Beliefs and Social Structure: Determinants of Female Labour Participation in an Ecuadorian Andean Community

Abstract

The aim was to identify the determinants of female labour participation and willingness to work in one of the poorest agricultural areas of the Ecuadorian Andes with a high indigenous population. A stratified random sample of 268 women by parish was used to collect the information. Social, demographic, and economic variables, as well as norms, values, beliefs, and social structure were consulted. A hypothesis test of means was used to identify possible differences in cultural variables. In addition, the identification of the determinants of labour market insertion was estimated through discrete choice models (logit), selecting the one that best classified the data by means of the confusion matrix. Significant differences were identified in the beliefs analysed based on prejudice towards women’s work (gender roles), attitudes towards children, and intersectionality (discrimination based on ethnicity) by estimating mean differences, considering education, marital status, ethnicity, and labour insertion as grouping variables. The results of the estimations indicate that female labour force participation depends on age, marital status, experience, number of children, education, ethnicity, head of household, social structure (ethnicity), and dimensions of beliefs and values. Willingness to work is explained by attitude towards children, experience, and age. Beliefs and social structure can therefore enhance female labour.

1. Introduction

Ecuador at the macro level has been incorporating the gender perspective in its development plans, proposing equality, cohesion, and social integration to guarantee a dignified life with equal opportunities for all people. Despite this, Ecuador is a multi-ethnic and multicultural country and there are few studies that address this perspective, which is relevant given the indigenous component of the population, especially in the Andean region, which leads to beliefs and social norms that limit the transformation of traditional gender roles (Mosquera 2018), negatively impacting female labour participation, even more so in rural and agricultural areas.

In this sense, the canton of Guamote, one of the poorest areas of Ecuador, located in the Ecuadorian Andes, is characterised by the fact that more than 90% of its population is indigenous. Its main economic activity is agriculture for self-consumption or sale in nearby markets, with strong male migration in search of sources of employment, which has implied a feminisation of agriculture, scarcity of job opportunities, low level of schooling, and low access to health.

Analysing female labour participation associated with traditional factors such as age, marital status, educational level, being head of the household, and in particular, the impact of cultural factors associated with attitudes towards children, marriage, and family, allows for an understanding of the phenomenon of female labour in this rural context with a strong ethnic component. This allows for the design of local public policies that contribute to the improvement of labour opportunities and transformation of gender roles, favouring changes in intergenerational beliefs that lead to a greater appreciation of female labour and less discrimination.

Understanding the phenomenon of women’s labour participation in this canton is relevant in the sense that it is one of the poorest areas of Ecuador, with many economic and social inequalities, linked to a high indigenous component that limits women’s empowerment derived from a patriarchal system, the feminisation of agricultural work, and scarcity of educational and employment opportunities.

In this way, the role of the social structure and cultural factors determine to a large extent the public policies of local governments to improve living conditions from a gender perspective, respecting the norms, beliefs, and customs of the area.

The research approach begins with a review of the literature on the subject, from the theories that explain female labour insertion to the empirical evidence of the influence of cultural, social, and demographic factors. Secondly, the methodology used is presented, which is based on a sample of 268 women in the canton of Guamote, both in rural and urban parishes. Hypothesis testing and modelling techniques are also proposed.

The results are divided into four parts: characteristics of the sample, beliefs and social structure, determinants of the likelihood of work participation, and determinants of willingness to work. Finally, these results are analysed in the light of the findings of other authors and conclusions are drawn.

2. Literature Review

The results achieved by economic growth in recent decades, deterioration of the environment and its ecosystem services, as well as inequality gaps (Sachs 2015), evident for example in Latin America, have made it possible to rethink the concept of development to achieve sustainable development, whose goals are oriented towards the preservation of nature or social well-being. The goal is that economic development moves towards the recognition of nature and its ecosystem services as an additional factor of production that has limits and is scarce. This more holistic conception of development is a complex process that demands the interrelation of at least the four dimensions that comprise it (Szauer and Castillo 2003; Sepúlveda 2008).

These dimensions are the environment, the economy, society, and institutions. An environmental dimension is required to: (1) conceptualise nature as an additional factor of production, just as capital or labour can be, (2) preserve ecosystem services for regulation and provisioning, and (3) understand that without them it is not possible to maintain the current quality of life or reduce existing gaps without generating more pressure on planetary boundaries (Sepúlveda 2008; Riestra 2018).

For its part, the economic dimension requires rethinking traditional business and production models to move towards circular, collaborative, and social economy models (third sector), which generate less impact on the environment and more economic growth oriented towards social welfare (Prieto-Sandoval et al. 2017; Riestra 2018).

The socio-cultural dimension, although related to the reduction in gaps in terms of access to education, health, and opportunities, from a gender perspective, is closely related to culture, norms, values, relationships, and networks that hinder or facilitate the transformation or valorisation of traditional gender roles and possibilities for women’s employment and empowerment.

The integration of these three dimensions is pursued through the institutional dimension, which aims to establish clear rules to generate incentives, at least in terms of gender, to create the necessary normative conditions for equity. While achieving equity is a necessary but not sufficient condition for achieving gender equality, it requires equal conditions of treatment and opportunities, which are not necessarily intended to change the gender division of labour (roles), but rather equality of social value and remuneration, which is addressed from a gender perspective.

In general terms, equal access to education, health, resources, and technology are the starting point for facilitating or hindering the generation of and access to employment opportunities, which would allow the women to have greater empowerment, independence in decision-making, confidence, and control over their lives.

Female labour force participation, therefore, is a complex and multifactorial problem, which must be approached from different disciplines, giving rise to theories that examine its determinants and explain the phenomenon. These theories provide an economic, sociological, psychological, feminist, organisational and demographic perspective on the problem, which are not mutually exclusive, but rather complement each other to strengthen understanding and overcome their individual limitations (Figure 1).

From an economic perspective, the theories associated with human capital (Becker 2009) stand out, in which it is proposed that investment in education and training increases job opportunities and wages, which in the case of women leads to greater labour participation. In the segmented labour market, two sectors can be identified: a primary sector with high conditions and salaries, where participation is mainly male, due to the existence of barriers and discrimination; and a secondary sector without labour benefits and low salaries, with greater female inclusion (Reich et al. 1973). Economic dependence explains that the decisions of whether to insert women into the workforce is not linked to their own preferences, but rather to the degree of economic dependence of their partners (Folbre and Nelson 2000; Bergmann 2005).

In addition, discrimination constitutes a barrier to entry into the labour market for women and extends to promotion and wage setting. Home economics argues that the incorporation of technology in domestic work has made these activities more flexible and facilitated the possibility of female insertion into the labour market (Goldin 2021).

These theories have led to the identification of economic determinants of female labour participation, highlighting education and training, labour market conditions in terms of job quality and availability, labour market policies and regulations associated with minimum wage, unemployment insurance and social security, occupational segregation that allocates only certain positions, and activities and occupations to the female labour force, as well as wage inequality that limits promotion and opportunities for women, family-friendly policies (parental leave, flexible working hours, and childcare) and migration, which impacts on the recognition of educational levels and experience.

Theories from the sociological perspective include the gender role theory, which states that traditional roles and stereotypes create a bias in women’s decision-making, as women must be responsible for domestic activities, childcare, and community management, limiting access to education and thus to job opportunities, which demands a gender-neutral position in organisations (Oakley 1974; Hochschild and Machung 1989; Acker 1990). This is exacerbated by the factors analysed by social structure theory, in which class, race and ethnicity exacerbate gender roles, widening the gap between men and women (Bergmann 2005).

This social structure phenomenon is reinforced through the proposal of the theory of socialisation and intersectionality. The first is based upon the fact that women’s labour participation is determined by socially acquired norms and values that determine expectations and beliefs regarding women’s work, which has been called the gender schema (Bales and Parsons 1956; Lipsitz 1981). The second deals with how work experiences are influenced by a host of social factors and structures, most notably race and class.

To these sociological theories, one can add the theory of social capital, in which horizontal or vertical networks and relationships facilitate access to resources and the appropriation of the benefits derived from them. It is usually indicated that women have little social capital, which limits their possibilities of education, employment, and the options of generating associative enterprises (Bourdieu and Richardson 1986; Coleman 1990; Putnam 1993).

These relationships are based on trust, norms and values (Putnam 1993), which is why women tend to have stronger relationships with family and friends (bonding) and weaker or non-existent relationships in terms of quality and quantity (bridging) or hierarchical relationships (linking), as stated by Woolcock and Narayan (2000).

These theories have made it possible to identify sociological factors that condition female labour market insertion and are linked to social norms, gender roles, stereotypes and cultural expectations, access to networks, discrimination, and prejudice in the workplace.

With respect to gender roles and traditional stereotypes, women are seen as the primary caregivers and men as the breadwinners, while social norms and cultural expectations around gender roles and appropriate behaviours for women can affect their ability to participate in the labour force and succeed in their careers, which is reinforced by family dynamics, such as the availability and involvement of a partner or spouse in household activities (Das and Mahanta 2023).

In addition, race, ethnicity, and immigrant status can interact with gender to shape women’s opportunities and experiences in the labour force, further limiting women who belong to certain ethnicities and races, such as indigenous and black women, respectively.

Psychological theories focus on understanding how self-beliefs, social interactions, and unconscious biases influence women’s employment opportunities and decisions. Women who believe in their capabilities (self-efficacy) have greater motivation that drives them to obtain and maintain employment, which is conditioned by the influence of women’s work experiences and trajectories (Subich 1998) from their environment and social networks (social cognitive career).

As Ryan and Deci (2000) argue, women’s career decisions are determined by their expectations of success in a particular area of work and the perceived value of that field of work (expectancy value), but there are also perceptions and behaviours that generate biases at the societal level (Greenwald and Banaji 1995), which tend to consider women less skilled or qualified than men (implicit bias).

In addition, feminist theories have provided explanations of women’s labour dynamics from the point of view of patriarchy and dual systems, reinforcing considerations made by other disciplines regarding the influence of race, ethnicity, class, discrimination, and inequality that in some cases are not addressed by regulatory institutions and policy frameworks.

Specifically, patriarchy theory argues that women’s experiences in the labour force are determined by the patriarchal structure of society, which is characterised by male domination and the oppression of women (Friedan 1963; Firestone 1970; de Beauvoir 2010). According to this theory, women’s employment opportunities are limited by patriarchal values and beliefs that see women as inferior to men and as primarily responsible for domestic and caring roles.

This theory is complemented by the dual systems theory, in which Hartmann (1976) and Eisenstein (1979) argued that capitalist and patriarchal systems work together to create a dual system of oppression for women that limits their opportunities and experiences in the labour force. They also proposed that the intersection of these two systems creates a unique form of exploitation and oppression for women that is different from that experienced by men.

There are additional factors that can be analysed in a more managerial setting, linked to organisations, that place the emphasis on career paths, working conditions, diversity policies and job descriptions associated with gender roles, which have given rise to the theory of the glass ceiling, tokenism, the dual-career couple, diversity management, and gender role spillovers.

In the first case, the theory holds that women face invisible barriers that prevent them from advancing to higher-level positions in organisations, despite their qualifications and efforts (Reskin 1988). According to this theory, these barriers are often subtle and unconscious and created by organisational practices and cultures that favour men. In the second, women are often the only or one of the few in a particular position or group (Moss and Kent 1996).

The third examines the impact of dual-career couples on organisational policies and practices, such as recruitment, retention and promotion, while indirect gender role effects argue that gender roles and expectations can affect how people are perceived and treated in the workplace. According to this theory, women may be disadvantaged due to gender roles and stereotypes, which can lead to discrimination and prejudice in the workplace. Finally, diversity management theory suggests that organisations can take proactive steps to manage diversity, including gender diversity, to improve organisational performance and create a more inclusive and equitable workplace.

Considering all theories, empirical evidence from studies of female labour force participation linked to socio-economic factors does not allow for the identification of unique behaviours, as they depend on the conditions of the population under study.

Most research incorporates traditional variables associated with age and marital status (Contreras and Plaza 2010; Eckstein and Lifshitz 2009; Rodríguez-Garcés and Muñoz-Soto 2015), in addition to household headship (Rodríguez-Garcés and Muñoz-Soto 2015, 2018) and family income level (Castro et al. 2022). Added to this are human capital variables measured through education or training (Law and Wye 2023) and the combination that can be generated with variables such as the number of children (Rodríguez-Garcés and Muñoz-Soto 2015).

In this context, Eckstein and Lifshitz (2009) managed to show that educational level, as an observable variable, explains a good part of female participation and the difference is determined in married women by unobservable variables associated with changes in domestic activities and the financial burden of raising children and even of age. Some research seems to show that the number of children under the age of five per woman has a negative impact on the probability of forming part of the labour market (Contreras and Plaza 2010; Rodríguez-Garcés and Muñoz-Soto 2018; Shittu et al. 2019), but it is influenced by the level of education (human capital), so that fertility decreases its impact as education improves (Kouogueng 2016). Thus, there is empirical evidence of a positive relationship between educational attainment and female labour force participation, regardless of the number of years of education (Rodríguez-Garcés and Muñoz-Soto 2018; Gupta 2023), which is affected by marital status or raising young children (Gershon and Nwonuala 2021).

When including cultural factors and social behaviour aimed at reinforcing machismo in the studies, for the case of Chile, a reduction in the female participation rate in the labour market is evident, which could offset the positive effect of educational level (Contreras and Plaza 2010). This effect was evaluated by Espino and Sauval (2016), also in Chile, identifying that cultural factors constitute a barrier to entry into the labour market for women, associated with traditional gender roles that encourage informal working conditions or lower wages compared to men.

Rodríguez-Garcés and Muñoz-Soto (2018) analysed the effect of social norms, perceptions towards children, subjective evaluations or prejudices towards women’s work, attitudes towards marriage and work-family balance, such that a more conservative attitude towards marriage, child-rearing, and the difficulties of reconciling family and work are obstacles to female labour market entry. These perceptions may evolve over time in an intergenerational manner in the sense that as female labour participation increases, the benefits and value of work are transmitted to children, diminishing the obstacles associated with culture and facilitating the transformation of gender roles (Farré and Vella 2013).

3. Materials and Methods

For the study of female labour participation and willingness to work, the analysis of labour conditions, socio-economic conditions and conditions associated with social norms and values, an instrument was applied in September and October 2022 to a sample of 268 women, including three parishes: one urban (La Matriz) and two rural (Cebadas and Palmira).

The sample was estimated at 95% confidence and 5.85% for an economically active female population of 8645. Initially a 5% error was considered, but restrictions of accessibility, language, and willingness to answer the questionnaire made it necessary to assume a larger error.

The stratification of the random sampling was distributed among the three parishes according to the proportion of the economically active population in each, so that Cebadas had 50 women, Palmira 72, and La Matriz 144.

Parishes are the lowest political-territorial division in Ecuador. There are 1499 parishes in Ecuador (1140 rural and 359 urban). The urban parish is one that is circumscribed within the metropolis or city; it has all the infrastructure necessary to be a major city. Rural parishes are those that are separated from the main city or metropolis; they tend to be comarcas or a group of areas whose inhabitants live from agricultural work and the countryside.

The instrument was divided into three sections. The first one related to socio-economic and family conditions that are traditionally consulted in labour market insertion studies, such as age, marital status, children, household headship, educational level, ethnicity, and work experience. The second section related to work status, the type of economic activity carried out, the level of income received, the number of hours worked, and whether access to employment benefits is available. If unemployed, the reasons and duration of unemployment were asked, as well as the willingness to look for a job. The last section, based on the proposal by Rodríguez-Garcés and Muñoz-Soto (2018), was related to cultural aspects in which 14 statements were evaluated on a Likert-type scale, where 1 is totally disagree, 2 is disagree, 3 is neither agree nor disagree, 4 is agree, and 5 is totally agree.

The statistical validation of the instrument was carried out through Cronbach’s alpha, whose results show values above 0.70 for each section, as suggested by the literature. The first section obtained 0.76, the second 0.83 and the third 0.86, guaranteeing the internal consistency of the instrument.

Based on these approaches, the dimensions of prejudices about women’s work, attitudes towards children, and aspects of intersectionality associated with the link between gender roles and the valorisation of women’s work were analysed. In the first case, questions were asked about whether men are the providers of household resources, the equality of productivity between men and women, the relationship between women’s work and child rearing, divorce, and women’s productive role.

In the second case, the issues were associated with the satisfaction of parenting and freedom associated with parenting, financial burden of parenting, importance of children in the family, and responsibility of children for the care of adults. The last dimension included limited opportunities and job discrimination for women, as well as the influence of indigenous ethnicity on these opportunities, pay, and discrimination.

Given that the variable of interest is dichotomous, a value of 1 indicated that women participate in the labour market and zero that they do not. The estimation considered a discrete choice model of logit or probit type, in which case the selection was based on the ability to correctly classify the data (accuracy), the estimation not being feasible through a linear regression model because it is not guaranteed that the estimated values are between 0 and 1, a condition relative to the predicted probability. The errors are not normally distributed, and the variances are homoscedastic.

Thus, the estimation proposed corresponds to the one proposed in Equation (1):

$${lab}_i={\beta }_o+{\beta }_1{x}_i+{\beta }_1{z}_i+{\epsilon }_i$$

(1)

where female labour participation (${lab}_i$) depends on a set of socio-economic and family variables ($x_i$) and social norms, attitudes, and prejudices ($z_i$).

The difference in the estimation of the probit or logit models lies in the distribution function used, either normal in the first case or logistic in the second, but in both cases, it is guaranteed that the estimated variable is indeed a probability ranging between 0 and 1.

From the estimation, the marginal effects were analysed either in elasticities or partial derivatives, depending on the measurement of the explanatory variables, correct specification of the data through the confusion matrix, sensitivity and specificity, as well as the regression diagnostic curve (ROC).

Similarly, the determinants of willingness to work in women who are not employed were identified. Since it is a dichotomous variable that takes the value 1 if willing and zero if not, a discrete choice, logit or probit model was estimated (Equation (2)):

$${willing}_i={\beta }_o+{\beta }_1{x}_i+{\beta }_1{z}_i+{\epsilon }_i$$

(2)

where female labour participation (${willing}_i$) also depends on a set of socio-economic and family variables ($x_i)$ and on beliefs, prejudices, and social structure ($z_i$).

4. Results

4.1. Characteristics of the Sample

From the results obtained, the women are characterised by being young, married, not being heads of household, having at least one child, and self-identifying as indigenous. Indeed, 84.33% of the women surveyed are under 45 years old and more than half are under 30 years old, which is characteristic of the population in the sector, where most people are young. On marital status, 52.99% indicate that they are married, of which only 15.44% are heads of household, while among single women, the percentage rises to 33.33%. Of the proportion indicating that they are married, the majority have at least one child (90.14%), although 52.34% have more than three children, while single women have either no children (81.17%) or between one and two (18.26%).

As indicated, the predominant ethnic group in the area is indigenous women who are part of the Puruhuá people (86.57%), while a few mestizo women participated (13.43%). Of the married women, 97.18% are indigenous, with only 24.32% being heads of household. If we compare the proportion of indigenous women with more than three children with mestizo women, we can also identify a difference in favour of the former, 31.89% in this case as opposed to 11.11% in the case of mestizo women.

An important variable that needs to be evaluated when working with labour insertion in accordance with the theory of human capital is the level of education, whether formal or through training. In this case, the results show that 12.31% have no education at all, 29.48% have only primary education, and 39.18% have completed secondary school. The proportion of women with tertiary-level education is 18.66% and only 0.37% have a post-graduate-level education. All the women who report having no education at all are indigenous.

If we also link the number of children with the level of education attained, we can see from the tests of independence for qualitative variables that there is a statistically significant inverse relationship: the higher the number of children, the lower the level of education. Women who have more children are those who have a lower educational level; in fact, most women who have between three and four children have not completed any level of education or only primary education (86.4%), while this proportion in women who are mothers of five or more children rises to 94.12% (

Table 1. Relationship between educational level and number of children.

	Number of Children
Level of Education Attained	None	1 to 2	3 to 4	5 or More
None	0.00%	9.89%	18.18%	47.06%
Primary	5.05%	30.77%	68.18%	47.06%
Secondary	59.60%	41.76%	13.64%	5.88%
Tertiary	34.34%	17.58%	0.00%	0.00%
Post-graduate	1.01%	0.00%	0.00%	0.00%
Total	100%	100%	100%	100%

).

Women who do not have any children are mostly those with secondary (59.60%), tertiary-level (34.34%) and post-graduate (1.01%) education. The only woman who reports having post-graduate education does not have any children.

When asked about their current educational situation, only 23.13% indicate that they are still studying, while those who dropped out of school report that they did so due to lack of resources (50%), work or domestic chores (13.11%), age (11.17%) and pregnancy (9.22%), among other reasons.

If these results are analysed according to ethnicity, 97.08% of the women who dropped out of school due to lack of resources were indigenous, as well as for pregnancy and domestic work. Mestizo women are more likely to report having dropped out of school due to the natural completion of their education cycle.

When analysing the characteristics associated with working conditions, it is evident that 58.96% have no experience, and those who do have experience have an average of 7.74 years of experience. On employment, 59.33% indicated that they did not work and those who did work were mainly in commerce (38.53%) and agriculture (21.10%). In terms of remuneration, 63.30% receive the equivalent of less than one basic salary, 25.69% receive one basic salary, and only 11.01% receive more than one basic salary, so that 55.96% consider that the income they generate is not enough to cover their basic needs.

If we also consider labour benefits and hours of work, we find that 55.97% work more than 8 hours a day and 75.23% do not have any insurance, neither “peasant” nor social security, so that 77.06% do not receive any type of labour benefits.

Comparing these general results with respect to each ethnic group, the proportion of indigenous women who work is 37.31%, while the proportion of mestizo women is 58.33%. In the first case, the proportion of women working in commerce and agriculture is higher, as are those who receive less than the basic salary, work more than 8 hours, do not have insurance, and do not have employment benefits (

Table 2. Labour characteristics by ethnicity.

	Indigenous	Mestizo
In work	37.93%	58.3%
Agriculture and trade	70.45%	14.29%
Less than basic salary	67.05%	23.81%
More than 8 hours work	56.82%	52.38%
No insurance	78.41%	61.90%
No employment benefits	82.95%	52.38%

).

Thus, although indigenous women work less than mestizo women, their working conditions are more precarious, not only in terms of remuneration but also in terms of employment benefits and the number of hours they work. In this aspect, Mosquera (2018), in his study for some areas of the Ecuadorian Andes, explains that women dedicate more time to work than men, and this is even more so in indigenous communities linked mainly to agricultural activities in rural areas.

With respect to women who do not work, 61% do not seek employment, indicating that the reasons for not working are linked to having household care duties (44.65%) and a lack of opportunities (20.75); hence, they devote their time to raising small animals or agricultural production for self-consumption.

Compared by ethnicity, 47.9% of indigenous women do not work to devote themselves to household chores, while among mestizo women, only 13.3% do not work. Of the proportion of women who do not work, 62.5% of indigenous women do not seek employment, while among mestizo women this proportion drops to 46.7%.

4.2. Beliefs and Social Structure

The social norms and values that create women’s perceptions of their productive role, child-rearing, and discrimination in the workplace linked to their ethnic group and social structure were addressed through approaches that are grouped into three dimensions: the first relating to prejudice towards women’s work, the second to attitudes towards children, and the last to intersectionality or the influence of being indigenous on opportunities, remuneration, and discrimination.

Standardising the Likert-type scale to a value between zero and one, results closer to zero indicate less agreement with the statements and thus fewer obstacles in terms of women’s perceptions of achieving labour market insertion. In more conservative social structures, in which traditional gender roles are more entrenched, the social valuation of women’s work is lower, and ethnicity generates a certain degree of differentiated behaviour. These are reflected in values close to 1.

Thus, when analysing the 14 statements, the perceptions that reflect a more conservative attitude (values above 0.50) are associated with the great importance of children in the home and high level of responsibility they must have in the care of adults, and the high level of discrimination against women in general and indigenous women in particular.

These results are reflected in the value of the index constructed from the average of the approaches (0.6363) and in the overall assessment of the dimensions. In the case of attitude towards children and intersectionality, these are above 0.50, which shows the conservative behaviour of the women, as well as the social structure that recognises the obstacles in labour insertion linked to the upbringing of children and discrimination by gender and ethnicity.

In the case of attitude towards children, a value of 0.6333 was obtained and in intersectionality 0.7414, while in prejudice towards women’s work related to the value that women assign to this role compared to men, the results show an average behaviour of 0.5087, which does not really show a less conservative perception either.

If we compare these results by ethnicity, we observe that in the three dimensions the value is higher for indigenous women than mestizo women, and these differences are statistically significant in the prejudice towards work, in the attitude towards children and in the overall index. Although the difference exists in intersectionality, it is not significant (

Table 3. Hypothesis testing for the difference of means of the dimensions by ethnic group.

Dimension	Indigenous	Mestizo	Difference
Prejudice towards women’s work	0.5096	0.4212	0.0883	***
	(0.0065)	(0.0196)	(0.0184)
Attitude towards children	0.7471	0.6796	0.0674	***
	(0.0072)	(0.0179)	(0.0196)
Intersectionality	0.7436	0.7277	0.0158
	(0.009)	(0.017)	(0,023)
Overall index	0.6440	0.5861	0.0579	***
	(0.0042)	(0.0121)	(0.0116)

Note: Values in brackets represent standard errors; significance level at 1% (***).

).

The difference in the assessment of the index and its components when comparing working and non-working women is in favour of the former, in the sense that they are less conservative, except in the case of the attitude towards children, which is slightly higher in non-working women, although in none of the cases are statistically significant differences identified.

If we compare the results obtained considering the condition of household headship, we obtain statistically significant differences in prejudice and attitude towards children but not in intersectionality and the overall index. Nonetheless, these differences do not always indicate that women who are heads of household believe that there are fewer obstacles to women’s labour market insertion (

Table 4. Hypothesis testing for the difference in means of the dimensions by household head status.

Dimension	Head of Household	Not Head of Household	Difference
Prejudice towards women’s work	0.4797	0.5025	−0.0227	*
	(0.0142)	(0.0073)	(0.016)
Attitude towards children	0.7714	0.7292	0.0421	***
	(0.0181)	(0.007)	(0.0166)
Intersectionality	0.7564	0.7375	0.0188
	(0.0222)	(0.008)	(0.019)
Overall index	0.6410	0.6350	0.006
	(0.1075)	(0.004)	(0.102)

Note: Values in brackets represent standard errors; significance level at 1% (***) and 10% (*).

).

Indeed, in the case of prejudice towards work, the value is higher for women who are not heads of household, considering that the productive role can also be assigned to women and there are no differences in terms of productivity by gender. In the rest of the dimensions and overall index, the value is higher for female heads of household.

Comparisons by educational level and marital status also show statistically significant differences for each category and dimension, as well as for the overall index. In the first case, the ratings are higher for women with no education at all and decrease with higher education, while in the second case, women in civil partnerships along with divorced and widowed women rate the attitude towards work and intersectionality more highly, while married women rate the attitude towards children more highly.

4.3. Determinants of the Likelihood of Work Participation

To identify the determinants of the probability of female labour force participation, as explained above, discrete choice, logit or probit models were used, considering the set of variables associated with beliefs and social structure, as well as those associated with human capital theory (education and experience) and the socio-economic variables that are traditionally incorporated. The results in terms of coefficients and marginal effects obtained from the partial derivatives for the logit estimation (selected) are presented in

Table 5. Estimation of female labour force participation through logit model.

	Logit Model
Variable	Coefficients	Marginal Effect
Age	0.2270	0.0403	**
	(0.098)	(0.017)
Marital status	−0.4042	−0.0718	**
	(0.188)	(0.032)
Experience	1.9140	0.3400	***
	(0.299)	(0.035)
Children	−0.4629	−0.0822	**
	(0.227)	(0.039)
Prejudice and education	0.5680		*
	(0.342)
Prejudice		0.2705	*
		(0.160)
Education		0.0513	*
		(0.030)
Intersectionality and ethnicity	0.3510		*
	(0.194)
Intersectionality		0.0879	*
		(0.047)
Ethnicity		0.0462	*
		(0.025)
Attitude towards children	1.3842		**
	(0.617)
Attitude towards children		0.0533	**
		(0.022)
Head of household		0.1554	**
		(0.067)
Constant	−2.2298		***
	(0.679)

Note: Values in parentheses represent standard errors; significance level at 1% (***), 5% (**), and 10% (*); marginal effects correspond to partial derivatives.

.

The results included the variables age, marital status, experience, number of children, and three factors that were constructed from the interaction of female work bias and education, intersectionality, and ethnicity, as well as attitude towards children and head of household, considering the behaviours discussed in the previous section.

In order to select the model that best classifies the data, the confusion matrix was used (

Table 6. Confusion matrix for the logit model of female labour force participation.

		Logit Model
Classification of the model		Observed Values		Total
		Work	No work
Work		71	31	102
No work		38	128	166
Total		109	159	268
Sensitivity		65.14%
Specificity		80.50%
Correctly classified		74.25%
Correctly classified and adjusted for the short point probability	76.12%

), in which the observed values are compared with the values estimated by the model in terms of whether women work, and the model effectively classifies them as such (sensitivity); and when they do not work, the model also defines them as such (specificity).

The probability cut-off point for dividing working and non-working women is 0.50, but it can be adjusted according to the model to use the one in which the difference between sensitivity and specificity is zero. When this adjustment is made, the correct classification in the models is equal (76.12%), which is why the logit model is selected, given that a greater number of variables are significant.

Thus, according to the results obtained from the logistic regression, the probability of female labour participation increases with an increase in age, experience, educational level, or a low number of children; or being mestizo (mixed-race), head of the household, or single.

For the dimensions used in all estimations, concerning women’s social beliefs about women’s work prejudices, attitudes towards children and intersectionality, a transformation of the scale was made so that higher values indicate more liberal perceptions and facilitate interpretation. These dimensions were found to be statistically significant in explaining the probability of labour participation, in the sense that an improvement in the perception of the value of women’s work and possibility of carrying out productive activities; as well as a less conservative attitude towards children, a reduction in the perception of discrimination, and the scarcity of opportunities considering the ethnic component; have a positive impact on the probability of employment.

The graphical analysis of these results is presented in terms of the correct classification of the selected model (logistic regression), showing sensitivity as the correct classification of working women and specificity the correct classification of non-working women, as well as the short point of the probability of assignment or classification with which the accuracy was adjusted, which as the literature indicates must exceed 70% in order for us to accept the model (Figure 2).

Additionally, the regression diagnostic plot (ROC) shows the ability to classify the data correctly and is evaluated through the area under the curve, with a value of 1 being the perfect prediction and a value of 0.50 being the minimum acceptable value to classify women workers as workers. In the case of the regression, the area under the curve is 0.7862, which means that the model is able to adequately predict labour market insertion (Figure 3).

4.4. Determinants of Willingness to Work

A similar approach was used to analyse the probability of being willing to work in terms of the variables associated with cultural factors and those proposed by economic and sociological theories that explain the behaviour of the female labour market. The estimation was performed both in terms of logistic regression and the probit model, as well as marginal effects as partial derivatives, presenting the results of the logit model that was selected in the

Table 7. Estimation of willingness to work through logit model.

	Logit Model
Variable	Coefficients	Marginal Effects
Prejudice	−1.8729	−0.3190
	(1.589)	(0.267)
Attitude towards children	6.5202	1.1107	**
	(2.993)	(0.486)
Intersectionality	0.5250	0.0894
	(1.494)	(0.254)
Experience	0.9711	0.1654	*
	(0.548)	(0.091)
Education	−0.2072	−0.0035
	(0.294)	(0.050)
Age	−0.1576	−0.0268	*
	(0.095)	(0.0158)
Constant	2.0038
	(2.571)

Note: Values in brackets represent standard errors; significance level at 5% (**), and 10% (*); marginal effects correspond to the partial derivatives.

.

In the estimation, the variables that were found to explain the likelihood of being willing to work are associated with attitude towards children by cultural factors and with experience and age. A less conservative attitude towards children, more experience, and a younger age have a positive impact on the likelihood of being willing to work or seek employment.

While intersectionality and prejudice towards women’s work were not statistically significant, fewer barriers in terms of opportunities and discrimination (intersectionality) in general were perceived. The ethnic component raises the probability of willingness to work; the opposite is true for prejudice, wherein a less traditional position reduces the probability of willingness to work.

According to the correct classification of the data, the selected model corresponds to the logistic model, which allows us to classify women who work as workers (sensitivity) and those who do not work as workers (specificity) in 76.73% of the cases (

Table 8. Confusion matrix for willingness to work.

	Logit Model
Classification of the Model	Observed Values		Total
	Are willing	Not Willing
Are willing	113	32	145
Not Willing	5	9	14
Total	118	41	159
Sensitivity	95.76%
Specificity	21.95%
Correctly classified	76.73%

).

The graphical analysis first shows the sensitivity and specificity curves, as well as the optimal short point of the probability at which both are equalised, validating that the maximum values are between zero and one, and the sinusoidal shape is evident (Figure 4).

With respect to the diagnostic logistic regression plot (ROC), the area under the curve does indeed exceed the minimum allowed of 0.50, standing at 0.7074, so that the model adequately predicts the probability of willingness to work (Figure 5).

In this way, the logistic model allows us to explain the behaviour of the probability of women’s willingness to work or their intention to keep looking for a job, which in some cases exceeds one year, with a higher proportion of mixed-race women wanting to get a job.

5. Discussion

Female labour participation has been approached in this study from an economic and sociological perspective, considering in the former the elements provided by: human capital (Becker 2009), relating to education, training and experience; the segmented market (Reich et al. 1973), with the barriers to market access and more precarious employment benefits for women; and the theory of economic dependence (Folbre and Nelson 2000; Bergmann 2005), including aspects such as being head of the household and the theory of discrimination through some approaches to the intersectionality dimension (Goldin 2021).

With respect to the second, women’s perceptions that constitute their beliefs and expectations were analysed based on approaches which evaluated the possibility of transforming traditional gender roles and their link with the ethnic or social structure component, as proposed by authors such as Oakley (1974), Hochschild and Machung (1989), Acker (1990), and Bergmann (2005). Based on the theory of socialisation and intersectionality, the emphasis is placed on society and the impact it has on the construction of beliefs, perceptions, and patterns associated with gender roles and the valorisation of female work, reinforcing the feminist theory of patriarchy (Friedan 1963; Firestone 1970; de Beauvoir 2010).

In this sense, the variables used to measure the economic or demographic component that is traditionally incorporated, as proposed by Contreras and Plaza (2010), Eckstein and Lifshitz (2009) and Rodríguez-Garcés and Muñoz-Soto (2015, 2018), were age (Berniell et al. 2023), experience, marital status, being head of the household, educational level, and number of children, as well as those associated with working conditions such as salary, benefits, and hours dedicated to the working day.

To measure the social component explained through sociological theories, three dimensions associated with prejudice towards women’s work, attitude towards children, and intersectionality were considered. Prejudices tried to show women’s perceptions and beliefs regarding the possibility of transforming gender roles, equality in productivity, and the relationship between divorce and work and parenting and work. Attitudes towards children related to the financial and time burden of children, as well as the importance of children in marriage and caring for the elderly. Finally, intersectionality assessed women’s perceptions of the scarcity of opportunities, wage differentials and discrimination, both in general terms and the difference that may arise when considering ethnicity.

As can be identified in the literature (Kouogueng 2016; Rodríguez-Garcés and Muñoz-Soto 2018; Gupta 2023), relationships were established between explanatory variables, such as the influence of educational level on the number of children in the household or ethnicity on marital status, and even on being head of the household or not.

Additionally, according to Contreras and Plaza (2010), the social structure represented by ethnicity, as well as the more conservative beliefs of indigenous communities, offset the positive effect that education has on female labour participation; in fact, including the level of education in the model without the interaction with the dimension of prejudice is not statistically significant.

It is thus possible to prove that social structure influences norms and values, with indigenous women being more conservative than mestizo women, which is reflected in a lower rate of labour market insertion, lower wages, longer working hours, and fewer benefits (Das and Mahanta 2023). Thus, as proposed by Mosquera (2018), indigenous women have limited employment opportunities due to both the cultural component and their limited access to education, dedicating themselves to caring for the home, children, and agriculture. In Ecuador, this generates a process that has been called the feminisation of agriculture, used for self-consumption or sale at small fairs.

This reality in Ecuador means that the problem of the gap in labour insertion must be addressed not only from the point of view of the traditional variables, but also considering ethnic, social, and belief components through work on the land.

6. Conclusions

The results suggest that norms, beliefs, and social structure are determinants of the probability of labour participation among women in the canton of Guamote, showing an increase in this probability when women report fewer conservative behaviours and beliefs, which is determined by the ethnic group to which they belong, with mestizo women showing less patriarchal behaviour than indigenous women.

It is necessary to consider that the population under study are indigenous people of the Puruhuá ethnic group (86.57%) and mestizos (13.43%); there are no women who define themselves as white, Afro-Ecuadorian, or Montubian. The white population in Ecuador is less than 10%, which is why the constitution itself establishes respect for indigenous ethnic groups and defines the country as multicultural and plurinational, which makes it even more difficult to analyse the implications in terms of social structure.

The results in terms of other ethnic groups must be analysed in a particular way, as the patriarchal social structure is predominant in indigenous ethnic groups, where traditional gender roles are clearly defined, which limits women’s empowerment and thus the possibilities for economic independence. Women must be dedicated to the care of the household, children, and the elderly, as well as to agricultural activities and community management.

The process of feminisation of agriculture as a work alternative for women in rural areas, or self-employment, conditions labour insertion, as these are precarious activities, lacking in work benefits, established working hours, and pay commensurate with the minimum wage. The main reason for this phenomenon is based on the scarcity of job opportunities in these areas, limitations in terms of education and work experience, and of course impossibility of not attending to reproductive roles.

In addition to the variables explained through sociological theories, variables related to the economic perspective were incorporated, including age, education, number of children, experience, marital status, and being head of the household. We found, as reported in the literature, that the probability of female labour participation increases with increasing age, experience, educational level, or a low number of children; or being of mixed race, head of the household, or single.

With respect to the probability of willingness to work, it rises with a less conservative attitude towards children, more experience, and a younger age. So, while it is not possible to identify any element of social structure linked to ethnicity, it is possible to identify the perception of women in relation to children and how that perception affects the possibility of seeking employment.

Considering the results obtained, it is evident that not only traditional variables explain the probability of participating in the labour market and willingness to seek employment, but also variables associated with norms, beliefs, and social structure, even more so in areas with a significant component of certain ethnic groups, whose experiences and patterns are much more patriarchal.

Although the conclusions of the study coincide with the evidence found in the literature, they are limited to the reality of the Ecuadorian Andes, with high levels of poverty, an essentially indigenous or mestizo population and a social structure that responds to traditional gender roles and a patriarchal system. For further studies it would be necessary to compare the results with other areas of the country, such as the coastal region with different cultural characteristics.

Similar results could be found in other Latin American countries where different indigenous ethnicities are preserved, with social behaviours that correspond to those identified in this study, as is the case in countries such as Peru, Bolivia, and even Mexico.