Women empowerment refers to the procedure by means of which women acquire the ability to organize themselves in increasing their own self-sufficiency, to declare their independent right in making decisions and to control assets that will support in challenging and eradicating their own dependence (Keller, B., & Mbwewe, D., 1991). It has been specified by considering a multidimensional problem that can be designated by education, occupation, discussion of family planning with partners, decision making inside the household, free movement, marriage age, political representation and legal rights (Alsaawi, M., & Adamchak, D.J., 2000; Al-Riyami, A.A., & Afifi, M., 2003). Many national development plans are increasing their importance in the integration of women towards the development procedure and hence, their involvement in economic activities alongside men (Dixon, R.B., 1978). In order to achieve the targets set up by Sustainable Development Goals (SDGs) in 2015, which draws attention to guarantee the full and efficient participation of women and equal opportunities in all sectors, women empowerment has been considered as a prerequisite (UNDP, 2016; Molina, 2025).

It is a long-standing matter that in a society with strong patriarchy as Bangladesh, women are assigned a lower status as men who have the absolute power of administering households and society as a whole, whilst women are quite often confined to household duties (Balk, D., 1997). Bangladesh has made huge progression towards the workforce participation of women. The industrial sector, especially the apparel industry and services sector fueled the higher growth in the workforce participation of women though the most number of women are recruited in the agriculture sector. Women are becoming more engaged about their career than before in these times (Islam, N., & Khatun, J., 2019). However, the overall national rate is still too low and the rate falls behind the rate for male counterparts to a large extent in Bangladesh (Kokitula et al., 2019; Pomi, 2021).

Education can be considered the factor that assume more power or autonomy for women in our country (Paul et al., 2016). However, a large section of the Bangladeshi women are illiterate or low-educated. They rely heavily on their father or husband for fulfilling their needs. Because of this dependence, they have no choice but to follow decisions of their father or husband (Tabassum et al., 2019). Even after completing their education, a large number of women have been choosing to be homemaker. (Mahmood, F., & Rezina, S., 2016). Even though women's workforce participation is increasing, Lack of job opportunities is still apparent widely in rural areas despite that the participation of women in the workforce is increasing. (Kalam, I.M.S., & Amin, M.M.A., 2016).

It has been found that organizations having a greater number of women involved in the leadership roles perform better than organizations having lesser women in the leadership roles (Tyagi, S., 2015). However, women's participation in leadership/managerial positions has consistently been under the radar than men in these economically in dispensible industries of Bangladesh (Islam et al., 2017). Women become unproductive counterparts in society because of lack of financial independence, which also minimize the opportunities for them to contribute to economic development (Niaz, M.U., & Iqbal, M., 2019). Tradi-tionally, geographic mobility has been remarkably limited of the poor women in the rural sides of Bangladesh. Effective restriction of mobility for women is fairly severe today in many scenarios (Islam, N., & Khatun, J., 2019).

Some of the major reasons for women for not partaking in the workforce are domestic responsebilities, family-specific roles and commitments to their community. Work-family disputes are commonly ascribed to the fewer number of women in the workforce (Salleh, S.N.S., & Mansor, N., 2022). Becoming a parent is associated with a change to more traditional gender roles, with men expanding their involvement in the workforce and women reducing their participation in the workforce (Abroms, L.C., & Goldscheider, F.K., 2002; Baxter et al., 2008; Christie-Mizell, C., 2006; Gjerdingen, D.K., & Center, B.A., 2005; Sanchez, L., & Thomson, E., 1997). 

Another major constraint on women's workforce participation have been identified in the form of long standing health issues among women (Majeed et al., 2014; Pit, S.W., & Byles, J., 2012). Apart from all of these, cultural barriers, gender discrimination,  social hazard, physical challenges and lack of family support makes it worse for working women to continue their jobs (Mahmood, F., & Sonia, R., 2016). In comparison to the last decades, even though women empowerment rate has increased in Bangladesh, but it is still quite lower in comparison to the global progress (Islam, N., & Khatun, J., 2019). For a sustainable development of Bangladesh, increasing female employment rate is a must. 

Objectives of the study

This study aims to analyze the factors influencing women empowerment, identify barriers and propose strategies to enhance women's participation in Bangladesh, contributing to the advancement of gender equality and overall societal progress.  The study is intended to be carried out keeping in view the following objectives:

1. To investigate the influence of education and other socioeconomic factors on women empowerrment and household decision making.
2. To analyze the effect of domestic violence on women empowerment.
3. To assess the effects of healthcare facilities and its smoothness on women empowerment.
4. To investigate the effects of social and gender norms existing in Bangladesh affect women empowerment.
5. To analyze how the socioeconomic and demographic covariates affect women's participation in the workforce.

Women empowerment has been the subject of concern within the policy makers of developing countries, the inclusion of women into the process of uplifting the society and consequently, their partaking in economic activities alongside men, have gained significance in several national and international development plans (Paul et al., 2016). The economic development, empowerment, gender equality and upliftment of society is largely influenced by the participation of women in the workforce (Mehtap et al., 2016). Women's involvement in the workforce not only boosts the economic progression of a country but also other social sectors to expand (Islam, N., & Khatun, J., 2019). Education level, religion, age, place of residence, marital status, family financial status and media coverage are some of the key elements that affects women's the empowerment of women (Haque et al., 2011). As stated by the International Labour Organization, women's participation in the wokforce enhanced by 35 percent between 2008 and 2017 in Bangladesh (Nawaz, F., & Haque, T., 2020). The percentage is following an upward trend since. 

Educational is considered as the principal requirement for the empowerment of women in all aspects of the society. The most crucial factor for social change is higher education (Malik, S., & Courtney, K., 2011). According to a study, women's educational attainment is considered more essential than their workforce participation (Hossain et al., 2012). The fact that higher education plays a crucial impact on women empowerment was proven on the basis of results which highlighted that almost 88% of the respondents in the Pakistani society agreed to the fact that women empowerment is amplified by higher education (Razzaq, S., & Mehmood, S.T., 2022). 

Literacy rate for women has imposed a positive impact on the ideological, political and social spheres in accordance with a study conducted in the North-Eastern region of India. This region is influenced greatly by positive changes of education and training in achieving this result (Pegu, A., 2015). The opportunity cost of not participating grows along education level, and hence it is suggested that higher education refers to a higher probability in the employment market (Khandker, S.R., 1987). Education has a beneficial influence on female labor force participation of women (Contreas, D, & Plaza, G., 2010; England et al., 2012). Women with higher education degrees are more inclined to participate in labor force in comparison with the other groups (Kotikula et al., 2019). Women as head of the household, with higher education, residence in urban area have played a positive impact on female workforce participation. However, ownership of land, lower level of education, women's marital status, having infants have an adverse impact on the empowerment of women (Rahman, R.I., 2005) 

Even though women who received higher education have the highest level of empowerment among other classes, husband's education also plays a crucial role to understand the needs of wife and involving wife in decision making (Haque et al., 2011). A study that observed more than 70 developing countries concluded that Educated women are in a much better position to participate in financial and non-financial decision making in contrast to women who did not attain any formal or technical sector education, according to a study that observed the conditions of women from more than 70 developed countries (Le, K., & Nyugen, M., 2021). When women receives more and more education, an awareness of her rights increases in her which subsequently increases women empowerment (Urooj et al., 2022; Baig et al., 2020; Shetty, S., & Hans, V., 2015). As the level of educa-tion of women and their husband's increases, the status of women empowerment increases simultaneously (Urooj et al., 2022; Islam et al., 2023).

Marriage is considered as a known associate of women workforce participation in Bangladesh (Mahmud, S., & Bidisha, S.H., 2018). Marriage penalizes only urban women's participation in the workforce, while it is associated with a greater probability of workforce participation for men residing in the urban areas, and for both men and women in rural areas, according to a report of the Labor Force Survey (LFS) data in Voices to Choices. Some cases show that marriage imposes a negative effect on the workforce participation of women (Blau et al., 2014; Nor & Said, 2016). Moreover, single/divorced/widowed women leads by 19.3 percent in workforce participation than married women according to a study conducted in Malaysia using the Labour Force Survey data. Married men are more likely to join the workforce than married women because the husband is the main provider for the family (Nor, & Said, 2016). The workforce participation of married women and their age shows an inverted U-shaped pattern. The pattern specifies that the participation of women in the workforce increases initially but decreases later on. Additionally, women who were previously married have a higher probability of participating in the workforce and its percentage increased by 22.3 percent, thus announcing the probability of a push factor among the divorced, separated and widowed women to participate in the workforce in order to survive (Salleh, S.N.S., & Mansor, N., 2022). 

Another constraint to women's workforce participation in Bangladesh and other countries is the presence of children, particularly younger than 5 years old (Solotaroff et al., 2019). According to a study conducted using the Fifth Malaysian Population and Family survey (2014), nearly 65 percent of women stated care provision, particularly taking care of their children, is their major reason for not participating in the labour force. Besides, women's participation in the workforce is affected significantly by the number of children (Salleh, S.N.S., & Mansor, N., 2022). An argument emerged that if women are relieved from the weight of childcare and at the same time if they are given opportunity to obtain education from educational institutions, the probabilities of their workforce engagement would enhance significantly (Ejaz et al., 2010; Azam & Rafiq, 2014).

Perceptions and attitudes are heavily interrelated with female workforce participation. Women who consider the outside environment as safe and sound and who are comfortable travelling outside the homes by themselves are 9.6 percentage more likely to participate in the workforce. Women with conservative viewpoint are less likely towards participation in the workforce. Women who state themselves being depressed are 10.5 more likely to be working (Kotikula et al., 2019). Women's labor force participation is also restricted by social and cultural norms (Dildar, Y., 2015). Such norms have narrowed the choice of the labour force participation for women (Heintz et al., 2018; Nazier, H., & Ramadan, R., 2018). Disregarding the encumbrance of unpaid care activities amid working women and how social and cultural norms affect women's ability to enter and remain in the workforce subscribe to the ongoing gender gap in workforce participation (Ferrant et al., 2014). 

Though mostly in urban areas, creation of large-scale jobs in the manufacturing sector has been contributing to employment growth. This extension is another factor in the female employment growth that has brought a massive number of women into the workforce (Farole, T., & Cho, Y., 2017). Another constraint on the job accessibility to women is occupational sex-segregation. Gender norms often dictate the concept of “women's work” in Bangladesh that further affect the perceptions formed among people about what is acceptable for men and women (Kotikula et al., 2019). However, a promising trend was shown according to the Labor Force Survey (2016) that many manufacturing and higher skilled occupations displayed a high portion of female workers. Organizations can create their own practices, norms and cultures that are more convenient for female stuff (Hossain, J.B., & Kusakabe, K., 2005). In case of accessing job opportunities, accessibility to transportation is of key significance. Recent researches show that excessive levels of gender-based violence in public transportation and the bordering public spaces in many parts throughout the world is quite visible (Gonzalez et al., 2015). A research by BRAC in Dhaka area suggests that around 94 percent of women have faced any forms of sexual harassment in public transportations whereas 20.5 percent of women stopped using public transportations for these conditions (The Daily Star, 2018). The practice of Purdah and limitation of mobility is speculated to prevent women from searching for beneficial jobs outside their home (Kabeer, N., 2013; Ahmed, T., & Sen, B., 2018; Asadullah, M.N., & Wahhaj, Z., 2019). Moreover, a comparatively low proportion of working women move out of their their community to work (Kotikula et al., 2019).

Another major obstacle in the women workforce participation is the financial status of the families. A woman who belong in a low-earning family are often forced to participate in the workforce because of financial difficulties. Economically sound and wealthy women are less active in the workforce (Azam, A., & Rafiq, M., 2014; Shi, Y. et al., 2016). Moreover, women living in joint families participate in the workforce, surpassing the women living in nuclear families (Azam, A., & Rafiq, M., 2014). Urban women participate more in the workforce than rural women (Hussain et al., 2016; Nor et al., 2014). The workforce participation of women is often not her own choice. As Bangladesh has a strong patriarchal society, such decisions are often dictated or guided by the male members of the family (Kabeer, N., 2013). In reference to leadership competence, women have gained a legitimate space in political organizations that can uplift their marginalized position, even though they still remain a minority (Khan, M.M.R., & Ara, F., 2006). Media exposure is another factor that heavily influence the empowerment of women. Women with media accessibility have higher percentage of workforce participation than those with few or no accessibility to media (Chaudhry, I., & Nosheen, F., 2009; Khan, S.U., & Awan, R., 2011; Nayak, P., & Mahanta, B., 2012; Paul, S., 2016). Education, training and exposure to information media are playing the pivotal role in the empowerment of women in Bangladesh (Parveen et al., 2005).

A diversified scene can be found within the tribal women, where they participate to the family economy positively along with men in subsistence agricultural activities except ploughing. Additionally, managing household chores, child rearing is also the responsibility of the women. The work of tribal women has always been more than their counterparts as they have round annually, unlike men who enjoy seasonal and periodic leisure (Pappala, A.N., 2020).

Data Source and Study Area

For the analysis of women's empowerment, data from the nationally representative Bangladesh Demographic and Health Survey (BDHS), administered by National Institute of Population research and Training (NIPORT) of the Ministry of Health and Family Welfare have been utilized. We have used two waves of BDHS data, collected in 2017-18 and 2022 for identifying the changes in women empowerment status and the factors affecting it. These were nationally representative cross-sectional surveys based on a two-stage stratified sample of households. Only ever-married women aged between 15-49 years have been interviewed in these surveys. As the missing values have been eliminated, we were left with 18866 and 18987 observations, respectively.

Covariates and their measurement

The covariates used in the study are age groups, division, type of place of residence, highest education level, religion, relationship to household head, owns a mobile telephone, wealth index combined, currently residing with husband/partner, husband/partner's education level, education gap between husband and wife, media exposure and differences between number of sons and daughters (categorized).

Construction of Women Empowerment Index

Four dimensions of women empowerment have been used to calculate women's empowerment index as there exists no specific concept of women empowerment in the datasets. The survey, however, covered some important dimensions related to it. The dimension “Smoothness to medical facilities” was calculated using four indicators, namely: “getting permission to go for medical facilities”, “Getting money needed for treatment”, “Distance to health facility” and “Not wanting to go alone”. In order to construct the index without a hitch, we recoded the responses of these indicators (i.e. 0 for “big problem”, 1 for “not a big problem” and 2 for “no problem”). The second dimension “Domestic violence” contained five dimensions and was used to ascertain women's attitude towards wife beating. Women's opinion about the justification of their husband beating their wife if “she goes out without telling her husband”, “neglects her children”, “argues with her husband”, “refuses to have sex with him” or “burns the food” were recorded. As this index is negatively correlated with women empowerment, we recorded the responses inversely from the first indicator for analysis purposes (i.e. 1 for “no” and 0 for “yes”. The third dimension “Household decision making” contained four indicators that are defined as: people who usually decides on “respon-dent's health care”, “large household purchases”, “visits to family or relatives” and “what to do with money husband earns”. We recoded the responses as: 0 for “respondent has no role” and 1 for “respondent plays a role”. The final dimension in calculating the index was “Respondent currently working”, which took value 0 if the “respondent was not working” and 1 if the “respondent was working”. A summative index of women empowerment was calculated by taking the sum of the four dimensions. In order to understand the women empowerment index more clearly, we categorized the variable in five categories: completely disempowered for score 0, less disem-powered for score 1, moderately empowered for score 2, highly empowered for score 3 and completely empowered for score 4 or more.

Bivariate Analysis

For examining how the dimensions of women empowerment changes with the change in the covariates throughout the study years, we conducted the measure of association with the Pearson Chi-square test. The hypotheses that need to be tested can be stated as follows:

H_0: No association exists between selected covariate and the dimensions of women empowerment. 

vs

H_1: Association exists between selected covariate and the Low Birth Weight.

The following illustrates the definition of the Chi-square test statistic:

Where,

O_ij= The number of observations in the cell (????,) is denoted by i=1,2,⋯,r and 

j=1,2,⋯,c.

The expected cell value.

The test statistic follows a Chi-squared distribution with (r − 1)×(c − 1) degrees of freedom. Moreover, the p-value collected from this test is used to make decision.

Ordinal Logistic Regression Model

Ordinal logistic regression model is a statistical analysis method that can be used to model the relationship between an ordinal response variable and one or more explanatory variables, be it continuous or categorical. It can be considered as an extension of logistic regression where the independent variables are linearly related to the logit of a binary response. An important assumption of ordinal logistic regression is the assumption of proportional odds: the effect of an independent variable is constant for each increase in the level of the response. Hence the output of an ordinal logistic regression will contain an intercept for each level of the response except one, and a single slope for each explanatory variable. A cumulative logit parameterization is used in ordinal logistic regression models. However, there are more than one ways in which this can be done. Table 1 shows the parameter-zations used for the cumulative logit model, where J represents the number of levels in the categorical response variable, and p represents the number of explanatory variables. Models 1 and 2 are the most common parameterizations, where the outcome of interest is observing “Y less than or equal to j” where j is one of the ordered categories the response variable.

Table 1: Parameterization models involved in ordinal logistic regression.

There is a direct correspondence between the slope and the ranking in Model 1, as it associates a negative sign. Thus a positive coefficient indicates that the increase in the value of the explanatory variable is positively correlated with the likelihood of a higher ranking.

Ordinal probit model

 Suppose that i be the index respondent i, i = 1,…,n, where n is defined as the sample size. Let yi be the response of the ith individual to the survey question, and assume that this can take one of the integer values 1,2,3,…,j. Let  be the underlying latent variable that represents the respondent i's inclination of agreeing to the statement advanced. Let xi be a vector of characteristics relevant in explaining the attitude of a respondent. The ordered probit model is based on the assumption that yi* depends linearly on xi, according to

 yi*=xiʹβ+µi; where i=1,2,…,n

µi~N(0,1)                                                                                                       (1)

b is a vector of parameters that does not contain an intercept. These parameters will ultimately be interpretable in the same way as slope parameters in linear regression. y* is unobserved, but the relationship between y* and the observed variable y is

y=1 if -¥ < y* < k1

y=2 if k₁ < y* < k2

y=3 if k₂ < y* < k3

….

y=J if k_J-1 < y* < ¥                                                                                                 (2) 

The parameters k_j,  j=1,…,J-1, are known as "cut-points", or sometimes “threshold parameters”.

As all of the J-1 cut points are free parameters, the intercept of the model specified here is absent. If one of the cut points were normalized to zero, then the intercept parameter would become identified and would appear in the model. The need to normalize either the intercept or one of the cut points, and in addition to set Var(ui)=1 in (1), is in order to set the otherwise arbitrary scale of the latent variable y*. The log-likelihood function shall now be constructed. Suppose that Pi(y) be the probability that the response of the ith respondent is y. This probability is

Pi(y)= P(k_y-1 < y_i *<k_y)= F (k_y - x_i 'b) -F(k_y-1- x_i 'b)                          (3)

Where F(.) is the standard normal cumulative distribution function. So, based on a sample (yi, xi, i=1,…, n), the log-likelihood function is

=                (4)

The log-likelihood (4) is maximized with respect to the elements of b along with the cut-points k₁, k₂, …,k_J-1, by an iterative procedure, to give maximum likelihood estimates (MLEs) of both sets of parameters. Asymptotic theory involving the sample information matrix is used for estimating the asymptotic standard errors for both sets of parameters. When there are only two possible outcomes on the Likert scale, for example agree/disagree, ordered probit simplifies to the more familiar binary probit model, with the only difference that the single cut-point, although equal in magnitude to the intercept arising from binary probit, is opposite in sign.

Software and technical supports

The analysis procedures were performed using the statistical software packages: IBM SPSS Statistics version 20, MS Excel 2019, MS Word 2019 and R programming language, each for different purposes. The datasets (BDHS 2017-18 and BDHS 2022) were extracted in SPSS format. Data cleaning and statistical analysis were done on IBM SPSS Statistics version 20 and R programming language. We used MS excel 2019 in visualizing the data. Finally, we used MS Word 2019 for preparing our report.

Bivariate Analysis of Women Empowerment Index

The bivariate analysis has been performed to investigate the percentage distribution of the Women Empowerment Index (WEI), constructed with the association of the dimensions, with various covariates. Table 2 and 3 represents the bivariate analysis of the Women Empowerment Index (WEI) by different covariates of the BDHS 2017-18 and BDHS 2022 datasets, respectively. WEI by age groups displayed that women aged greater than 30 are more empowered than women aged less than 30. According to the study, the percentage of women who are assumed to completely empower decreased in the 2022 dataset. The aggregate of the higher groups has also faced a decline through the study years. However, the highly empowered group has seen a rise from the 2017-18 dataset. Khulna, Mymensingh, Rajshahi and Rangpur division had the highest percentage of empowered women in the 2017-18 dataset. The percentage dropped significantly in the 2022 dataset, where Khulna, Rajshahi and Rangpur division had the highest percentage of empowered women among the divisions, while Mymensingh division faced a decline in the percentage of empowered women. Sylhet division had the lowest percentage of empowered women throughout the study years. Even though the rural women had a greater advantage in the completely empowered category, they trailed by the urban women in the highly empowered category. Moreover, the rural women are ahead in the completely disempowered and less empowered group in both datasets.

Table 2: Women empowerment index by covariates from BDHS 2017-18.

Although women with less education had more empowerment rate in the 2017-18 datasets, their number in the dataset was lower. The scenario changed in the 2022 dataset, where the total percentage of the top two tiers in the women empowerment index was 84.7%, which is quite above than all the other groups and also a few percentages above the previous study year. This ascertains the importance of higher education in women empower-ment. Women following Buddhism and Christianity had more empowerment rate than Islam and Hinduism follower women in both study, though the number of respondents among them were many times lower than Islam and Hinduism followers. Respondent who are foster child and not related with the household head was seen to have the highest empowerment rate. However, only around 10 responses were recorded among the two groups. Women who are household head are more empowered than other categories, according to the total of the top two tiers of the women empowerment index. Although mother-in-law of the household head is seen to be more empowered, the number of them in the sample is quite insignificant in front of the other categories. Women who are mother of the household head also have a high empowerment rate throughout the study years, even the percentages increased throughout the study years. Even though women who do not own a mobile phone had more percentages in the top completely empowered tier, women owning a mobile phone placed higher in the total percentage of the top two tiers among the two datasets. According to the wealth index category, poorest people have a large percentage of women in the completely empowered tier; however they sit in a lower position than all the other categories. In the highly empowered tier, richest women are at the top among all the other categories. Moreover, the poorest women have the highest percentage in the bottom tiers. Women who live with their husband have higher empowerment rate than women living elsewhere in both the study periods. However, the percentage of empowered women who lives elsewhere has increased in the 2022 dataset.

Table 3: Women empowerment index by covariates from BDHS 2022.

The total percentage of the top two tiers of the Women Empowerment Index are almost equal throughout the education level of the husbands, though husband with less education were seen to have a greater advantage towards women empowerment in the 2017-18  study. This trend was present in the 2022 study, but the total percentage of the top two tiers was more than other categories among those respondents whose husbands received higher education, implying importance of higher education of the spouses. Percentage of women empowerment is higher among women who studied equal number of years as their husband than other categories. However, the percentage declined through the years. The 2017-18 study showed that women empowerment was higher among women not exposed to media, although the women whom were less empowered were also higher in this category. The 2022 study displayed the importance of being exposed to media for ensuring women empowerment when women empowerment rate was higher, both in the top two tiers in the women empowerment index among women who are exposed to media. This emphasizes the importance of media among women, which can also take effect by creating awareness among women. Women with more sons are more empowered than women with equal sons and daughters or more daughters throughout the study periods. However, the aggregate of the top two tiers of the women empowerment index is higher among women with more daughters in the 2022 dataset.  ANOVA F-test and Pearson's Chi-square tests have been performed throughout the analysis to assess the goodness of fit of the associations. As the covariates are tested to be significant, we carry the covariates into the ordinal logistic regression model.

Determinants of Women Empowerment: Ordinal Logistic Regression Approach

We fitted an ordinal logistic regression model to determine the adjusted effects of selected covariates on women's empowerment, as the empowerment index is an ordinal response variable. We found that the model has a p-value of less than 0.05, which suggests that there is a significant improvement in fit compared to the null model. Hence, the model is showing a good fit. The goodness-of-fit test was also conducted. A goodness-of-fit test, in general, refers to measuring how well do the observed data correspond to the fitted (assumed) model. A goodness-of-fit statistic indicates a poor fit if the significance value is less than 0.05. As the significance level of this model is greater than 0.05, the model adequately fits the data. The following table displays the Ordinal logistic estimates of women empowerment index with the covariates in the BDHS 2022 dataset. The fully significant categories are marked with three stars.

Table 4: Ordinal logistic estimates of women empowerment index with the covariates in the BDHS 2022 dataset.

Age acts as an important determinant for estimating the empowerment of women. Here, the age of women is categorized into 2 groups and the group of women aged greater than 30 has been taken as the reference category. The results show that as age increases, the coefficient value also increases, which indicates a positive correlation between the age of women and women empowerment. The divisions where the respondents reside also have a significant impact on women empowerment. Sylhet division has been taken as the reference category here. The results show the highest empowerment rate in Rangpur division, followed by Rajshahi, Khulna, Dhaka, Mymensingh and Chittagong. The lowest empowerment rate is seen in Barisal division compared to Sylhet division. Place of residence has a significant impact on women empowerment. Taking “women living in the rural areas” as a reference category, the women living in the urban areas shows negative correlation with women empowerment index. This indicates an opposite effect on women empowerment for women living in the rural and urban areas. Highest education level of women also plays a significant role on women empowerment. Taking higher education as reference category, the result shows negative coefficient among other categories. It indicates that higher education rate is more correlated with women empowerment than other categories. The determinant religion shows that women who follow Islam religion have the highest empowerment behind Christianity, where the reference category is Christianity. However, the number of Christian women is quite few in the survey. Although the table shows that Buddhism has a positive correlation with the reference category, their relationship is not significant.

Owning a mobile phone also has a significant part in women empowerment. Women owning a mobile phone have a negative correlation with women not having a mobile phone, which implies that ownership of a mobile phone has a significant effect on women empowerment index. A significant relationship between wealth index and women empowerment can be seen from the study. Taking richest as the reference category, women who belong to the poorer category of wealth has higher empowerment. There is a positive correlation between the reference category and other categories. Women currently living with husband/ partner have a positive impact on women empowerment than women living elsewhere, which is also taken as the reference category. Husband/partner who completed their primary education has a positive and significant impact on women empowerment index, taking the category “unaware” as the reference category. There is a negative correlation between women exposed and not exposed to the media, taking “exposed to media” as a reference category. It also implies that media exposure plays a significant role in women empowerment.

Women who are the household head, women who are either the mother or mother-in-law of the household head have a higher empowerment rate among the other categories of the covariate “Relationship to Household Head”. However, there remains doubt about its significance in the study. Similar questions rose in the cases of determinants such as: education gap between husband and wife and difference between sons and daughters.

Determinants of Women Empowerment: Ordinal Logistic Regression Approach

An ordinal probit regression model was fitted to determine the adjusted effects of selected covariates on women's empowerment, as the empowerment index is an ordinal response variable. We found that the model has a p-value of less than 0.05, which suggests that there is a significant improvement in fit compared to the null model. Hence, the model is showing a good fit. As the significance level of this model is greater than 0.05, the model adequately fits the data. The following table displays the Ordinal probit estimates of women empowerment index with the covariates in the BDHS 2022 dataset. The fully significant categories are marked with three stars.

Table 5: Ordinal probit estimates of women empowerment index with the covariates in the BDHS 2022 dataset.

Taking women aged above 30 years as a reference category in the age groups covariate, the coefficient of women aged less than 30 is -0.236, which indicates a negative impact on the women empowerment index. It suggests a lower likelihood of women aged less than 30 years being involved in women empowerment than women aged above 30. The value also suggests that an additional unit change in the value of the first category decreases the likelihood of being empowered by 0.236 units. The women from Rangpur division have a higher likelihood in the women empowerment process than other divisions, taking Sylhet division as the reference category. As all the coefficients of the divisions are positive, women from Sylhet division have a lower likelihood in this regard. The urban women have a negative coefficient value when the rural women category taken as the reference category. This also implies that the likelihood of these two categories is opposite to each other in regard of women empowerment. 

Women who received higher level of education have a greater likelihood in regard of women empowerment index, as all the other education categories have a negative coefficient value. Women who received no education are less likely to contribute towards women empowerment. Women following Christianity have a higher likelihood of being involved in the women empowerment process, though their number is quite low in the survey. Islam and Hinduism categories both have almost similar coefficient values and they are almost equally likely towards women empowerment progression. Though the Buddhism category has a higher coefficient value than these two categories, its significance level is not analogous to the model. 

Women who own a mobile phone have a higher likelihood of being involved in women empowerment process, as the women who do not have a mobile phone have a negative coefficient value. The earlier category is taken as index the reference category here. The dimension wealth index combined shows that women belonging to the poorer category have the highest likelihood of being empowered than other wealth statuses, when the richest status is taken as the reference category. The poorer category is followed by middle class category with coefficient 0.128, which differs slightly than the coefficient of the poorer category. Women currently living with their husband/ partner have a higher likelihood towards empowerment than women living elsewhere. Its coefficient 0.046 implies that any additional unit change in the first category will increase the women empowerment likelihood by 0.046 units. Women whose spouses received primary education have a higher likelihood of being empowered than other categories, when the category unaware is taken as the reference category. All categories have a positive coefficient value in this regard. Women not exposed to media have a lower likelihood of being involved in women empowerment when women exposed to media is taken as the reference category. Its coefficient value indicates that an additional unit change in the first category decreases the women empowerment index by 0.105 units.

Women as mother-in-law followed by mother and head of household have a higher likelihood of being empowered than other categories of the covariate “Relationship to household head”. Wife who studied more than their spouses has a higher likelihood towards empowerment. Women with more female children are more likely to be empowered. However, the significance value of these three covariates is not appropriate for this model.

For acknowledging the situation of women empower-ment, a Women Empowerment Index (WEI) was constructed that ranged from 0 to 4 and consisting of four dimensions: Women's participation in household decision making, opinion on domestic violence, smoothness to receiving medical facilities and the employment status of women. The study also compares the situations of the dimensions throughout the two study years. A bivariate analysis that was performed to investigate the association between women empowerment index and various covariates, suggested that women who were aged above 30 had a higher empowerment rate throughout the study years. Khulna and Rajshahi division was consistently among the top division in regard of women empowerment status, though the percentage dropped considerably throughout the study years. Both the respondents and their spouses who received higher education kept increasing throughout the study years. Women following Buddhism and Christianity had higher empowerment rate among the religions. Women who are household head are more empowered in terms of sample adequacy than other categories throughout the study years. Owning a mobile phone also had an association with the empowerment of women. Women living with her spouses/partners had a higher empowerment rate than women living elsewhere. Overall, the women empowerment rate has seen a decrease throughout the consecutive study years. Though, the number of less empowered women has decreased, the amount of empowered women has also seen a little bit of downfall in Bangladesh. So we recommend increasing the female literacy level at each level. The Government should take steps to ensure the maximum enrollment of girls in school and also increase the facilities of higher studies for women. Awareness should be created among the household to improve the scenarios in decision making criteria of women. Also voices against domestic violence should be preached among the masses.

M.A.U.: contributed in conceptualizing, designing and supervising the study. F.H.U.: contributed in data analysis and report writing. N.N.L.; and M.R.: contributed in data collection and analysis.

First and foremost, the authors are grateful to Almighty Allah. The authors are also thankful to anonymous reviewers and editors for their helpful comments and suggestions.

The author declares no conflict of interest.