WALKABILITY IS AN urban quality indicator related to how much the built environment encourages or inhibits walking. It has been linked to many individual and environmental benefits, including increased physical activity levels (SALLIS et al., 2015) and satisfaction with the environment (LEE et al., 2017). In this sense, walkable urban spaces can promote economic vitality and increase the quality of life for residents (GEHL, 2013; SALLIS et al., 2016).

Walkability is widely influenced by the compactness of the built environment, which enriches accessibility to destinations within short distances, minimizing possibilities of driving rather than walking (LO; HOUSTON, 2018). Such built environment features are opposite to the tendencies of urban sprawl in contemporary urban development (FRANK et al., 2019), offering many sustainability benefits such as the reduction of individual motor-vehicle transportation and subsequent reduction in the emission of hazardous air pollutants (RISSEL, 2009; GILES-CORTI et al., 2010).

An array of studies explores active travel behaviors, mainly by foot, establishing causal relationships between land use mix and travel by foot in diverse settings (CHRISTIAN et al., 2011; RACHELE et al., 2018; SUGIYAMA et al., 2019; LEÃO et al., 2020). It is understood that more non-residential destinations are present in an area with different land uses, consequently, pedestrian movement is more likely (CERIN et al., 2007). The relevance of land use mix is so latent that evidence suggests the possibility that the perceptions of users regarding land use mix may be relatively accurate, even if local residents make more vehicle trips for their daily activities (KOOHSARI et al., 2015). Its conceptual origin dates back to the “diversity” proposed by Jane Jacobs (JACOBS, 1961), who argued that the richness of experiences and stimuli in living within a city results from lively urban spaces.

Even though the concept of “diversity” and, therefore, land use mix, are very much connected to the urban quality understood as “Vitality”, it is in walkability research that land use mix as a construct is applied more profusely (FRUMKIN et al., 2004). Mainly in quantitative approaches, land use mix is conceptualized when measuring the heterogeneity of functionally different uses (LESLIE et al., 2007). The literature on land use mix usually considers the concepts of richness and evenness of use, derived from ecological studies (PEET, 1974; MAGURRAN, 1988).

Evenness indicates the proportion of land uses, representing the extent of variation in the distribution of uses (HAJNA et al., 2014), while richness indicates the total quantity of land uses, which point out the possibility of the existence of non-residential uses and seem to be the determining factor to diversity (KRETZER; SABOYA, 2020). Currently, land use mix is most widely quantified (FRANK; ANDRESEN; SCHMID, 2004; GEBEL; BAUMAN; OWEN, 2009; LEE, 2010; GRASSER; TITZE; STRONEGGER, 2016) through a variation of the equations of Shannon’s entropy, a measure of the evenness of uses, representing the extent of the variation in the distribution of land uses (HAJNA et al., 2014).

However, the possibility of existence of non-residential land uses seems to be the determining factor to land use diversity (KRETZER; SABOYA, 2020). Therefore, studies indicate simpler metrics as being as coherent in the objective measurement of land use mix, such as the proportion between residential uses and other types of destinations, obtained from simpler and more readily available information (SONG; MERLIN; RODRIGUEZ, 2013; SUNG; LEE; CHEON, 2015). By underlining such quantitative discussions, it can be understood that increasing equity in urban environments is essential for the promotion of social diversity with access to such walkable and compact infrastructure (FAGAN; TRUDEAU, 2014).

Urban theorists have long hypothesized that urban sprawl, marked by a low land use mix, reinforces a large array of gender2 disparities, including walking behavior (HANDY, 2004). Initially, walking differs between men and women mainly as the female participation in the job market increases (CRANE, 2007). Women have been known to make more trips for shopping for food or domestic chores and less leisure trips (POLLARD; WAGNILD, 2017). In the same sense, the understanding that women have more complex travel patterns and are more likely to engage in trip chaining, from non-work to work trips, is widely and historically recognized (McGUCKIN; MURAKAMI, 1999; SCHWANEN; KWAN; REN, 2008; BROWN; SMITH, 2016; LO; HOUSTON, 2018). Thus, it can be inferred that women’s walking is considerably influenced by land use mix, where living in more diverse environments will probably promote greater walking opportunities than living in areas without non-residential uses.

Despite the large volume of current research on walking and plentifulness of available evidence, the study of gender in walking remains limited. Especially in walkability research and its components, such as land use mix (OWEN et al., 2004), gender is rarely deeply investigated. The limited scholarly attention to women’s experience as pedestrians indicates a research gap in understanding women’s transportation patterns (GOLAN et al., 2019). The existing literature that focuses on analyzing such patterns in the context of urban characteristics generally points to substantial disparities between men and women, including their perception of attributes of the urban landscape (CLIFTON; LIVI; PARK, 2005), fear of crime (PAIN, 1997; KOSKELA; PAIN, 2000), and influence of land use patterns over their walking levels (CLIFTON; LIVI; PARK, 2005).

The research gap related to the influence of physical elements on women’s specific active behaviors is even more latent in countries ranked between low and middle income, such as Brazil (ADLAKHA; PARRA, 2020). In such contexts, active travel promotion is paramount due to the rapid expressive increase in population (BOUTAYEB; BOUTAYEB, 2005) and deep social disparities. In the same sense, researchers have long pointed out that existing evidence lacks sociodemographic moderations (DYCK et al., 2015), such as the effect of individuals’ income, since the labor division within families is determined, in part, by their income earning potential.

Considering all methodological and conceptual needs presented above and the latent research gap on gender differences within walkability analysis, the main objective of this study is to empirically verify the influence of land use mix on women’s walking behavior, moderated by sociodemographic characteristics in a Brazilian city. To do so, logistic regression modeling was conducted, considering women’s walking trips, income, employment status, age, educational attainment, and land use patterns of their zone of residence.

Only from this broader consideration of moderating characteristics on the relationship between walking behaviors and the built environment will it be possible to document the effects of land use mix on travel considering gender differences (BOARNET; HSU, 2015). In our structurally rigid social paradigm, urban land use policies may play an important role in forwarding more gender-sensitive land use policies and design guidelines. On a more urgent note, considering pandemics like COVID-19, land use policies promoting dense mixed-use neighborhoods can contribute to the reduction of both non-communicable diseases and infectious diseases, equitably creating activity-friendly communities (ADLAKHA; SALLIS, 2020).

Considering the phenomenon of the influence of land use characteristics in women walking as contemporary and contextual, hence, inseparable from reality, the ideal research strategy is the Case Study (YIN, 2001). The city of Londrina (Paraná, Brazil) was selected as a case, mainly due to the availability of current and precise data. In 2020, the Instituto de Pesquisa e Planejamento Urbano de Londrina (IPPUL) (INSTITUTO DE PESQUISA E PLANEJAMENTO URBANO DE LONDRINA, 2019), conducted an Origin and Destination (OD) research to prepare its Mobility Plan, in compliance with the Federal Law nº12587. Such raw data is available at the website of the city of Londrina (http://ippul.londrina.pr.gov.br/index.php/plano-de-mobilidade.html) and was systematized in a Geographic Information System environment (ESRI ArcMap 10.6) by the authors.

The OD survey was conducted with a sample of 5,131 households, totaling n=13779 individuals, both men and women (n is the number of individuals who compose the sample of the N – total population). This data is aggregated in the 91 city zones within the municipal limits, however, only 84 of these zones are contained in the urban perimeter and were considered (Figure 1). The seven zones excluded present rural characteristics that are not contemplated by this research’s theoretical and analytical bases.

The results of Multiple Logistic Regression analysis (Table 2) indicates that for every walking trip conducted by women of lower-income levels (two minimum salaries or less), the OR of their higher-income counterparts are much lower for the two consecutive higher income categories (OR=0.69 and OR=0.43, respectively). These results align with the literature that indicates that lower-income individuals of all genders present higher levels of active travel for their own locomotion (AUGUSTO et al., 2017). Such a scenario is widespread in low and middle-income countries where social disparities are abrupt. Further, evidence indicates that, in such contexts, motor vehicle ownership rates decrease for women in lower-income groups (ADLAKHA; PARRA, 2020), impacting their transportation options.

In parallel, the adjusted Odds Ratio for education indicates that women with higher educational levels have a much lower tendency to walk, opposing to poorly educated or illiterate women. The general literature on the moderation effects of educational attainment in walking reports mixed results, indicating stronger associations in less-educated adults (FORSYTH et al., 2009) and possibly more walking in individuals with higher educational levels (OWEN et al., 2007). However, it can be conjectured that in lower and middle-income countries such as Brazil, where educational attainment is historically dependent on social status, the evidence obtained here is relevant in enlightening such paradigm, specifically for women that are, moreover, associated with less car usage than men (VANCE; BUCHHEIM; BROCKFELD, 2004).

When it comes to age, adult women (OR=1.02) and children (OR=1.63) have shown to have a higher tendency to walk over older women. It can be conjectured that active behaviors decrease with age due to lack of available quality infrastructure, increasing the functional limitations of the elderly (RANTAKOKKO et al., 2010). This is in line with evidence from the literature that indicates a decrease in walking levels in higher ages (CAUWENBERG et al., 2012). Considering population aging tendencies, active travel may play an essential role in preventing chronic diseases and cognitive decline in older women (WEUVE et al., 2004) and a consequent decrease of the burden on public health (CAUWENBERG et al., 2011). Considering this scenario, walking should be encouraged through the provision of quality-built environment features.

It must be noticed that the difference between older women and adult women in walking, when interpreting an adjusted OR, is not too discrepant. This can be construed as a social phenomenon of the reality of a low-income country, where government provisions of solid social providence policies that may enable the elderly to retire promptly are compromised, maintaining their need to work and travel. In relation to the results on younger women walking, they seem to be contrary to evidence from the literature that indicates lower levels of activity among young girls (BUCK et al., 2014). It can be conjectured that, in low-income countries such as Brazil, women still play a determinant role in family dynamics, being attributed large responsibilities from younger ages.

The employment status explainable variable, binary by nature, presents results indicating that employed women have a much lesser chance (OR=0.35) of walking than unemployed women. This reinforces the research evidence that transportation behavior is largely determined by an individual’s financial status (BOARNET; HSU, 2015). It can be highlighted that household and child-oriented responsibilities are constant key factors, even with the increased participation of women in the job market (CRANE, 2007). The obtained results are parallel to those for the income and education variables, indicating that a relatively higher female job market participation, higher educational levels, and a consequently higher earning potential are linked to less walking behaviors, a reflection of social and economic structural difference between genders and within genders when it comes to transportation. Women are more prone to do more food and grocery shopping, child serving, and household errand trips comparable men (POLLARD; WAGNILD, 2017), and, if they are employed there are greater chances that these trips will be conducted by car.

Considering that the base level of housework for women is much higher, even if a woman makes more money than her spouse (BITTMAN et al., 2015), it can be inferred that her possibilities of transportation choices are limited due to time management. Further, women are understood to suffer significantly larger stress levels than men when walking in public places. Such factors may also contribute to the option not to walk when the income available allows for such. However, consequently, walking for leisure is also significantly below recommendations (ADLAKHA; PARRA, 2020). In conjunction with the fear promoted by social conditions and public space design, a spatial expression of the patriarchy, that imposes notions on women’s roles and appropriate places (PAIN, 2001), walking is a generally suppressed behavior among women, being avoided when income, education, and employment status allow for other opportunities.

Contextualizing such results in the city of Londrina (Brazil), it can be initially discussed that demographically, its population surpasses the 500,000 mark. It is composed mainly of women (263,628 women or 52% of the city residents). The fact that the female population makes up the majority needs to be emphasized, making the results found in this research even more tangible. Further, it must be considered that, from this percentage of women, 7.1% are over 60 years old, a fair part of the population that, according to the results obtained here, walks less than other age groups and needs policies and incentives to increase walking behaviors (INSTITUTO BRASILEIRO DE GEOGRAFIA E ESTATÍSTICA, 2010).

As for income, structural social disparities compose the city of Londrina, where the poorer 80% concentrate 42.77% of the municipality’s income, while the richer 20% concentrate 57.23%, according to the Instituto Brasileiro de Geografia e Estatística (2010). Considering these data, this research’s results and the evidences in literature, lower-income women walk more; however, they are more susceptible to health issues and lower leisure walking than higher-income women who walk less in general. Thus, despite both income categories being deeply separated by our social paradigm, in this case study, it is clear that all groups of women need more attention for increased walking behaviors, being through infrastructure or social incentives.

When it comes to built environment characteristics, land use mix was represented here as the proportion of non-residential uses on the individuals’ housing zone. OR indicated that women residing in areas with a high proportion of non-residential uses, therefore with greater availability of destinations, have a 2.4 greater chance of walking than women residing in areas with a low proportion of non-residential uses. This is an indication of the possible strong relationship between women walking and the land use composition around their household. On the other hand, for the proportion of residential uses on present on the individuals’ housing zone, OR indicate that, for each walking trip conducted by a woman living in an environment with low proportions of residential uses environment, 0.77 are conducted by women residing in places with a high proportion of residential uses. These results also indicate that the prevalence of residential uses, and the consequent lack of diversity between different use categories, might be detrimental to women’s walking behavior.

The question as to whether land use patterns influence women’s walking behavior in Brazil is enlightened by the obtained results, indicating that more compact communities seem to provide opportunities for walking. However, in terms of gender equality, women have historically been understood to engage in fixed activities more frequently than men (PICKUP, 1988), hindering transportation freedom to consider more sustainability-friendly modes such as walking (SCHWANEN; KWAN; REN, 2008). Such concept is expressed in the results obtained, where more educated women with jobs and higher income seem to generally walk less. In such a structurally rigid social paradigm, urban land use occupation policies may positively contribute to design guidelines encouraging denser occupation through more gender-sensitive land use policies.

Solid literature evidences indicate that activity rates are higher in areas with greater land use mix (FRANK; ANDRESEN; SCHMID, 2004; EWING et al., 2014). However, in a paradigm that exposes patriarchal power mechanisms to women in the attribution of greater responsibility for domestic labor despite their participation in the labor market (SCHWANEN; KWAN; REN, 2008), it seems to be relevant that urban design policies consider providing opportunities for greater within-household gender equality through the provision of destination options within walking reach, so that populations such as those of working women might have more destinations worth walking to, so that these women may get more exercise (LO; HOUSTON, 2018)

The literature indicates that, in high-income countries, walkability and land use mix establish a clear link to real estate prices, connected to more developed areas (PIVO; FISHER, 2006). In Brazil, there is no record of specific research on this topic. It can be conjectured that high-income women possibly live in areas with less land use mix, a reflection of traditional Brazilian occupation patterns, where large areas of single-use vertical residential buildings or large horizontal private condominiums are directed to higher-income classes. However, further evidence is needed to establish clear causal trends.

A binary multilevel logistic regression analysis in a representative case study with an up-to-date database reveals evidence of a trend that the walking behavior of women is sensitive to land use occupation and sociodemographic differences, especially related to income, as suggested in previous studies. The evidence obtained in this study indicates that, in line with the literature, land use mix affects women’s active travel behavior. Further, higher income and education – directly related to employment – seem to be expressive characteristics in negatively determining their walking behavior. These are indications that working women that are more spatially and temporally constrained due to household and childcare activities have a reduction in freedom to explore urban opportunities and transportation options. Urban design heavily contributes to such burdens. However, it possesses the power to ameliorate them by providing safe and accessible alternatives, as is active travel.

The analysis conducted here proposes several significant contributions. First, this study adds to the debate of the growing Brazilian scientific literature on the objective quantification of walkability, here specifically of land use characteristics – a proxy for walkable environments – moderated by gender discussions and socioeconomic characteristics. Although relevant to support evidence-based public policies and scientific-academic research, it can be pointed out that, in a context marked by a socio-political framework of patriarchal relations, from social vulnerability to structural income disparities, political actions for environmental improvements aimed at walking women should always consider such intangible barriers. Furthermore, the obtained evidence must be taken with critical consideration of our social urban scenario, marked by structural crime, insecurity, and a persistent lack of infrastructure and maintenance. Thus, our results must be considered in the context of the general issues of public spaces for pedestrians in Brazilian cities.

Other limitations include the variety of land use measures applied. Such research field is under intense methodological changes and still relies on inconclusive findings regarding the ideal measure for land use mix quantification. Understanding representations and varying compositions of land use classifications and their impact on walking behaviors, including gender differences, remains uncertain. Future studies can further contribute to such research efforts.

For the present study, such data was not available. However, it is paramount that future studies control the sociodemographics of race when studying gendered walkability3. The existence of spatial differences structured within gender, especially concerning race, income, and age, should not be overlooked. The relationship between women and public spaces, including their fear, is shaped by their identity. Therefore, considering it to be uniform is not ideal. Evidence is also needed on gender differences in general walkability, comparatively.