residence and nativity that were reinforced by returns to average characteristics. Nevertheless men had greater BMI returns associated with occupations and age. Before the transition, male BMI return advantages were supported by returns to socioeconomic status, and occupation returns and returns to age reinforced the male pre-transition advantage to current net nutrition (Bleakley & Costa, 2013, pp. 5–10). The majority of women’s pre-transition advantage was due to returns to height, and women had considerable return advantages from greater returns to average characteristics, such as age, nativity, residence and occupations.

Across-group Difference-in-decompositions

The female–male across-group difference-in-decompositions are shown in panel C of Table 3. If a component is positive, women did better than men with the transition to social feminism and negative if women did worse. From levels, women did better than men with the rise of social feminism, and from proportions, women did better than men due to non-identifiable sources in the intercept. These differences were attributable to nutrition, disease and urbanisation. However, the increase in the intercept also includes changes in the social, legal and political positions of women, which were associated with the rise of social feminism. Furthermore, women’s change in BMI associated with unidentifiable sources in the intercept was offset by men’s return to stature and cumulative net nutrition. Among observable characteristics, women did better than men due to age and occupation BMI returns, while male BMI returns were associated with nativity, complexions and residence. Women’s returns to average characteristics were greatest for height and complexions, whereas men’s returns to average characteristics were greatest for age, nativity, residence and occupation. While women’s overall BMIs and current net nutrition were greater with the transition, the positive autonomous intercept indicates that women’s BMIs increased with social feminism. However, women’s BMI returns associated with height, complexion, nativity and residence were greater relative to men before social feminism and women’s suffrage.

Within-group Difference-in-decompositions

Isolating female and male within-group differences before and after the rise of social feminism isolates how women after the transition compared women before the transition, relative to how men after the transition compared to men before with the transition. The female within-group decompositions are shown in panel A of Table 4 (Equations [23] and [24]). Panel B (Equations [27] and [28]) shows the male within-group decompositions. Panel C (Equations [29] and [30]) shows the female–male within-group difference-in-decompositions with the transition to social feminism.

Women’s Within-group Decomposition

Women’s within-group’s autonomous intercept difference indicates that their BMIs observed after 1900 were greater than women observed before 1900 transition. By proportion, women were better off after the transition to social feminism; however, the relationship between women’s BMI and stature was greater prior to the rise of social feminism. In addition to stature, women’s BMI returns associated with nativity were greater prior to the transition, and compositional differences were small. The largest contribution to the female–male within-group BMI difference was age and complexion, and the complexion average difference was small. Alternatively, women’s BMIs after the transition associated with age and residence were attributed to compositions. The structural difference between occupations was small, while the occupational difference was large, indicating much of the within-group female difference associated with occupations were compositional. The overall results for the within-group female decomposition are mixed.

Men’s Within-group Decompositions

The male within-group intercept difference indicates that men’s BMIs observed prior to the rise of social feminism were greater than men observed after. By proportions, men were considerably better off prior to social feminism. Independent of characteristics, men’s BMIs were higher after 1900, as illustrated by the positive autonomous intercept difference. Nevertheless, like women, men’s relationship between BMI and stature was larger prior to the rise of social feminism, and there was little difference between average statures, indicating the preponderance of the BMI return associated with male stature was attributable to structural differences. In addition to stature, male age-related BMI returns were greatest prior to the transition, with comparable compositional differences. It was male occupations after the transition to social feminism that had the greatest structural returns, with little differences in the composition component. While the difference was small, male BMI structural returns associated with residence, nativity and complexion were greater after the transition to social feminism, and there were small compositional differences associated with nativity. Nonetheless, compositional effects associated with age and residence distorted the male BMI relationship. Overall, male BMI results indicate that the male within-group BMI results were associated with structural returns to characteristics rather than compositional differences.

Within-group Difference-in-decompositions

The within-group difference-in-decompositions indicate the female BMI advantage with the rise of social feminism was greater than the male difference. Much of the increase in women’s BMIs after the rise of social feminism was attributable to non-identifiable characteristics after the rise of social feminism relative to before the transition. Moreover, the proportional within-group difference-in-decompositions intercept indicates the transition increased women’s BMIs relative to men independent of the changed in structural and compositional differences. Women after the transition were better-off than women before compared to the male difference associated with complexion and ages. However, differences in female returns to characteristics were offset by returns to average characteristics, indicating that the difference was due to compositional changes. Male BMI returns after 1900 were larger due to height, nativity, residence and socioeconomic status. Furthermore, the within-group male structural returns differences associated with height were greatest prior to the transition. Within-group differences-in-decompositions associated with complexions and age for women were greatest with the transition; however, complexion and age were similar, off-setting the structural and compositional difference.

Conclusion

Throughout economic development, the role of women and men’s relative living standards changed with the rise of social feminism. Prison records are one source to compare late nineteenth and early twentieth century physical measures; however, results using prison records must be taken with caution because criminals represent net nutritional conditions among the middle and lower classes, and there is a need for more studies for individuals in higher socioeconomic groups. In developing economies, greater emphasis is attributable to occupations where physical strength is required (Marques et al., 2018, p. 158); however, this emphasis on physical strength decreased as labour markets develop and less importance placed on physical strength and more emphasis on cognitive ability. Late nineteenth and early twentieth century US economic development improved the relative status of women relative to both men before and after the transition, and twentieth century women did better relative to nineteenth century women. The main sources of across-group variation were non-identifiable characteristics, height and nativity, indicating that relative net improvement in women’s net nutrition was related to non-observable factors, such as improved social standing and better conditions after passage is the Nineteenth Amendment. Nativity and socioeconomic status were the primary sources of female–male within-group variation, yet both of these favoured men relative prior to the rise of social feminism. Subsequently, twentieth century women’s net nutrition improved relative to men and relative to women in the nineteenth century with women’s suffrage and the rise of social feminism.

Footnotes

Acknowledgements

The author would like to appreciate comments from John Komlos, Tom Maloney and Paul Hodges. Shahil Sharma, Chinuedu Akah, Meekam Okeke, Ryan Keifer, Tiffany Grant, Bryce Harper, Greg Davis, Lee Carson and Kellye Manning provided research assistance.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author received no financial support for the research, authorship and/or publication of this article.

Note

References

Barker

D. J. P.

(1992). Fetal and infant origins of adult disease. British Medical Journal, 301(6761), 1111.

Becker

(1981). A treatise on the family. Harvard University Press.

Bereczki

, Reschler-Nicola

, Marcsik

, Meinzer

, & Baten

(2018). Growth disruption in children: Linear enamel hypoplasias. In Steckel

, Larsen

C. S.

, Roberts

, & Baten

(Eds.), The Backbone of Europe (pp. 175–197). Cambridge University Press.

Bleakley

, & Costa

(2013). Health, education, and income in the United States, 1820–2000. National Bureau of Economic Research.

Blinder

A. S.

(1973). Wage discrimination: Reduced form and structural estimates. Journal of Human Resources, 8, 436–455.

Bodenhorn

, Guinnane

, & Mroz

(2017). Sample selection biases and the industrialization puzzle. Journal of Economic History, 77(1), 171–207.

Brands

H. W.

(2010). American colossus: The triumph of capitalism,1865–1900. Anchor Books.

Buechler

(1990). Women’s movements in the United States: Women suffrage, equal rights, and beyond. Rutgers University Press.

Burkhauser

, & Cawley

(2008). Beyond BMI: The value of more accurate measures of fatness and obesity in social science research. Journal of Health Economics, 27, 519–529.

10.

Burnette

(2013). The changing economic roles of women. In Whaples

& Parker

(Eds.), Routledge handbook of modern economic history (pp. 306–315). Routledge.

11.

Bush

(1987). The theory of institutional change. Journal of Economic Issues, 21, 1075–1116.

12.

Card

, & Krueger

(1993). Minimum wage and employment: A case study of the fast-food industry in New Jersey and Pennsylvania. American Economic Review, 84(4), 772–793.

13.

Carlino

, & Sill

(2001). Regional income fluctuations: Common trends and common cyclers. Review of Economics and Statistics, 83(3), 446–456.

14.

Carson

S. A.

(2008). Health during industrialization: Evidence from the 19th century Pennsylvania state prison system. Social Science History, 32(3), 347–372.

15.

Carson

S. A.

(2009). Racial differences in body-mass indices of men imprisoned in 19th century Texas. Economics and Human Biology, 7(1), 121–127.

16.

Carson

S. A.

(2011). Height of female Americans in the 19th century and the antebellum puzzle. Economics and Human Biology, 9, 157–164.

17.

Carson

S. A.

(2012a). Family size, the physical environment, and socioeconomic effects across the stature distribution. Journal Homo of Comparative Human Biology, 63(2), 136–147.

18.

Carson

S. A.

(2012b). Nineteenth century stature and family size: Binding constraint or productive labor force? Review of Economics of the Household, 10(1), 39–52.

19.

Carson

S. A.

(2012c). The body mass index of Blacks and Whites in the United States during the nineteenth century. Journal of Interdisciplinary History, 42(3), 371–391.

20.

Carson

S. A.

(2013). Socioeconomic effects on the stature of nineteenth century US women. Feminist Economics, 19(2), 122–143.

21.

Carson

S. A.

(2014). The relationship between 19th century BMIs and family size: Economies of scale and positive externalities. Journal Homo of Comparative Human Biology, 65, 165–175.

22.

Carson

S. A.

(2015a). A weighty issue: Diminished 19th century net nutrition among the US working class. Demography, 52(3), 945–966.

23.

Carson

S. A.

(2015b). Biology, complexion, and socioeconomic status: Accounting for 19th century US BMIs by race. Australian Economic History Review, 55(3), 238–255.

24.

Carson

S. A.

(2016). Body mass index through time: Explanations, evidence, and future directions. In Komlos

& Kelly

(Eds.), Handbook of economics and human biology (pp. 133–151). Oxford University Press.

25.

Carson

S. A.

(2018a). Black and White female body mass index values in the developing late 19th and early 20th century United States. Journal of Bioeconomics, 20(3), 309–330.

26.

Carson

S. A.

(2018b). Net nutrition and the transition from 19th century bound to free-labor: assessing dietary change with differences in decompositions. Journal of Demographic Economics, 84(4), 447–475.

27.

Carson

S. A.

(2018c). The weight of 19th century Mexicans in the western United States. Historical Methods: A Journal of Quantitative and Interdisciplinary History, 51(1), 1–12.

28.

Carson

S. A.

(2019a). Late 19th and early 20th century native and immigrant body mass index values in the United States. Economics and Human Biology, 34, 26–38.

29.

Carson

S. A.

(2019b). Changing institutions, changing net nutrition: A difference-in-decompositions approach to understanding the US transition to free-labor. Review of Black Political Economy, 46(1), 65–94.

30.

Carson

S. A.

(2020). Net nutrition, insolation, mortality, and the antebellum paradox. Journal of Bioeconomics, 22, 77–98.

31.

Carson

S. A.

(2021). Nutrition within the household: 18th through early 20th century female and male statures. Journal of Biosocial Science, 1–22. 10.1017/S0021932021000250

32.

Costa

(1993). Height, wealth and disease among the native-born in the rural antebellum north. Social Science History Association, 17(3), 355–383.

33.

Davidson

, Rose

, Gutman

, Haines

, Condron

, & Condran

(2002). The quality of African-American life in the old Southwest near the turn of the 20th century. In Steckel

& Rose

(Eds.), The backbone of history: Health and nutrition in the western hemisphere (pp. 226–280). Cambridge University Press.

34.

Ellis

(2004). His excellency George Washington. Knopf.

35.

Fogel

R. W.

(1994). Economic growth, population theory and physiology: The bearing of long-term processes on the making of economic policy. American Economic Review, 84, 369–395.

36.

Fogel

R. W.

, Engerman

, Trussell

, Floud

, Pope

, & Wimmer

(1978). The economics of mortality in North America, 1650–1910: A description of a research project. Historical Methods: A Journal of Quantitative and Interdisciplinary History, 11(2), 75–108.

37.

Fogel

R. W.

, Engerman

, Floud

, Steckel

, Trussell

, Wachter

K. V.

, & Georgia. (1979). The economic and demographic significance of secular changes in human stature: The US 1750–1960 (Working Paper). NBER.

38.

Fortin

, Lemieax

, & Firgo

(2011). Decomposition methods in economics. In Card

& Ashenfelter

(Eds.), Handbook of labor economics (Vol. 4, Part A, pp. 1–102). Elsevier.

39.

Haines

, Craig

, & Weiss

(2003). The short and the dead: Nutrition, mortality, and the ‘antebellum puzzle’ in the United States. The Journal of Economic History, 63(2), 382–413.

40.

Himes

(2011). The demography of obesity. In Cawley

(Ed.), The Oxford handbook of the social science of obesity (pp. 35–47). Oxford University Press.

41.

Hodgson

(2009) On the institutional foundations of law: The insufficiency of custom and private ordering. Journal of Economic Issues, 43, 143–166.

42.

Jennings

(1992). Not the economy: Feminist theory, institutional change, and the state. In Dugger

, & Waller

(Eds.), The stratified state: Radical institutionalist theories of participation (pp. 117–152). Routledge.

43.

Komlos

(1987). The height and weight of West Point cadets: Dietary change in antebellum America. Journal of Economic History, 47(4), 897–927.

44.

Komlos

(2019). Shrinking in a growing economy is not so puzzling after all. Economics and Human Biology, 32, 40–55.

45.

Komlos

, & Brabec

(2010). The trend of mean BMI values of US adults, birth cohorts 1882–1986 indicates that the obesity epidemic began earlier than hitherto thought. American Journal of Human Biology, 22(5), 631–638.

46.

Komlos

, & Carson

S. A.

(2017). The BMI values of the lower classes likely declined during the great depression. Economics and Human Biology, 26, 137–143.

47.

Kose

, Kuka

, & Shenhav

(2021). Women’s suffrage and children’s education. American Economic Journal: Economic Policy, 13, 374–405.

48.

Marques

, Matos

, & Meinzer

(2018). Proliferative periosteal reactions: Assessment of trend in Europe over the past two millennia. In Steckel

, Larsen

C. S.

, Roberts

, & Baten

(Eds.), The Backbone of Europe (pp. 137–174). Cambridge University Press.

49.

McLannahan

, & Clifton

(2008). Challenging obesity. Oxford University Press.

50.

Moehling

, & Thomasson

(2020) Votes for women: An economic perspective on women’s enfranchisement. Journal of Economic Perspectives, 34(2), 3–23.

51.

North

(1990). Institutions, institutional change and economic performance. Cambridge University Press.

52.

Oaxaca

R. L.

(1973). Male female wage differentials in urban labor markets. International Economic Review, XIV, 693–709.

53.

Oaxaca

, & Ransom

(1999). Identification in detailed wage decompositions. Review of Economics and Statistics, 81(1), 154–157.

54.

Pope

(2021). Shape of BMI-mortality risk associations: Reverse causality and heterogeneity in a representative cohort of US adults. Obesity, 29(4), 755–766.

55.

Robb

(1994). Skeletal signs of activity in the Italian meal ages: Methodological and interpretive notes. Human Evolution, 9(3), 215–229.

56.

Schneider

(2017). Children’s growth in an adaptive framework: Explaining the growth patterns of American slaves and other historical populations. Economic History Review, 70(1), 3–29.

57.

Sokoloff

, & Villaflor

(1982). Early achievement of modern stature in America. Social Science History, 6(4), 453–481.

58.

Steckel

(1995). Stature and the standard of living. Journal of Economic Literature, 33(4), 1903–1940.

59.

Steckel

, & Rose

(2002). Patterns of health in the western hemisphere. In Steckel

& Rose

(Eds.), The backbone of history: Health and nutrition in the western hemisphere (pp. 563–582). Cambridge University Press.

60.

Teele

D. L.

(2018). How the west was won: Competition, mobilization, and women’s enragement in the United States. Journal of Politics, 80(2), 442–461.

61.

Waaler

H. T.

(1984). Height, weight and mortality: The Norwegian experience. Acta Medica Scandinavia, 679(suppl.), 1–51.

62.

Williams

, Meinzer

, & Larsen

C. S.

(2018). History of degenerative joing disease in people across Europe: Bioarchaeological inferences and lifestyle and activity from osteoarthritis and vertebral osteophystosis. In Steckel

, Larsen

C. S.

, Roberts

, & Baten

(Eds.), The Backbone of Europe: Health, diet, work and violence over two millennia (pp. 253–299). Cambridge University Press.

63.

Yun

M.-S.

(2008). Identification problem and detailed Oaxaca decomposition: A general solution and inference. Journal of Economic and Social Measurement, 33, 27–38.