Introduction

Like many other post-communist countries, Slovakia experienced a fertility transition in the early 1990s, achieving stable fertility below the replacement level. However, while differences in regional fertility levels increased at the turn of the century, the total fertility rate (TFR) decreased significantly. The TFR dropped from 2.1 in the early 1990s to 1.2 children per woman in the early 2000s. As a common phenomenon in post-communist countries, this trend has been the focus of much research. This study, in particular, aims to contribute to understanding this trend in Slovakia, where fertility is slowly increasing but remains below the replacement level, currently standing at 1.6 children per woman.

The political changes in Eastern and Central Europe at the end of the 1990s profoundly impacted various societal levels. The socialist system, seemingly stable and stagnant, was replaced by unprecedented social and economic changes (Sobotka, 2011). The establishment of democracy and the transformation of the Slovak economy led to the privatisation of state property and the restoration of private property and business. The emergence of unemployment and the fundamental changes in housing and family policy created a sense of social insecurity. Ultimately, these societal changes heralded ideological and cultural change, significantly altering population reproduction.

The previous socialist conditions had ensured a high degree of reproduction uniformity. It was especially noted in early fertility, a strong inclination towards the two-child family model, and low childlessness. Šprocha and Tišliar (2016) noted that in the post-communist period, due to societal changes, divergent tendencies are manifested in this area, and the reproductive paths of women differ. In addition, Potančoková (2011) reported that the significant factors affecting birth rates in the transformation period were decreased fertility intensity, childbirth postponement, and more children borne out of marriage. The power of these tendencies varied across Slovak regions due to differences in demographic, economic, social, and cultural factors. However, there is general agreement that these factors indirectly but significantly affect current fertility characteristics and inter-regional differences (Abdennadher et al., 2022; Campisi et al., 2020; Iwasaki & Kumo, 2020; Lieming et al., 2022).

The study evaluates the effects of socioeconomic determinants of fertility, explaining differences between regions in Slovakia. The regional fertility in Slovakia varies significantly. Herein, we identify significant fertility predictors based on regional strengths and characteristics. The study's originality lies in finding significant predictors behind the fertility rates between the regions of Slovakia while connecting the spatial context with the determinants explaining the differences.

Theoretical framework

Demographic development is a multidimensional process influenced by a combination of factors, and their interactions may vary in different regions and periods. Scientists from various fields have investigated the complex interplay of factors that shape demographic trends and changes in population composition (Bongaarts, 2009; Bontje, 2020; Pampel, 2011; Raymo, 2015; Sobotka & Fürnkranz-Prskawetz, 2020; Tulchinsky & Varavikova, 2014). Birth rates and fertility intensity are critical demographic components that determine population increase or decrease, and determinants with varying effects influence the levels. In addition, the set and structure of fertility determinants differ in countries and regional areas depending on the specific cultural, social, political, and economic situation. Moreover, the determinants interact and vary in relative importance in different contexts, so understanding specific population dynamics is crucial in establishing fertility and birth patterns.

Determinants affecting natality and fertility can be perceived in several broader categories (El-Ghannam, 2005; Marenčaková, 2006; Adhikari, 2010; Wang & Sun, 2016), while in the conditions of the Slovak regional structure, the level of fertility varies most often depending on social, economic or cultural factors (Potančoková et al., 2008; Šprocha & Bačík, 2021a, 2021b; Šprocha & Bleha, 2018; Šprocha & Tišliar, 2019; Šprocha et al., 2022).

Sociocultural conditions

Sociocultural conditions may contribute to regional variation in fertility levels by influencing individuals living in certain areas. A major societal shift in Slovakia in the past three decades is increased women's education, and its association with fertility has been extensively researched. Results indicate that increased education is consistently related to lower TFR values (Adsera, 2017; Cheng et al., 2022; Gray & Evans, 2019; Maulida et al., 2023; Shirahase, 2000). Šprocha and Tišliar (2019) add that education is one of the critical factors in the intensity and timing of maternity and family starts and the realisation of fertility outcomes in Slovakia.

D'Addio and D'Ercole (2005) explained observed changes in OECD countries’ fertility rates by socio-economic influences, stating that as a result of increasing women's education, women's participation in the labor market also increases, the average age of first-time mothers increases, and women's higher education contributed to higher contraceptive use. More educated women are also more likely to delay childbearing until an older age because they participate in the labor market, are more economically independent, are more likely to have a career orientation and build a job position. Consequently, they often have fewer children than initially planned (Ní Brolcháin & Beaujouan, 2012; Iacovou & Tavares, 2011; Testa, 2014). The desire for children increases at higher income levels, increasing the chances of providing for a bigger family, and higher income, as well as better employment prospects, are the result of accumulated human capital in the form of higher educational attainment.

Becker (1992) assumes that people with higher education might want to have a larger number of children, thus pointing to the positive effect of income—but this is in contrast to the negative effect of the costs of lost opportunities (Nisén, 2016). However, age is a crucial determinant of fertility because conception ability decreases with age. Fekiačová (2019) adds that postponing parenthood until an older age can negatively affect the intensity of fertility in highly educated women in the form of a smaller number of children, or it can lead to final childlessness, which may indicate lower TFR values in the region. The opposite trend can be observed among less educated women when they enter motherhood soon after completing their education, thereby trying to reduce the uncertainty associated with the official labor market (Hechter & Kanazawa, 1997). In addition, Potančoková et al. (2008) cite ethnicity and cultural determinants of fertility, where Slovak Roma women with low education secure a more certain source of finance, which often indicates higher TFR values.

Further research provides evidence of an unequal relationship between education and fertility. The negative gradient dominates primarily in Central and Eastern European and German-speaking countries (Beaujouan et al., 2016; Klesment et al., 2014; Nisén et al., 2021; Wood et al., 2014). Deviations from this pattern have been reported in northern and northwestern European states (except Finland), where the gradients have narrowed and are often no longer observable (Jalovaara et al., 2019; Testa, 2014).

The theory of gender equality (Anderson & Kohler, 2015; McDonald, 2013; Mills, 2010; Neyer et al., 2013) also points out the negative impact of women's higher education on their fertility. However, the effect is not uniform and varies by gender and parity. The demographic revolution theory further elucidates the relationship between education and fertility (Axinn & Barber, 2001; Lesthaeghe, 2014; Sobotka et al., 2008). As women's educational attainment increases, fertility rates tend to decline, often associated with improved gender equality and socio-economic development in many developed countries.

Religiosity, explaining deviations in reproductive outcomes, has long been perceived as one of the significant and consistent factors determining actual fertility, especially in Slovakia (Šprocha & Tišliar, 2019). Numerous studies confirm its positive influence on fertility (Philipov & Berghammer, 2007; Frejka & Westoff, 2008; Zhang, 2008; Dilgmaghani, 2019; Herzer, 2019; Götmark & Andersson, 2020; Bein et al., 2021a; Buber-Ennser & Berghammer, 2021). All religiosity rates are generally related to a higher ideal number of children, a higher probability of having another child, and a higher expected and actual number of children, primarily with the Catholic faith in the European environment. Among various dimensions of religiosity, religious involvement is a more significant predictor of fertility than religious tradition, affiliation, and self-rated religiousness (Philipov & Berghammer, 2007; Dilmaghani, 2019; Bein et al., 2021b; Perry & Schleiffer, 2019; Buber-Ennser & Berghammer, 2021), which promises higher TFR values.

The scientific community has focused on clarifying the relationship between religion and fertility, and on explaining the differences in its influence on fertility in different countries. For example, religion is important in the lives of half of American women but in less than one-sixth of European women. In addition, Frejka and Westoff (2008) report that women in Northern and Western Europe are less religious. However, they have the same or even higher fertility than American women, and significantly higher fertility than Southern European women. The authors consider that a slight increase in European fertility could theoretically be expected if Europeans were as religious as Americans. Buber-Ennser and Berghammer (2021) add that the positive influence of religion on actual fertility outcomes is much more substantial in Western Europe than in Central and Eastern Europe. Moreover, the effects of religion in the latter countries are considered generally weak and inconsistent, and religious significance and relationship to fertility there would benefit from further investigation.

Many studies have examined variations in fertility levels in different ethnicities (Adebowale, 2019; Bagavos et al., 2008; Booth, 2010; Chui & Trovata, 1989; Dubuc & Haskey, 2010; Forste & Tienda, 1996; Jasilioniene et al., 2014; Martin, 2019; Muhammad, 1996; Urale et al., 2019), which explain in connection with cultural, social, economic and minority postulates. Some ethnic and cultural groups maintain relatively high birth rates despite the global shift towards smaller families, and traditional gender roles, religious beliefs, cultural norms, and socio-economic conditions all influence the outcome. Moreover, established patterns are not absolute determinants because individual choices and circumstances vary considerably within a given ethnic group (Forste & Tienda, 1996; Majo, 2014; Šprocha, 2014).

Another important differentiating factor in fertility outcomes is marital status, which affects basic demographic processes such as birth rate and mortality, as well as marriage and abortion rates. Concerning fertility, marriage has long been considered its prerequisite (Ridfuss & Parnell, 1989), and their high degree of correlation in the past has been empirically documented (Magdalenić, 2016). However, with the onset of the demographic transition, this relationship began to loosen, and new forms of family behavior appeared in society. Several studies have led to the clarification of the relationship between marital status and fertility and the description of differences in the level of fertility between subpopulations with different marital status (Van Bavel et al., 2012; Hiekel & Castro-Martín, 2014; Magdalenić, 2016; Meggiolaro & Ongaro, 2010; Nedomová, 2015; Perelli-Harris et al., 2009; 2014; Perelli-Harris, 2014; Raley, 2001; Ridfuss & Parnell, 1989). Behind the increase in non-marital births in the European area is an increasing number of conceptions and births within cohabitation, pointed out by Perelli-Harris et al. (2012), while in most of the observed countries, the percentage of births to unmarried women decreased or remained relatively similar. According to Magdalenić (2016), the reproductive function of marriage thus loses its relevance. In addition to the type of partnership itself, fertility is also affected by its instability. In line with Meggiolaro and Ongaro (2010), divorce can theoretically be considered a depressive factor, but empirical studies have not consistently confirmed this hypothesis. Remarriage and cohabitation increase the probability of giving birth after the dissolution of marriage, whereby separated childless women have a higher risk of conception. However, according to Van Bavel et al. (2012), divorced women have lower fertility than continuously married women, even when re-partnering. In the case of men, it is the opposite.

Slovakia also captures something similar, where with the onset of the second demographic transition (cf. Mládek, 1998) already in the early 1990s, new forms of family behavior begin to appear with a gradual decline in cohabitation in marriage and an increase in partnerships cohabitation. The period of the last three decades due to population aging (Kačerová, 2009; Šprocha & Ďurček, 2019) also brings changes in the age composition of the Slovak population due to population aging (Kačerová, 2009; Šprocha & Ďurček, 2019) with the expansion of thirty- and forty-year-old age groups, which enters into marriage more often. These changes are behind the significant increase in the share of the reproductive Slovak population living in marriage and only a slight increase in the divorced. Up to 85% of the population lived in marriages in Slovakia in the last census of 2021, and 17% of the reproductive population were divorced. Among the EU countries, Slovakia still has a higher proportion of married people. The shift in society from the traditional model of the family, which typically involved a married couple and their children living together, to modern and postmodern forms of family relations with an inclination towards individualism, personal freedom, and independence (Kraus et al., 2020; Mendelová, 2018) causes a general decline in marital cohabitation with the prospect of further decline, then would be the relevance of using the women's divorce rate in the research low.

Nevertheless, within the demographic of reproductive-aged women, married Slovak women continue to outnumber divorced women, with a notable correlation observed between Slovak regions and TFR. Šprocha and Tišliar (2021) also indicate that middle-aged and older women enter the marriage union, where the union remains more stable if women with a higher education who postpone motherhood enter it. However, many factors influence the formation of the regional structure of the reproductive population according to family status. Demographic factors include primarily the intensity and timing of marriage and divorce. Therefore, in the north and east of Slovakia, a region with a more significant representation of married women was formed (cf. Bleha et al., 2014). In the western region, it is precisely the hinterland of the capital city that, due to the migration attraction of the productive population, registered a similarly higher proportion of married women in the last period (cf. Pregi & Novotný, 2019). The southern region of Slovakia is represented by a higher representation of cohabitation and a higher representation of divorced women in the reproductive population, which some authors (Džupinová et al., 2008; Korec, 2005; Šprocha & Ďurček, 2017) explain by a higher representation of low-income persons, higher unemployment and a lower level of development of local economies.

Economic conditions

The income-fertility hypothesis describing the relationship between income and fertility suggests that fertility falls as income rises. D'Addio and D'Ercole (2005) consider women's income and earnings as critical influences for childbearing, and they record the complexity of the relationship between income and fertility. Findings have shown that fertility rates can reflect the difference between current and past income levels of each cohort, and wealthier OECD countries have higher fertility rates and higher average age at first birth (Herzer et al., 2012), while in all countries, women with higher household income levels have fewer children compared to other women. Although there is a general correlation between income and fertility, this relationship can vary across countries and cultures (Fox et al., 2019; Luci-Greulich & Thévenon, 2014).

Individual preferences and choices regarding family size can differ significantly, even within income groups. Shifts in family policies (Adema & Thévenon, 2014), changes in the spatial organization of the economic sphere (Ciminelli et al., 2021; Wachsmuth, 2022), and selective processes of international and internal migration (Rees et al., 2017; Billari & Dalla Zuana, 2013; Pregi & Novotný, 2019) contribute to this. In addition, Fox et al. (2019) suggest a stronger convex relationship between fertility and income in Western European countries, while in Eastern European countries, where only Poland and the Slovak Republic seem to have transitioned to a positive fertility–income relationship.. The income level at which the association between income and fertility changes from negative to positive was much lower in the East than in the West but varied in both regions.

Some research has documented an inverse relationship between fertility rates and women's participation in the labor market (see Bernhardt, 1993; Mishra & Smyth, 2010; Altuzarra et al., 2019; Del Rey et al., 2021). However, some studies point to a change in the correlation between total fertility and the women's employment rate or the labor force participation rate (Ahn & Mira, 2002; D'Addio & D'Ercole, 2005; Del Boca et al., 2003) or they track their joint growth (Hwang et al., 2018) in some OECD countries. In fact, according to Klasen et al. (2021), heterogeneity in returns to women’s own characteristics and family circumstances—including education, income, and fertility explains most of the between-country differences in participation rates, indicating that the economic, social, and institutional constraints that shape women’s labor force participation are still largely country-specific.

Hwang et al. (2018) see the change in correlation and the increase in both indicators as an increase in childcare substitutability, i.e. the degree of perception of market-provided child care by parents as a sufficient substitute for the mother’s time. As the value of substitutability increases, the female labor force participation rate increases while a convex relationship (U-shaped) emerges in interaction with TFR, the existence of which can be explained by a combination of behavioral and compositional effects. According to Thévenon and Luci (2012), fertility trends also depend decisively on mothers' ability to combine work and family life, pointing to a higher fertility rate in countries where women have greater access to the labor market (Northern European countries and France). Fertility outcomes can be influenced by the availability of formal childcare facilities, which (Ridfuss et al., 2010; Haan & Wrohlich, 2011) is considered an essential prenatal tool of family policy mitigating the conflict between work and family. Jung et al. (2019), as well as Wood and Neels (2019), state that despite the generally widespread hypothesis about the positive impact of available childcare on the level of fertility in the developed world, the results of several empirical studies are ambiguous and contrary to hypothetical expectations. A positive relationship between these indicators was recorded most often only in regions with high participation of women in the labor market. However, the research results of Ridfuss et al. (2010) showed that increasing the availability, i.e., capacity, of facilities from 0 to 60% for preschool children in Norway led to an increase in fertility by 0.5–0.7 children per woman under 35 years of age.

In addition to women's participation in the labor market, unemployment also affects the timing and intensity of fertility. Still, according to some authors, the effects are unclear (Andersen & Özcan, 2021; Bono et al., 2015; Huttunen & Kellokumpu, 2016). D'Addio and D'Ercole (2005) point out that in the 1990s, there was a change in the correlation between fertility and unemployment, with the fertility rate of most OECD countries being higher in times of low unemployment and falling with its increase. Adsera (2011) states that high and persistent unemployment is associated with delayed childbearing and, as a result, likely fewer children. However, Fernandez-Crehuet et al. (2020) report that fertility and unemployment rates are unrelated in the long run. Andersen and Özcan (2021) also concluded that unemployment positively affects the transition to motherhood and has no significant effect on second-order children, possibly due to the specificity of the Danish context. Finally, studies record that unemployment does not necessarily negatively affect fertility, and its negative impact is selective.

Housing availability further determines fertility, but its influence in Slovakia has rarely been analyzed. However, Katuša (2012) demonstrated that many young Slovak couples could not afford appropriate housing for child-rearing, and this was a prominent cause of the decline in birth rates and fertility, regardless of employment, education status, and other factors. European researchers (Campisi et al., 2020; Makszin & Bohle, 2020; Mulder & Billari, 2006; Stoenchev & Hrischeva, 2023; Zavisca & Gerber, 2016) have led efforts to understand how housing conditions, especially ownership, are related to the transition to adulthood, the formation and dissolution of partnerships, and fertility.

In addition, Zavisca and Gerber (2016) identified housing as a primary source of socioeconomic difference, with the home critical for everyday life and family dynamics, consumer lifestyle, subjective well-being, and family planning. High housing prices also mean that individuals and couples may face financial constraints that make it difficult to afford the costs associated with raising children. According to studies by Zhang et al. (2012) and Saguin (2021), an increase in housing prices leads to an indistinctive decrease in the birth rate, but changes in housing prices do not have an immediate effect on fertility; they can even also have a positive impact (Clark & Ferrer, 2019; Simo-Kengne & Bonga-Bonga, 2020). However, housing availability problems affect the timing of births more significantly than the intensity of fertility (Kostelecký & Vobecká, 2009), and the relationship between housing affordability and fertility, therefore, varies depending on the region or country’s social, economic, and cultural background.

Urban–rural conditions

The degree of urbanization similarly plays a role in the fertility rates. Urban areas tend to have a higher prevalence of apartments with smaller living spaces, which is natural due to population density and limited land availability. Research by Kulu and Vikat (2007) revealed that people living in apartments have lower fertility than those living in single-family homes. The smaller living spaces of housing may force families to limit childbearing due to space constraints. In contrast, the family environment often associated with family houses may facilitate reproduction in rural areas (Felson & Solaún, 1975).

High housing costs can lead individuals to migrate from urban centers to rural or suburban environments, where they can generally have more children. These selective steps for increasing fertility outcomes may also contribute to patterns of low urban fertility (Kulu, 2013; Rusterholz, 2015). However, studies also show that selective moves from urban centers to family-friendly environments do not cause significant differences in urban–rural fertility levels (Kulu & Washbrook, 2014).

Marenčáková (2001) considers the size of the settlement to be an important differentiating factor for several demographic phenomena and processes in Slovakia and adds that the difference in the reproductive behavior of the population of cities and rural communities is generally accepted. Scientific studies (Kulu & Boyle, 2009; Vobecká & Piguet, 2012; Kulu, 2013; Kulu & Washbrook, 2014; Riederer & Buber-Ennser, 2019; Lopéz-Gay & Salvati, 2021; Rodrigo-Comino et al., 2021; Salvati, 2021) describe and explain the differences in fertility levels based on the urban–rural dichotomy or along the city-suburb-rural gradient. Most confirm that rural areas and small towns have higher fertility than large cities. Urban–rural fertility disparities decrease over time (Bleha et al., 2020; Kulu, 2013) when factors describing the economic environment, family and gender norms, and population composition are considered (Riederer& Beaujouan, 2024).

While urbanization generally correlates with lower birth rates, there can be considerable variation between countries and regions. According to Billari and Kohler et al. (2002), European regional fertility regimes have rapidly changed due to common socio-economic transformations. However, these changes vary across the geographical gradient north–south and west–east of European cities (Rodrigo-Comino et al., 2021). These findings largely mirror the dominant phase of the urban life cycle (Morelli et al., 2014). Northwestern cities have entered a phase of reurbanization (Dembski et al., 2021), leading to new forms of urban expansion such as polycentric development. This trend has seen inner cores gradually attracting a younger population and couples with a high propensity to marry and have children, indirectly contributing to a higher average birth rate.

Conversely, Eastern European cities are still undergoing the final phase of suburbanization (Stryjakiewicz, 2022; Szmytkie, 2021), characterized by intense population growth in suburban districts and significant stability or even shrinkage of inner urban cores. Rodrigo-Comino et al. (2021) state that suburbanization is only associated with younger and larger families—and thus higher fertility levels—in Eastern and Southern Europe. Compositional and contextual factors and the influence of selective migration and housing conditions can explain differences in fertility between urban and rural areas (Kulu, 2013; Kulu & Washbrook, 2014; Nestorová Dická et al., 2019; Riederer & Buber-Ennser, 2019).

Other conditions

The theoretical overview of the factors that influence the level of regional fertility indicates the authors' intention to include them in their research. However, in addition to the above, regional fertility can be conditioned by other indicators not included in our study. Studies, such as Hesketh and Xing (2006), Wu et al. (2006), Tafuro and Guilmoto (2020), Chao et al. (2021), and Becquet et al. (2022), have pointed out that the gender ratio of the reproductive population can also play an important role in the given issue. Slovakia's secondary sex ratio shows a slight predominance of the male gender, similar to the case in other developed regions (Orzack et al., 2015), which can be used as a reflection of living conditions and health status (Chao et al., 2022) while deteriorating conditions are associated with a decrease in ratio values and vice versa. Golian and Liczbińska (2022) also confirmed what was stated for Slovakian conditions, while in the period of formation of the current reproductive population, there are no significant exogenous shocks that could significantly affect the sex ratio at birth and thereby affect the demographic status of the current reproductive populations in the regions of Slovakia. In almost every region of Slovakia, men slightly predominate in the reproductive population, or the ratio is mainly balanced. In the end, even the initial survey of the relationship between the fertile population of women and the level of fertility between Slovak regions did not confirm significance.

However, the migration factor can also affect the total fertility rate. In regions with significant migration, the birth rate can be affected by changes in the population's demographic composition (e.g., Bagavos, 2019; Sabater & Graham, 2019). In Europe, some authors register that economic and migration factors often explain changes in fertility (cf. Majelantle & Navaneetham, 2013; Sobotka et al., 2011), and migration may be responsible for reshaping the ethnic and social composition of many highly developed countries (Sobotka, 2008). According to research by Parr (2021) on the fertility replacement level in the presence of positive net immigration, Slovakia's current migration replacement TFR is low. This is related to its low birth rate and low net migration, and the relatively lower life expectancy at birth also plays a role compared to Western, more developed countries. This means that migration does not significantly impact the overall demographic situation in the Slovak regions and, therefore, not on the level of fertility. Compared to Western, more developed countries, where migration can be a more significant factor, its influence is minimal in Slovakia. As for life expectancy, although it is relatively lower compared to Western countries and may affect the demographic structure of the population, it does not directly affect the fertility rate. Fertility levels are more influenced by other social, economic, and cultural factors, which are the primary focus of our research.

In the same way, access to health care plays an important role in the issue of this research. Regions with better access to health facilities and information on reproductive health tend to have lower fertility due to increased use of contraception and better outcomes in the area of maternal health, as confirmed by studies such as Yüceşahin and Özgür (2008), Phillips et al. (2019), Herrera-Almanza and Rosales-Rueda (2020). From another point of view, according to Brodeur et al. (2022) and Jones et al. (2023), access to reproductive health care services, including family planning resources and prenatal care, can affect fertility rates. Stefko et al. (2018) demonstrated the diversity of healthcare facilities' performance in Slovakia's regions. However, over time, there is an indirect dependence between the variables and the results of the estimated efficiency in all regions, and thus, all regions of Slovakia have increased their productivity compared to ten years before the COVID-19 pandemic. Technological improvements significantly impacted this improvement (Vaňková & Vrabková, 2022). Even though Soltes (2016) states that the situation in individual regions is uneven and there are regional differences in access to care, all Slovak districts have medical facilities of varying quality. It is important to note that access to health care can affect a population's overall health, but the direct effect on fertility levels is unclear. Our research focused on other factors that directly and significantly impact fertility levels, such as economic, social, and cultural determinants. Although access to health care is important for the population's overall health status, based on available data and research, it was not considered a key factor that would directly influence the fertility rate in Slovak regions.

In addition to the determinants mentioned above, the researchers also investigated the relationship between abortion, average life expectancy (Trynov et al., 2020), the size of the university population, the share of employees in agriculture (Campisi et al., 2020), or also investigating the quality of life and the level of fertility (Palomba et al., 2018; Koert, 2021). These factors play a significant or less significant role in influencing overall fertility in the regions.

Abortion rates can have an impact on the level of fertility in Slovak regions, but this relationship is often complex and influenced by many other determinants that influence reproductive behavior. In our research, we focused on identifying factors that directly and significantly affect the level of fertility. In the initial phase of the investigation, no significant relationship between abortion rate and fertility level was demonstrated. This confirmed that abortion is not the primary determinant of fertility in Slovak regions in the context of our research.

Educating women plays a significant role in the final level of fertility. Research indicates that women's low education is directly linked to a higher fertility rate (cf. Šprocha et al., 2020). Although higher education also affects fertility, its effect may be more indirect and complex. Higher education is often associated with postponing parenthood, lower fertility rates, and greater emphasis on career and personal development (Šprocha & Bačík, 2021a, 2021b). As part of our research, we decided to focus on women's low education because its impact on fertility is more direct and pronounced in the context of Slovakia.

The level of employment in agriculture indicates a society's development level, which can impact the overall fertility level. Less developed regions often show higher fertility rates, influenced by various social and cultural factors. However, in the case of Slovakia, employment in agriculture is very low, representing only 2% of the economically active population, and this share does not differ significantly in individual regions. Although theoretically, higher employment in agriculture can be related to higher fertility, empirical data for Slovakia do not confirm this relationship. Based on the analysis of regional differences, we found that higher employment in agriculture does not automatically mean higher female fertility. Therefore, we decided not to include this factor in our research because it would not contribute to a more accurate understanding of the determinants of fertility in Slovak regions.

The influence of quality of life on fertility is complex and may vary according to specific conditions in a given country or region. A higher quality of life can lead to lower fertility due to better economic stability, higher education, and better access to health care. Conversely, lower quality of life may be associated with higher fertility due to traditional values and lower access to contraception and health care. In our research, however, we found that the quality of life factor penetrates our analyzed social, economic, and cultural determinants of fertility. Specifically, we considered different aspects of quality of life within these determinants, such as economic stability, education, and access to health care, and analyzed their individual impact on fertility levels. Therefore, we decided not to include quality of life as a separate factor but rather to integrate its various aspects into the broader context of our research.

Research focus

To understand what drives the differences in the rate of participation in fertility in the regions at the beginning of the twenty-first century, we create a unified empirical framework that enables comparative analyses in space Slovakia. The research goal is to recognize the effects of socioeconomic determinants affecting fertility in the Slovak regions. This objective can only be investigated to the extent that the items in the last population census in Slovakia and the public databases of the Slovakian statistical office are available. Therefore, answers to how a region’s culture contributes to fertility may remain unanswered. Correlations between fertility rates and selected determinants may provide insight into some correlates or primary predictors of fertility in these groups. They may open avenues for further, more purposeful research into the difference in fertility levels in Slovak regions.

Data and methods

Data

The study relies on publicly available statistics from the Statistical Office of the Slovak Republic (SOSR) and the Center for Scientific and Technical Information of the Slovak Republic (CSTI SR). The primary determinants are selected indicators from the demographic, social, and economic statistics in the public DATAcube database. The primary database consisted of data on women’s birth rate and fertility, followed by data on the unemployment rate of women, the average monthly income, and the number of completed housing units. The second section of the statistical data includes key data about religion, nationality, women's family status, and women's education from the last population census in 2021.

The CSTI SR delivers data on the number of children, pupils, and students at various types of schools in the Slovak regions. Childcare substitutability in kindergartens for children aged 3–6 years was investigated concerning fertility levels and nature in Slovak regions. The last census revealed the ethnic structure of Slovakia's inhabitants, and this information was used to estimate the Roma population, which has significantly different reproductive behaviors (Nestorová Dická, 2021). Correcting this data with the Atlas of Roma Communities (2019) gave a reliable reality compared to the implemented state population censuses (Šprocha, 2014).

Spatial research was implemented at the LAU1 level, covering Slovak districts at the regional level. Bezák (1996) proposed merging the Bratislavan and Košice city districts in separate spatial units for research purposes. They were not integrated as part of the study of the determinants of fertility, and the existence of 9 urban districts of Bratislava and 22 urban districts of Košice was preserved due to their diversity in terms of birth rate and fertility of the population, as well as socioeconomic determinants.

The basic spatial units consisted of 79 districts, diverse in spatial and population size. For example, Prešov, Nitra, and Žilina districts have more than 160,000 inhabitants. In contrast, Stropkov, Turčianské Teplice, Banská Štiavnica, and Medzilaborce consist of less than 20,000 people. Bezák (1996, 1997) pointed out Slovak regional inequality and injustice. The available data for regional analysis is only for these spatial units. Statistical analyses of women's fertility determinants provide ample opportunity to investigate socioeconomic effects in the aforementioned regional structure. The reference period for analyzing fertility data and selected determinants was 2019–2021, and the census data was from 2021.

Methods

Due to the primary purpose of the research, which is a regression analysis of the influence of socioeconomic determinants (Table 1) on the level of TFR fertility in the regions as a dependent variable, the key was the selection of determinants as independent variables, the choice of which was conditional on the results of many professional studies listed in the theoretical framework dealing with various socioeconomic effects on fertility.

Table 1 Overview of selected regional fertility indicators, determinants, and their statistical characteristics, 2019–2021

Initial regression analysis of the influence of selected socioeconomic indicators in spatial units on the fertility level revealed only fragile dependencies, and the correlation strength with different indicators varied in some regions. The reason is significant regional diversity in terms of fertility and socioeconomic determinants.

That led us to create a regional typification of spatial units, where individual types are quasi-homogeneous sets of spatial units with similarities in fertility nature and level. Factor analysis (FA) was crucial for the typification of spatial units. It reduced the number of intercorrelated input variables without much information loss and created new variables, i.e., factors (Nestorová Dická, 2013). FA assumes that each entering trait can be expressed as a linear combination of a few common latent factors.

The input database consisted of 10 variables related to various aspects of fertility and fertility (Table 1). The rate of Kaiser–Meyer–Olkin (KMO; 0.83), which tests the suitability of input data for FA by comparing the size of experimental correlation coefficients to the size of partial correlation coefficients, evaluated the original input variables as highly suitable for FA.

The factor model with the Principal component analysis (PCA) method was aimed at identifying interrelated groups of variables. The PCA method is one of the most frequently used in FA, identifying linear combinations of observed variables that maximize the overall variability of the data. FA used the PCA method to extract the components from the reduced correlation matrix, in which FA monitored the initial estimates of the communalities of the individual variables. For FA, the most significant factors determined according to Kaiser's eigenvalue criterion, i.e., factors with an eigenvalue greater than one, are considered statistically significant. Two primary factors can thus clarify the structure and fertility level in Slovakia regions (Table 2), which explain almost 91% of the total variability of the input variables.

Table 2 Factor loadings of the rotated factor matrix

Extracted factors with factor scores for each spatial unit express the degree of influence of individual factors (Fig. 1). The mentioned factors became the basis for regional typification. Relatively homogeneous regional classes concerning fertility's nature and intensity due to applied cluster analysis were created (Fig. 2). Where spatial units were classified through a hierarchical procedure using Euclidean distances, to which Ward's clustering method applies. Finally, the discriminant cluster analysis results verified the optimal spatial unit distribution for the selected number of types. The 10 resultant regional types were entered into regression analysis of selected fertility determinants in Slovak regions (Fig. 2 and Table 3).

Fig. 1
figure 1

Extracted factors differentiating fertility nature and intensity in the Slovak regions

Fig. 2
figure 2

Regional typification of fertility nature and intensity

Table 3 Fertility indicators in Slovak district clusters

The second part of the research focused on the knowledge of the effects of socioeconomic determinants on fertility in individual regional types, i.e., using regression analysis at two hierarchical levels (see Götmark & Andersson, 2020). Research first established the relationship between individual regional types, i.e. the investigated units here represent unique regional types with average values of TFR and socioeconomic determinants. The second level represents investigation within individual regional types at the level of spatial units, i.e. the investigated units here represent regional-type spatial units with their values of TFR and socioeconomic determinants. All analysed regional types had equal weight. For each socioeconomic determinant, a regression trend, correlation, and determination coefficient R2 were obtained at both levels, which, according to Klein (2020), provides the level of total variability explained by the independent variable. The model is more "effective" if its value is closer to 1, meaning the linear regression model explains a larger percentage of variability. Řehák (2023) adds that the coefficient of determination also represents a measure of statistical dependence between the independent and dependent variables, and its higher value indicates closer statistical dependence. The value of R2 expresses dependency strength, and the value of the coefficient β1 the character of the influence in the linear Eq. (1). A positive coefficient is established when Y increases as X increases, and negative values indicate an indirect effect. A linear regression model at the intraregional level was applied only for regional types containing more than one spatial unit.

$$ {\text{Y }} = \, \beta 0 \, + \, \beta 1X \, + {\text{ e}} $$
(1)

In the final phase of the investigation, the effects determinants of regional fertility were detected using the multiple linear regression method (MLR). The MLR evaluated the relationship between a continuous target, i.e., the factors from factorial analysis, and selected 9 socioeconomic determinants. The multiple regression model is developed from a simple regression model, where the dependent variable Y is a function of several independent variables X1, X2, X3,…, X9, and the residual component (error terms). The study uses an equation model (2), with Y as the dependent variable, X as the independent variable(s), β0 is the point where the regression line intersects the Y axis, β1 to β9 are the regression coefficients that determine the direction of the line, and e is the measurement error. After controlling for the effect of other predictors, the net effect of each independent variable on the dependent variable has also been measured through multivariate analysis MLR.

$$ {\text{Y }} = \, \beta 0 \, + \, \beta {\text{1X1 }} + \, \beta {\text{2X2 }} + \, \beta {\text{3X3 }} + \, \beta {\text{4X4 }} + \beta {\text{5X5 }} + \, \beta {\text{6X6 }} + \, \beta {\text{7X7 }} + \, \beta {\text{8X8 }} + \, \beta {\text{9X9 }} + {\text{ e}} $$
(2)

Results

As a post-communist country, Slovakia has had an insufficient level of fertility below the replacement rate since the beginning of the 1990s. Although some authors (cf. Šprocha & Bačík, 2021a, 2021b; Šprocha & Tišliar, 2016) emphasized the partial reversal of development trends during the new millennium, its increased level remains below the replacement rate, while the spatial picture of fertility in Slovakia is changing. Figure 3 highlights the spatial distribution of TFR values and fertility timing in Slovakia regions. A key factor in the spatial variability of fertility in Slovakia is the advanced process of postponing childbearing. Closely linked with it are indicators of the timing and age distribution of fertility, which contribute to the current differentiation of regions (Fig. 3). The existence of significant differences in the intensity and timing of fertility was also confirmed by studies such as Bleha et al. (2014), Šídlo and Šprocha (2018), Šprocha et al. (2019). During the new millennium, a vast region with very low fertility was formed in Slovakia, located in Slovakia's western and easternmost areas. The processes of postponing childbearing mark the districts of this region, and therefore, by a significant increase in fertility at an older age, and according to Šprocha et al. (2019), only by a very limited catch-up rate. Moreover, the capital region is forming an area with relatively favorable recuperation phase development. The region with a favorable level of fertility, a significantly smaller area, occupies areas in the middle east and north of Slovakia, in which the favorable level of fertility is a reflection of specific socio-demographic features (cf. Drinka & Majo, 2016; Nestorová Dická, 2021; Roupa & Kusendová, 2013).

Fig. 3
figure 3

Source: SO SR

Total fertility rate and timing of fertility in Slovakia regions, 2019–2021.

Factorial analysis and regional typification

The two factors extracted from factor analysis are independent variables with special links to human fertility nature and level in the Slovak regions. Figure 1 highlights the spatial distribution of factor scores and the relationship of regions to both generated fertility factors.

The first factor has a significant relation to indicators pointing to the age distribution of fertility, including the fertility of higher-order children and non-marital fertility; it can be interpreted as a factor in fertility timing, cohabitation, and family size. The factor covered up to 49% of the total variance of the variables. From a spatial perspective, the country's eastern region, particularly its southern and partly northern areas, exhibits a higher prevalence of fertility among younger women, along with increased rates of cohabitation and higher-order childbearing (Fig. 1). A peripheral location, underdeveloped economies with high long-term unemployment rates, inadequate or absent transport infrastructure, and lower human potential characterize regions. This is primarily evidenced by a low proportion of the population with higher education, a high proportion of residents from socially excluded environments, and marginalized Roma communities. The earlier onset of fertility, coupled with a higher prevalence of children born out of wedlock and in higher order in these regions, may be attributed to attempts to mitigate the financial uncertainty associated with the labor market (Potančoková et al., 2008) and ensure at least a certain income in unfavorable structural conditions, thanks to which the costs of lost opportunities related to childcare are low (Šprocha & Tišliar, 2019). In contrast, the North-Western geographical regions are prosperous and attractive to a higher-income population. The postponement of childbearing can be explained by the higher participation of women in the labor market; their higher education offers greater career prospects and economic independence in the labor market with building a job position (NíBrolcháina & Beaujouana, 2012), as a result, the costs of lost opportunities are high.

The second factor expresses the intensity of fertility, and it covers 42% of the total variability in the input information. Low and insufficient fertility, well below the replacement rate, is typical for many regions in the extreme east or west, south or central part of Slovakia (Fig. 1). Ethnic, cultural, and economic factors determine the higher intensity of fertility. Behind the low intensity of fertility can be peripherality or rurality (Nestorová Dická et al., 2019), bad economic situation, and a greater concentration of the population without religion or belonging to Protestant denominations. Finally, these communities generally have lower fertility levels, and often a higher percentage of the aged population.

Fertility regional typification in Slovakia was created by combining the obtained factor score levels. Figure 2 shows the ten regional types generated by cluster analysis, representing the different structures of human fertility and natality. Regional types A to E reflect low fertility and natality, with older women's more intensive fertility. They occupy the far east of Slovakia and the western region, except the capital region. Regional types F to J, which have higher fertility levels, represent the opposite. They are found in the central part of eastern Slovakia and the north. The positive extreme is represented by regional type J, which reached a higher fertility level above the replacement rate, with the fertility of primarily younger women and a higher birth rate of children of higher order. Its existence is connected with the Roma population, which has a significant presence mainly in the central part of eastern Slovakia. In contrast, the negative extreme is represented by region A, which has the lowest-low fertility according to the designation fertility levels by Kohler et al. (2002) and occupies Slovak districts primarily in the far east and western parts of the country.

The regression analysis applied to previous research results (regional types) established the effect of selected socioeconomic determinants on fertility (TFR) at two hierarchical levels. Because the cluster analysis was used, the variation in TFR values was relatively low for individual regional types inside, i.e., at the intraregional level. TFR values varied somewhat more between districts in regions J and F. Higher variability is between average TFR values of individual types, i.e., at the interregional level. Regional types G and I represent the highest levels of TFR (2.4–2.3), but Types A to H registered TFR values well below the replacement rate.

The key findings of the relationship between selected socioeconomic determinants and TFR in regional types are documented in Fig. 4. Regression analysis reveals that the fertility level of the population was negatively related primarily to the women's divorce rate, Childcare substitutability rate, urbanization, and monthly income at the interregional level. A positive trend was determined for the Roma ethnicity, the Catholic population, unemployment, and low education of women. However, residential building did not affect fertility. The women’s divorce rate had the closest relationship with fertility at R2 = 0.62 and Roma ethnicity at R2 = 0.44.

Fig. 4
figure 4

Total Fertility Rate (TFR) in Regional Types and its Relationship to Socioeconomic Determinants

Figures 2 and 4 highlight that regional type I, which represents only one district known for strong tradition, religious beliefs, and national unity with the low ethnic diversity of its population, records the lowest 3.7% average divorce rate with an extreme positive level of TFR. Type F, as the region of the capital with the hinterland, had the highest rate, with a value of up to 14%. The TFR is low below the replacement level. This stark contrast in divorce rates significantly impacts the TFR, which decreases with increasing divorce rates in regional types. On the contrary, the fertility level increased significantly with Roma ethnicity, where the average concentration of the Roma population varied from zero in types F and I to 8.1% in type J, where the TFR with a level of 2.3 children per woman represents the highest recorded value among the Slovak regions.

A significant correlation with TFR was also achieved with the degree of enrollment rate of children in kindergartens or the degree of urbanisation, while women's fertility decreases with their growth. The enrollment rate of children ranged from 535‰ in type J, with 2.3 children per woman, to 808‰ in type A, representing the lowest fertility level of 1.3. Similarly, it noted the indicator of urbanization, where values ranged from 12.3% in type I with a fertility level above the replacement rate to 82.7% in type F with a TFR of 1.6 children per woman. The amount of the monthly income also demonstrated a slight negative dependence. Research indicates that economically developed regions are directly related to a lower TFR. On the contrary, TFR values increased more significantly with the level of Catholicism and female unemployment. The proof is the generated types I and J, where the types reached the highest values of measures of Catholicism and simultaneously have the highest levels of TFR, i.e. more than 2.0 children. Regions dominated by the Protestant population record a lower TFR level below 1.5 children. Female unemployment was the lowest in the regional type of the capital city with hinterland (type F), with a value of 5.5%. The TFR of type F reached only 1.6 children per woman, and the highest fertility level, 2.3 children, was recorded in type J, with 20% of unemployed women.

From the point of view of the broader socioeconomic context, however, there are significant correlations also between the individual socioeconomic controls (Table 4) among the ten regional types. It was noted that the low educational level of women is significantly positively correlated with the Roma community, in which, at the same time, higher female unemployment persists with low monthly incomes, rates of children’s schooling, residential construction, or the rate of urbanization. The higher Roma community in regional types G, H, and J confirms higher unemployment and low incomes dominated by the rural population with a low level of residential development. The female divorce rate, on the other hand, indicates a significant relationship to urbanization with higher monthly incomes but with lower rates of Catholics, who are more likely to be concentrated in more rural regions. Residential construction primarily concerns regions with higher incomes, educational levels, and female employment. The aforementioned socioeconomic controls are documented by the Bratislava regional type F.

Table 4 Relationships with Total Fertility Rate (TFR) and Socioeconomic Determinants between Regional Types

The TFR and socioeconomic determinant relation are also analyzed at the level of individual districts within regional types. Correlations between TFRs and socioeconomic determinants are present in Fig. 5. However, the strength and nature of the relationships often differed from the previous level. Figures 4 and 5 highlight deviations in intraregional regression. We registered the most significant variations in four socioeconomic indicators. Prevailing opposite trends appear in the case of monthly income, Catholicism, and women's unemployment. On the other hand, residential building development showed a positive trend several times at the intraregional level, while independence was demonstrated only in one type, J. All regional types showed a negative association between TFR and urbanization, confirming the relationship reported in many studies.

Fig. 5
figure 5

Correlations between Total Fertility Rates and socioeconomic determinants

In five types, TFR increases with the increasing rate of women with low education, while this relationship is strongest in Bratislava type F. In five regions, the TFR decreases to the growing women's divorce rate, most in type C. Surprisingly, the opposite but significant trend was noted in type A. Regression analysis significantly and positively confirmed the association between fertility and Roma ethnicity in four types, but the relationship was negative in three.

Types C and F show a certain similarity in the observed trends, which, apart from the high intensity of dependence, differ only in the unemployment of women, where the opposite trends have taken place. Regional type D, significantly spatially disjointed with districts scattered throughout the country, did not capture a significant relationship to the given determinants.

Compared to the relationships found at the interregional level, no type showed complete similarity in the relationship of TRF to socioeconomic determinants. However, the most significant similarity was found in Bratislava type F, whose districts showed opposite trends only at Roma ethnicity and residential building. Regional type A shows significant differences, with confirmed tendency only in women's low education or the degree of childcare substitutability. The closeness of the relationships is not significant.

In assessing the intensity of the association between TFR and the individual determinants, it is evident that the individual region types are clearly associated with a different combination of socioeconomic determinants (Fig. 5). This is also evidenced by the one-way test ANOVA, which shows the presence of statistically significant differences between the region types and the socioeconomic indicators (Table 5). Thus, there are significant differences in TFR according to women's socioeconomic background.

Table 5 Differences in socio-economic determinants in the regional types

Determinants of the spatial differentiation of fertility

In three separate multivariate analyses, the factor scores of both factors and TFR represented the dependent variables. A multiple linear regression (MLR) model was used to determine the joint effect of the nine socioeconomic determinants, using the Enter method to determine statistical significance with respect to the dependent variable. Before conducting the multivariate analysis, regression assumptions such as normality of residuals, homoscedasticity, multicollinearity, and independence of residuals were tested. The significance level for all tests was p < 0.05 and p < 0.001. The scientific interest was directed to the knowledge of primary predictors of Slovak regional fertility.

A linear relationship was initially assumed between each independent variable and Slovak regional fertility. The MLRs included only statistically significant effects of the chosen variables. The values of the R, R-squared, and Adjusted R-squared multiple correlation coefficients revealed a gradual decrease in model significance. However, the models retained high quality with high significant values, and ANOVA confirmed they all explain a substantial percentage of regional variation. Table 6 highlights that the variables included in the models can explain a significant 92 per cent (p < 0.001) of the variance in fertility at fertility timing, a substantial 69 per cent (p < 0.001) in fertility intensity, and a significant 68 per cent (p < 0.001) in total fertility rate.

Table 6 Regression coefficients for regional fertility prediction

The prediction of the first factor value, which explains the difference in fertility timing in Slovakia's regions, was found to be statistically significant with only seven variables. Ethnicity and divorce rate did not affect the first factor and were excluded from multi-linear regression. However, multivariate analysis determined that low women's education, residential building, and women’s unemployment levels are significant predictors of fertility timing. The coefficients in Table 6 represent the effect of each independent variable on the dependent variable, which highlights that women’s low education significantly and positively affects the value of the fertility timing factor at p < 0.001. Increasing women's low education rate tended to increase the value of the given factor (β = 0.348). The coefficient of unemployment (β = 0.176) showed a significant (p < 0.05) positive effect on the timing of fertility, i.e. at the fertility of younger women. Other variables had a negative effect on the model. The low intensity of residential building development significantly and negatively affects the timing factor's values, indicating fertility in older women (β = −0.288). Similarly, the coefficients (β = −0.173, β = −0.152, β = −0.151, β = −0.118) showed significant influence on the timing of older women's fertility at p < 0.05. These are the respective β coefficients for child kindergarten enrollment, income, urbanization, and Catholicism. This indicates that when the enrollment rate of children in kindergartens increases, the income increases, the rate of urbanization and Catholics also decreases, and the factor timing values decrease, which indicates the fertility of older women. These findings significantly impact our understanding of fertility patterns in Slovakia's regions.

The prediction of the value of the second factor, which clarifies the difference in fertility intensity in Slovakia's regions, was statistically significant, with only four variables at the 0.001 significant level. Five of the original variables were excluded from the model. The MLR revealed that Roma ethnicity, monthly income level, residential building, and the women's divorce rate significantly affect fertility intensity in the Slovak regions. Notably, increasing Roma ethnicity tended to increase the values of the given factor, i.e., fertility intensity (β = 0.550). Similarly, fertility intensity in Slovakia's regions increases with monthly income or residential construction development (β = 0.479, β = 0.388). The women's divorce rate indicator was negatively affected in the model. The lower rate of divorced women has a significant and negative effect on the values of the factor, indicating lower fertility intensity (β = −0.533).

The final multivariate analysis was examined using the total fertility level, which also indicates the potential reproductive competence of the population. As a statistical measure, it estimates the average number of children a woman would bear in her lifetime if current age-specific fertility rates remain constant. It is a valuable tool for demographers and policymakers to understand and forecast population trends. The prediction of the level of TFR was statistically significant with five variables at the 0.001 and 0.05 levels, with Roma ethnicity (β = 0.697) and residential building (β = 0.302) indicating significant positive predictors of the level of TFR in the regions. Low levels of the above variables predict low levels of TFR. On the other hand, significant negative predictors of the level of TFR in the regions are unemployed women, enrollment of children in kindergartens, and divorced women in the region's population. If the values of the above variables increase (β = −0.345, β = −0.306, β = −0.275), the level of TFR decreases.

Discussion

The demographic transition and the related changes in people's reproductive behavior also affected the Slovak regions from the late 1990s (Mládek, 1999). The postponement of childbearing, the reduction of the marital birth rate, and the increase of one-child families and childless families due to the changed living, social, economic, and political conditions of the socialist model of reproductive behavior were the main reasons for the significant change in the fertility of the Slovak population with a decrease in the birth intensity and its stabilization far below the replacement rate (Šprocha & Tišliar, 2016). These findings are the driving force for the current research on fertility and natality under the conditions of Slovakia (Babinčák & Kačmárová, 2023; Bleha & Ďurček, 2019; Lentner & Horbulák, 2021; Šprocha & Fitalová, 2022; Šprocha & Tišliar, 2019; Šprocha et al., 2022). Moreover, the study's results highlight the main determinants of regional fertility in Slovakia, which are spatially significantly differentiated due to their nature and intensity.

The basic dimensions of Slovakia's regional fertility were developed through factor analysis, which provided a two-factor solution to the fertility assessment model. The two-factor model is supported by Šprocha et al’s (2022) independent research, but variations in input variables reverse the factors’ nature. The first factor indicates the nature of fertility, which differentiates regions in terms of the timing of fertility, and according to Šprocha and Šidlo (2016), the timing of fertility is the main factor in the variability of fertility in Slovakia. The second factor captures fertility intensity, which is influenced by demographic, socioeconomic, and cultural determinants, as well as the timing of fertility (Šprocha & Bačík, 2021a, 2021b). To construct primary regional fertility types, a two-factor model of fertility was applied using selected socioeconomic determinants at two hierarchical levels. The cluster analysis generated almost homogeneous regional types of fertility nature and intensity. Regions A-F show low fertility levels, and G-J have higher fertility above the national average. The number of clusters is purposefully higher than those examined in Šprocha et al. (2022). This provides a more detailed investigation of regression analysis with selected determinants.

Research on the inter-regional fertility level confirmed the expected relations for the selected indicators, especially about its nature, which is less connected with its strength. Women's divorce rates show negative dependence, which is consistent with the findings of some studies (see Meggiolario & Ongaro, 2010; Van Bavel et al., 2012) that demonstrate divorce has a negative effect on human fertility. A low Slovak female divorce rate is, therefore, a guarantee of higher TFR values, as evidenced by regional types I and J. The study found a positive trend in the Catholic religion, in line with the findings and theoretical concepts of Philipov and Berghammer (2007) and Zhang (2008), albeit one that is less pronounced as the women's divorce rate, which is consistent with their findings and theoretical concepts. Catholics are among the religious groups that support marriage and children and oppose contraception and abortion, which would decrease their fertility. Their less pronounced dependence confirms Šprocha and Tišliar's (2019) claim about losing the power of religiosity as a differential factor in reproductive behavior in Slovakia. However, it is still true that Slovak regions with a low concentration of Catholics achieve low TFR. The high and positive dependency on TFR among Roma ethnicity is related to their significantly different reproductive behavior (Nestorová Dická, 2021), which manifests itself in higher fertility intensity, especially in a socially excluded environment (Šprocha, 2014). Areas of higher concentration of the Roma population also guarantee a higher level of TFR. As in this study, Roma ethnicity was a significant predictor of fertility by Šprocha and Bleha (2018). Educating women plays a significant role in the final level of fertility, particularly in influencing the TFR. Research indicates that women's low education is directly linked to higher TFR, as lower educational attainment often correlates with earlier and more frequent childbearing (cf. Šprocha et al., 2020). Our analysis confirms this, showing a statistically significant relationship between lower educational levels and higher TFR across Slovak regions.

At the same time, socioeconomic controls revealed the connection of the Roma community with a low level of education, women's employment, and income, which was also pointed out by Rosičová et al. (2009), Preoteasa (2013), Andrei et al. (2014), etc. Similarly, urbanization, documenting the advancement of human civilization with social and economic progress, is more closely related to higher education (Marginson, 2016) and divorce rate (Zhang et al., 2014) than rural areas, as well as higher economic potential, which is documented by higher incomes or GDP (Henderson, 2003; Sancar & Sancar, 2017). Research indicates that economically developed regions are directly related to a lower TFR, as a higher level of education of women often correlates with later and less childbearing (cf. Šprocha et al., 2020). Several studies have also confirmed the relationship between religion, including Catholicism, and the divorce rate (e.g., Corley & Woods, 2021; Sander, 2019) and the availability of childcare facilities and unemployment for women and men (Kim, 2018; Legazpe & Davia, 2019).

Similarly, women's unemployment had a positive dependence. This is explained by Becker's neoclassical theory, which states that unemployment should reduce the cost of lost opportunities by providing time for childbearing and child care, thus promoting human fertility (Cazzola et al., 2016). Urbanisation exhibited a negative trend in fertility outcomes, which can be explained by the compositional or context hypothesis (see more Kulu, 2013; Kulu & Washbrook, 2014). Riederer and Buber-Ennser (2019) add that the main reason for the postponement of births in urban regions, where actual fertility rates are lower, in Eastern European countries may be precisely the difference in the context of the urban-suburban-rural gradient. Although monthly income shows a negative influence, it has a less pronounced tightness, which agrees with the earlier findings of Marenčáková (2006), who also confirmed the negative relationship between income and fertility in Slovakia.

The kindergarten enrolment rate has a significantly negative fertility impact in regions where the total fertility rate decreases as the enrolment rate increases. There are contrasting reports of childcare's influence on fertility. Some researchers consider a positive effect of available childcare but add that the results of several empirical studies do not support the hypothetical expectations (Wood & Neels, 2019; Ridfuss et al., 2010). Our research confirmed this, but the relationship may be related to many children from a socially disadvantaged environment without pre-primary education. However, Varsik (2019) considers that children from marginalized Roma communities have low enrolment rates, which conditions its negative effect on their communities. These communities are more concentrated in regional types with a higher level of fertility.

Only residential building demonstrated independence, but findings vary. Some confirm the causal relationship between housing affordability as the independent variable and fertility behavior as the dependent variable, often observed in less developed societies where the state controls the housing market (Felson & Solaún, 1975). Kostelecký and Vobecká (2009) found the existence of specific connections in Czechia when the improvement of the economic situation (increase in GDP and housing construction, decrease in inflation and unemployment) was accompanied by an improvement in the availability of owner-occupied housing and an increase in the birth rate. While the importance of housing availability has been scientifically proven, Salvati (2021) recorded that the positive impact of residential building was confirmed only in the initial phase of urban development. Considering the observed independence at the inter-regional level and the prevailing positive dependence at the intra-regional level, research in this direction requires further investigation into other housing factors that significantly impact fertility outcomes. For example, affordability (Clark et al., 2020) may significantly impact actual birth rates and TFR more than the development of new residential buildings.

However, the current nature and strength of the relationship between selected socioeconomic determinants and TFR are not identical within the individual types at the intraregional and interregional levels. From the assessment of the impact of TFR, it follows that individual regional types have different associations with the determinants studied, which was also confirmed by a one-way ANOVA test with statistically significant differences between regional types and socioeconomic determinants. Thus, the research demonstrated significant differences in TFR depending on the socioeconomic background of women, which was also confirmed in the study of Polykretis and Alexakis (2021).

The two MLR analyses determined that urbanisation, Catholicism, and women’s low level of education do not influence fertility level or intensity. However, these variables significantly affect fertility timing and differentiate younger and older women’s fertility, while the variable of Roma ethnicity is not statistically significant, which contradicts other analyses. In Slovak regions, fertility timing is not only associated with Roma issues but also in some "non-Roma" regions, fertility of younger women is significant, especially in the south of Slovakia (Levice, Poltár, Veľký Krtíš, Krupina districts), where fertility timing is also associated with the increase in out-of-wedlock births (Šprocha & Tišliar, 2021), which was predicted for both Hungarian and Roma populations. In addition, Roma ethnicity is a significant predictor of fertility intensity and level in Slovak regions. This is supported by the results of the studies, e.g. Šprocha et al. (2022), Nestorová Dická (2021), Szabó et al., 2021, Šprocha, (2014), Potančoková et al. (2008), which point to the higher reproductive characteristics of Roma population in Slovak regions. Šprocha and Bleha (2018) add that at least two-thirds of the overall variability is explained by some selected "socioeconomic" indicators and the share of the Roma population.

In the initial phase of the research, it was established that variations in fertility and birth rates are predominantly stratified along the urban–rural continuum. Notably, rural areas in Slovakia exhibit significantly higher fertility rates. However, as Nestorová Dická et al. (2019) highlighted, extreme rural municipalities characterized by low population density experience notably diminished population reproduction rates. Consequently, it becomes evident that fertility within Slovak rural areas primarily thrives within suburban municipalities and those with moderate population sizes. Nevertheless, two multiple linear regression (MLR) models indicate that the urban–rural gradient exerts negligible influence on the intensity and magnitude of fertility outcomes within the Slovak interregional framework, a finding corroborated by Šprocha et al.'s (2022) study in Slovakia. The distribution of the populace across urban and rural settings finds validation in fertility timing, as underscored by the research conducted by Riederer and Buber-Ennser (2019).

Unemployment has also become an important positive predictor of fertility timing, but it has a negative effect on the total fertility rate. Slovak economically developed regions (Korec, 2009) achieve higher fertility levels with high employment (Šprocha & Bleha, 2018) due to selective migration (Novotný et al., 2023), but also housing availability. On the other hand, Šprocha et al. (2022) supplement that young people in this area face problems related to high housing prices and low childcare substitutability rate (Križan et al., 2022; Madajová et al., 2021) due to insufficient capacity. Consistent with these conditions, an important finding of Mills et al. (2011) is that educated people are more likely to focus on building a career and, therefore, seek to combine parenthood at an older age, which affects their fertility outcomes, confirmed by the results of the first MLR.

The most important determinant affecting regional fertility differences is women's education, especially young women's education, according to research by Kostelecký and Vobecká (2009). The authors add that when women's education is controlled, housing affordability plays an important role in explaining regional differences in fertility—both the total fertility rate and the fertility timing. However, women’s low education was not significantly related to the level or intensity of fertility outcomes in Slovak regions. Götmark and Andersson (2020) support this finding in Eastern European countries, where there was no or a weak relation between education and fertility outcomes.

Conclusion

Slovakia's reproductive potential seems to have been stabilised for a long time below the level of insufficient population replacement. Therefore, research into understanding and predicting total and regional fertility with the prediction of further development is important for various scientific disciplines and interest groups.

Each determinant influencing fertility has specific spatial patterns with unequal regression coefficients at different regional levels, which cannot be summarised in a constant way. Therefore, implementing multivariate statistical techniques can adequately describe the relationships between fertility and its determinants. Mathematical-statistical techniques can potentially contribute to demography, geography, and social sciences by offering usually invisible solutions, enabling practitioners to better understand relationships' spatial perspectives. In this context, the study was to reveal the effect of a set of nine socioeconomic determinants on the level of regional fertility. To achieve this, spatial changes in the distribution of fertility and determinants were captured, and spatial heterogeneity in their relationships was examined using regression techniques.

We analyzed fertility variations across Slovak regions, examining the convergence or divergence of our regional findings about widely accepted theories on the determinants influencing fertility levels. The results highlighted that selected socioeconomic determinants were not unnecessary in any case, and they all showed certain regional connections about fertility nature and level. We have demonstrated that regional variations in fertility rates arise from economic, social, and cultural factors. The models differ regarding the nature and intensity of fertility in relation to the urban–rural dichotomy, the educational or Roma issue. The persistence of geographic variation will be important for understanding Slovak regional fertility levels.

Knowledge of prevailing geographic and population influences is essential to understanding Slovakia's regional fertility levels and intensity. Given that fertility is vital for social assessment and policy formulation, the study's findings could help local decision-makers and planners identify the socioeconomic conditions underlying fertility at the regional level and plan appropriate intervention strategies.