THE INCOMPATIBILITY OF TURKISH WOMEN’S EDUCATIONAL
ATTAINMENT AND OCCUPATIONAL
PARTICIPATION
Uzm. Aslı Ermiş
ABSTRACT
This study, in which European Social Survey second round data for Turkey are analyzed, explores the incompatibility of Turkish women’s educational attainment and occupational participation rates. Education could be seen as the primary determiner of the probability to be in the labour market; contrary to what is seen in other countries; however, the Turkish case shows a unique pattern. In Turkey, there is a simultaneously decreasing trend in women’s occupational participation rates even though the number of women in higher education has been rising, especially during the last two decades. As there may be a number of different reasons for this pattern, this study considers the impact of women’s and husband’s education together with age, the presence of children and traditional attitudes on Turkish women’s employment rates. The results display the fact that as the effect of women’s education on their probability to be in paid work is offset when marital status and children at home are considered; husband’s education affects this probability negatively. This first result is related to the first hypothesis that women tend to see education as an advantage for their domestic roles. Related to the second hypothesis; considering the results showing high educational homogamy for highly educated women; high education might be a motivation for a highly educated thus a better provider for the future family. Besides; the traditional attitudes of Turkish women, the double burden they face in the public and private spheres, the lack of necessary conditions in the work place and the unequal labour market circumstances affect their position –equally as a result of their own decisions as well as external impacts and pressures- in the labour market. As a conclusion, there are current constructive and positive aspects of the situation that there are progressive happenings in the labour market position of Turkish women despite the current continuing negative facts, which are expected to become different in the future with the help of positive social, economic and cultural changes.
Key words: Higher education, women in labour market, (in)compatibility between educational
background and occupational participation.
ÖZET
Türkiye’yi de içeren Avrupa Sosyal Anketi (ESS) ikinci kısım verisinin kullanıldığı bu çalışmada, Türk kadınlarının eğitime katılımları ile iş yaşamında yer almaları oranı arasındaki uyuşmazlık ele alınmıştır. Eğitim, iş gücü piyasasında yer almanın birincil etkeni olarak görülebilecekken, diğer ülkelerin aksine, Türk kadınının eğitim ve iş yaşamına katılımı oranları arasındaki denge farklı bir örüntü göstermektedir. Özellikle son 20 yılda yüksek öğrenime katılım oranındaki artış; yine aynı zaman diliminde düşen işgücüne katılım oranlarıyla ters düşmektedir. Bu eğilimin birçok farklı sebebi olmakla birlikte, bu çalışma konuyu; kadının kendi ve eşinin eğitimi, yaş, çocuk sayısı ve geleneğin etkisi gibi değişkenleri de göze alarak incelemektedir. Sonuçlar, kadının eğitiminin çalışma hayatında olması üzerindeki olumlu etkisinin, medeni durum ve çocuk sahibi olma gibi etiketler devreye girdiğinde zayıfladığını gösterirken, eşin eğitimi de bu durumu olumsuz olarak etkilemektedir. Bu sonuç, istatistiksel ve literatür çalışması sonuçlarıyla da bütünleştirilerek, çalışmanın birinci
hipotezine bağlı olarak, eğitimin ev içi sorumluluklarda bir avantaj olarak görülme eğiliminin varlığı olarak değerlendirilmiştir. İkinci hipoteze bağlı olarak ise, yüksek eğitimli kadınlarda görülen yüksek eğitimsel homogami sonucu; eğitimin, yüksek eğitimli; dolayısıyla kurulması beklenen aile için kadınla eş eğitim durumunda ve aileye iyi bir yaşam sağlayabilecek biriyle hayat kurma yolunda kısmen de olsa motivasyon olabileceği sonucuna diğer verilerin (istatistiksel ve literatür içinde) desteğiyle varılmıştır. Ayrıca, Türk kadınının geleneksel yapısı, kamusal ve özel alanda karşılaştıkları “çift yük” (ev sorumlulukları ve iş sorumlulukları), iş yerindeki gerekli şartların sağlanmaması ve eşitsiz iş yaşamı koşullarının da kadınların iş piyasasındaki pozisyonunu –hem kendi kararları
doğrultusunda hem de dış etken ve baskılar doğrultusunda- etkilediğini göstermiştir. Çalışmanın sonuç bölümünde, Türk kadınının iş piyasasındaki durumuna dair güncel olumlu durum ve yönlere
değinilmiş; olumsuz olarak süren bir takım şartlara rağmen geleceğe yönelik sosyal, ekonomik ve kültürel değişimlere dayalı yapıcı gelişmeler ve beklentiler sunulmuştur.
Anahtar sözcükler: Yüksek eğitim, iş yaşamında kadının durumu, bireyin eğitim durumu ve iş
piyasasındaki pozisyonu arasındaki uyum(suzluk).
Introduction
The phenomenon of “college educated housewives” has been an up-to-date topic for a long time in Turkey. For several decades now, Turkish women’s attainment in higher education has been
increasing dramatically; however the labour force participation rate of women in Turkey has remained low, and has been stable since the late 1980s. This study sets out to investigate the relationship between women’s educational attainment and labour force participation by considering some of the social, cultural and economic aspects of Turkish women’s position in the labour market and in wider society.
Since the 1960s there has been a significant rise in Turkish women’s educational level, including in higher education. As Table 1 shows, in 1960, 1.2% of Turkish women were enrolled in higher education, compared to 19% by 2000. This increase has resulted in a decrease in the gap between men’s and women’s educational attainment rates in Turkey. In 1960, the enrollment rate for men was nearly 4 times higher than that for women, but by 2000 men’s enrollment in higher education
institutions was only about 1.4 times greater. This education pattern over time for Turkey is similar to other OECD countries, in that the educational attainments of men and women have been increasing and converging. In addition, although Turkey has a low overall rate of higher educational enrollment when compared to the OECD average, the gap between men’s and women’s enrollment is no bigger than many other OECD countries (OECD 2006: 19).
Table 1: Percentages of Turkish men and women enrolled in higher education by decade (Source:
Turkish Institute of Statistics, Statistical Indicators 1923-2004 pp. 65)
Table 2: Percentages of Turkish men and women in paid work (employed), age of 15+, by 4 year
intervals (Source: Turkish Institute of Statistics, Statistical Indicators 1923-2004 pp.153)
0% 10% 20% 30% 40% 50% 60% 70% 80% 1988 1992 1996 2000 2004 Male Female The gap between men and women in higher educational enrollment has decreased over time (odds ratios for men: women)
1950 1960 1970 1980 1990 2000 4.00 3.66 3.83 2.74 1.93 1.36 0% 5% 10% 15% 20% 25% 30% 1950 1960 1970 1980 1990 2000 Male Female
As seen in Table 2, this trend in relation to Turkish women’s rising levels of education is not mirrored in the patterns of Turkish women’s labour force participation. Unlike the increasing educational attainment rates, the labour force rate of women in Turkey has remained low over time. As Table 2 shows, the labour market participation of women was 30.6% in 1988; but had decreased to 28.6 % by 1995 and 22.9% by 2004. This decrease is also reflected in the gap between men and women in their propensities to work. In 1988, the labour force participation level of men was 2.45 times greater than that for women and, in contrast to the decreasing gap between the educational attainment levels of the genders, the gender gap in labour participation rate increased to 2.83 by 2004. Turkey’s pattern in women’s workforce participation is unlike that of most other OECD countries, where as Figure 1 shows, the rate of labour participation for women has generally increased over time for many other countries. For highly educated women in particular, among all OECD countries, Turkey has one of the lowest labour force participation rates for women at 65% in 2004 compared to the OECD mean of 76%, even if the gap of participation in labour market between men and women with tertiary
education is not the worst among all countries. (OECD Statistics, Education at a glance, Table A10.1a)
Figure 1. Employment rates of women in OECD countries in percentages (OECD Economics
Department, 2004. Female Labour Force Participation: Past Trends and Main Determinants in OECD countries pp. 2, Figure 1)
Given these contrasting patterns, the main aim of this study is to evaluate the underlying reasons for the apparently negative impact of women’s rising education levels on women’s likelihood of working. Because the study draws on cross-sectional rather than longitudinal data, the study mainly focuses on the allied question: how does Turkish women’s education affect their likelihood of working?
In particular, this work investigates two types of explanations for the presented question. The first potential explanation is that college education is seen by Turkish women as a fulfilled aim rather than
The gap between men and women in labour market participation has increased over time (odds ratios for men: women)
1988 1992 1996 2000 2004
a path for a career. This is consistent with the work of Acar (1994) who argues that education is a “golden bracelet”1
for Turkish women not only in the economic sense but also as a means of prestige itself. Acar (1994) adds that although the “golden bracelet” effect played a crucial role historically for women, allowing them to participate more in the labour market; in society, this effect remained “as a minority effect”, implying that a smaller than expected proportion of women benefited from the expansion of education, which could also be seen parallel to the relatively lower educational attainment of Turkish women compared to women in other OECD countries, even if there is an increasing trend.
Having said that college education could be seen as an aim for itself rather than a particular means for a career, it could be added that even if higher education is seen as a path, it may be more oriented towards domestic career, rather than a labour market one, if higher education is seen by women as making them a more eligible spouse and perhaps also a better qualified parent. As Turner and Bowen (1999) suggest, women may have different and more domestic oriented life plans compared to men’s traditional “breadwinner” gender norm orientation. For women, education could be seen as a way of fulfilling their domestic motivations. To analyze this hypothesis, whether a woman is in paid work or not will be statistically evaluated by regressing it firstly with educational level and then adding variables such as marital status and gender role attitudes. I expect to find that being married or having traditional gender role attitudes are more significant predictors of being a working woman than the label of being highly educated in the Turkish social context.
The second potential explanation draws on the idea of educational homogamy and the links between husband’s education and the likelihood of (married) women to be in paid work or not. The idea is that women invest in education, partly because it gives them an advantage in the “marriage market”: it makes them more eligible candidates for a better spouse for themselves and their future family, and it gives them a greater chance of meeting a spouse with a high level of education because marriage markets are often structured by education (Mare, 1991). So, considering the idea that there is a general tendency of Turkish women to prioritize their roles in the private sphere, thus to cut down on paid work for the sake of family life; the second hypothesis of this study aims to investigate whether women in Turkey seek higher education because more educated husbands are expected to be paid more, and are better able to provide for their families, and as a result gaining access to these partners could be a substitute for women to work in particular cases. This is associated to the general
expectation for Turkish women to choose their domestic roles as a priority; particularly because it is generally not seen appropriate for a married woman to work, especially for those with younger children. In addition to the cultural aspect, considering that both work life and domestic
responsibilities are difficult to maintain, spouse choice is crucial for the future family life of a woman. As Blossfeld and Timm (2003) point out, educational expansion increases the likelihood of finding a spouse with the same educational level and as Kalmijn (1998) emphasizes, people tend to get married to people close in status. This implies that as the educational attainments of both women and men have increased in Turkey, the extent of educational homogamy is also likely to have increased. As the literature suggest, men and women tend to marry someone from the same educational level because this is linked to school completion time, also the same “time out of school” (Blossfeld and Timm, 2003).
Smits et al. (1998) suggest that educational homogamy means that ascribed status is replaced by achieved status, which suggests that getting a higher education may be seen by women as a way to be more competitive candidates for marriage, whereas a career path might be considered as a secondary aim. DiMaggio and Mohr (1985) emphasize that, for both men and women, the “cultural capital” effect – in other words, the benefit coming from the spouse’s cultural accumulation, prestige and way of life – is more significant than the effect of father’s education on the likelihood of being married to someone with at least the same status. They also suggest that from an “exchange perspective”, women and men expect their spouses to be educated because they would have more cultural capital. This could demonstrate that as a country is getting more industrialized as Turkey, education is one of the biggest sources of cultural capital, and thus essential for being a more desirable candidate as a
potential spouse. The fact that education provides cultural capital could make education a fulfilled aim for women rather than a career path. In other words, women may also be pursuing higher education to increase their changes of marrying a better provider for their future families rather than only because they particularly intend to work.
1 A concept in Turkish cultural context generally used for education referring it’s value as an
It is important to add that the DiMaggio and Mohr (1985) study suggests that women’s work decisions are economically rational and this fact is supported by the observation that Turkish women, as in other countries, are less well paid compared to men and generally disadvantaged by not being offered the same employee advantages as men (Reskin and Hartmann, 1986). This fact brings us to the point that additional to interference from outside, there are also reasons for Turkish women to decide not to work by their own will. Gündüz-Hoşgör and Smits’ (2008) analysis of survey data for Turkey suggests that of all the reasons women give for not being employed, looking after children and doing housework account for one third, which is higher than the other reasons. However, after indicating these results, Gündüz-Hoşgör and Smits (2008) add that one fourth of women said that they didn’t work because of the pressure from their families and/or husbands. The authors point out that husbands with more education generally are more likely not to let their wives work and that highly educated women in Turkey tend to marry highly educated men (which means high educational homogamy), which is consistent with the idea that highly educated women might tend not to work since they have the alternative to depend on their husband’s income, which would be high-level considering the high educational level of the husband. Even if higher education does not always guarantee higher-paid jobs; the general expectation would be in this direction that “the earnings of more educated people
are almost always well above average” (Becker, G., 1975). So, it could be pointed out that even if
there are women claiming that they are not working due to pressure from outside, it is equally likely that there is a high rate of women – perhaps particularly among highly educated women married to highly educated men – who are actively choosing not to work.
Given that Turkey is 54th out of 115 countries in equality of payment for men and women, according to the 2006 Global Gender Equality Report (TISK raporu, 2008), this suggests that while gender inequalities are not particularly worse in Turkey in comparison to other countries; men, because of the gender norms, are still seen as the main breadwinners, and women as added and/or secondary workers (Onaran and Başlevent, 2003). This fact may put Turkish women on an unequal footing in the labour market and lead them to decide not to be in the labour market considering the balance of costs and benefits it would bring. In other words, the choices of higher educated women may be active choices, and women may tend not to choose working as a rational decision since they expect to be
discriminated against and not to be paid equally.
It is also important to note that being a housewife is still seen as an occupational category in Turkey, and one that is still prestigious both economically and culturally for many women (Özer and Biçerli, 2003). Moreover, the boundary between “not working” and “being a housewife” is blurring. Özer and Biçerli (2003) emphasize that low educated women seem to be participating in labour force less than they actually are because they work in less prestigious jobs such as cleaning and thus they prefer to use the term “housewife”. In addition, “retirement insurance” has recently begun to be applied to housewives by a private Turkish insurance company, which again reflects the fact that being a “housewife” is seen as an occupation.
Data
The data I will analyze for my question of the incompatibility of Turkish women’s increasing
educational level yet stable occupational attainment are taken from the European Social Survey, round 2, in which Turkey is included2. I decided to use this data set after confirming that the data have the necessary variables for my study and after reaching representative results through descriptive statistics.
ESS is a multi-country survey (fielded in over 30 countries); of which the first round was conducted in 2002/2003. The second and third rounds were fielded in 2004/2005 and 2006/2007. The direction of the project is undertaken by a central coordinating team at the Centre for Comparative Social Surveys, City University, London, led by Roger Jowell.
The ESS questionnaire has two main parts (which includes about 120 items); the first one is the “core module”, which is asked in every round and additionally, there are two or more “rotating modules”, which are asked in a repeated pattern at intervals. The purpose of the core module is to evaluate and inspect change and continuity considering “media use, social and public trust, health, well being and security, religion, socio-cultural and socio–economical means, levels of national and ethnic and religious commitment, government and efficacy, level of political interest and taking part, social values and exclusion, political and moral values and demographics.”
Also, a supplementary questionnaire is applied at the end of the main interview. The first part of this questionnaire is a human values scale (part of the core), as the second part is designed to evaluate the
2 http://ess.nsd.uib.no/index.jsp?year=2005&module=main&country= (access with registration),
reliability and validity of the items in the main questionnaire. ESS is a 60-70 minute survey, including the background questions (http://ess.nsd.uib.no/, accessed in July, 2008).
As mentioned above, the second round of the survey, which includes Turkey, was conducted in 2004/2005. The Turkey data include 1856 respondents, of whom 1029 are women, which is the target gender for my study. Many respondents are from the biggest city of Turkey, Istanbul; which is taken as a region because of its different socio-economic structure and level of development. The main variables, which will be described in detail in the next section, will be women’s highest educational level, current employment status, marital status, presence of dependent children, region, husband’s educational level and questions linked to gender values in society.
Limitations of the data
There are some limitations to the data linked to its structure and sampling size. Firstly, the sample size might be problematic in that it includes a relatively small number of highly educated women;
however, this could also be considered the strength of the dataset because it reflects the society, since in Turkey, the percentage of women who are highly educated is still relatively low compared to those in other countries. Secondly, there is a problem that ESS is not a longitudinal data set and the trends over time are not seen; however, as mentioned earlier, this study will therefore take a cross-sectional approach suitable to the data.
Benefits of the data
The benefits of the ESS (second round) data are that the variables are used very eligible for the aims of this study since there are questions about how married women’s working life is perceived by both genders and how priorities are addressed. As the variables concerning the respondents’ and the partners’ main activities are described in detail, there are also questions about spouses’ educational levels, which is crucial for this research to see the effect of husbands’ education on married women’s decisions to work or not to work.
Variables
For the data analysis, a data set saved as “ESS data-women only” is used, which includes women in working age (18-55).
Education
The highest level of education of the respondent (edulvl) is the primary variable. Since this variable was coded into 6 levels of education except missing answers; they are recoded into three main variables: “low” (no primary& primary or first stage) “medium” (lower secondary& upper secondary) and “high” (first and second stage of tertiary).
Main activity
For the main activity variable, which records what the respondents had been doing mainly for the past seven days, firstly, the responses, which are not relevant for the study (retired, in the military, etc.) are dropped and only three responses for the purposes of running cross tabulations are kept; “in paid work”, “unemployed but looking for a job”, and “doing housework”. Secondly, for the purposes of the logistic regression models, a new variable is created as “activity” , which includes only paid work and housework, which are the main concerns of the research. For the logistic regression, dummy variables are created to see whether a woman is engaged in paid work (coded as 1) or in housework (coded as 0).
Year born
This variable is recoded as “age” to make it possible to see the effect of age and age squared on the likelihood of being in paid work rather than in housework in the logistic regressions. It is also used in interaction with women’s labour force participation, to see how every added year affects the odds of a woman to be in paid work considering also their educational level.
Region (within Turkey)
Since it is expected that the results for Istanbul, because of its more liberal structure, will be different, it is used (as it is also taken by ESS data) as a separate category (reference). For the purpose of a clear analysis; variables of region are coded again as South (recoded from Mediterranean part), North (recoded from Western Black Sea, Eastern Black Sea), West (recoded from Eastern Marmara, Western Marmara, Western Anatolia, Aegean Region) and East (recoded from North Eastern Anatolia, East, South East).
Women should be prepared to cut down on paid work for the sake of family
This attitude question is used to see how liberal women with different educational levels are, and how liberal-conservative attitudes regarding gender roles affect women’s likelihood to be in paid work. Additionally, this variable is expected to test the hypothesis that women’s life plans are more domestic and that this is one of the reasons they’re not working after graduating from a higher education institution.
Husband’s education
Husband’s education is taken into account to see whether it has an effect on the decisions of highly educated women to work or not. This is coded in the same way as women’s educational level (see above).
Children at home
Having children at home is used as a binary variable to indicate whether someone has a child living at home aged 12 or under or not.
Methods
Methods will include firstly cross tabulations to see the basic bivariate relationships between each of the major variables. Secondly, logistic regressions will be presented taking the odds of being in paid work rather than housework as the dependent variable. Logistic regression is a relevant method for this study, because the dependent variable is binary and because it is serving the aim of the research, that the main purpose is to see the reasons (effects) of different determinants (variables) on women’s labour market participation decisions linked to their education.
Hypotheses and research questions
The main question of the research is why women are going to college and yet there is still the phenomenon of “college graduated housewives”; in other words, why there is still not a satisfying compatibility between high education and labour market participation for Turkish women. Two main hypotheses will be tested to identify the underlying reasons for this phenomenon,
Hypothesis one denotes that Turkish women tend to see higher education as a progressive step for their domestic roles and go to college for the sake of having a degree rather than aiming a career. This hypothesis will be tested by modelling how the relationship between women’s education and work status are affected by marriage, the presence of dependent children, and liberal-conservative gender role attitudes.
Hypothesis two denotes that there is high educational homogamy in Turkey and one of the
motivations for women to go to college is to become a more competitive candidate for marriage, and to have a better chance of having a highly educated spouse (thus a spouse with more cultural capital and a better provider for the future family), rather than particularly having the aim of obtaining a job as an exclusive motivation. This hypothesis will be tested by exploring the extent of educational homogamy in Turkey and modelling how the relationship between women’s education and work status are affected by husband’s education.
Findings
To start with the descriptive statistical results, as seen in Table 3, the percentage of working women gets higher as the educational level increases. However, more than 10% of highly educated women report their main activity as “housework”, showing that there is still a significant rate of women, who are housewives, despite completing a higher education institution.
As Table 4 shows, high and medium educated women are less likely to be married than low educated women. This could be a result of age differences because low educated women are expected to get married earlier as highly educated women spend more time at school and get married at an older age.
Table 3. Main activity of women of working age considering highest educational level
Main activity within last 7 days Highest level
of Education
Paid
work Housework Unemployed Other Total
Low 4.2 85.9 0.9 8.9 100.0
Medium 22.9 47.4 5.7 24.2 100.0
High 61.4 11.4 9.1 18.2 100.0
Total 12.1 71.9 2.6 13.3 100.0
Table 4: Marital status of women in working age considering highest educational
level Marital Status Highest level of Education Married No longer
married Never married Total
Low 83.5 9.1 7.4 100.0
Medium 56.7 6.7 36.6 100.0
High 63.6 4.6 31.8 100.0
Total
75.6 8.2 16.2 100.0
Turning to the issue of educational homogamy, as seen from table 5, educational homogamy patterns for Turkish women are quite strong, particularly for low and highly educated women.
Table 5: Husband’s educational level considering (married women’s) highest
educational level
Husband’s educational level Highest
level of Education
Low Medium High Total
Low 73.0 23.3 3.7 100.0 Medium 21.3 57.4 21.3 100.0 High 3.6 21.4 75.0 100.0 Total 59.8 29.7 10.5 100.0
Table 6 (logistic regression results) presents the log odd of being in paid work rather than
in housework (dependent variable) by using women’s education as the main variable with other relevant variables.
Table 6: Log odds of women being in paid work versus housework, main effects (standard error in brackets)
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7
Educational Level (low education = ref)
Medium Education 2.15*** (.30) 2.03*** (.31) 2.05*** (.31) 2.01*** (.31) 2.52*** (.36) 1.88*** (.31) 2.18*** (.41) High Education 4.66*** (.54) 4.77*** (.56) 4.44*** (.54) 4.48*** (.55) 5.63*** (.74) 4.67*** (.61) 5.64*** (.86) Age .08 (.11) .36*** (.14) .21* (.13) .10 (.12) -30 ** (.13) .06 (.12) .40 ** (.16) Age squared -.002 (.00) -.01 (.00)*** -.003 (.00)* -.002 (.00) -.004** (.002) -.002 (.00) -.01*** (.00)
Marital status (married = ref)
Never married women 2.06*** (.42) 1.77*** (.71) No longer married women 1.41*** (.48) .76 (.57) Children at home -.82 (.32) .07 (.52)
Women to cut down on paid work for the sake of family (ref = disagree)
Disagree 1.05*** (.31) .83** (.37) Neither Agree Nor Disagree .02 (.50) .04 (.54)
Husband’s education (low = ref) Medium Education -1.25** * (.46) -1.36** * (.48) High Education -1.28** (.66) -1.44** (.71) Husband’s education N/A 1.07*** (.38)
Region (Istanbul = ref)
Central Anatolia Dropped
East -1.32** *(.47) -1.51** * (.55) West -.50 (.35) -.13 (.41) North .65 (.52) 1.00* (.58) South -1.11** (.51) -1.04* (.57) constant -3.80* *(1.88) -9.28*** (2.41) -5.48** * (2.03) -4.34** (2.00) -7.93** *(2.24) -2.28 (2.01) -9.26** * (2.80) G(df) 158.93 (4) 188.90 (6) 161.66 (5) 167.82 (6) 195.03 (7) 174.54 (8) 216.80 (15)
N=1029(only women) *p value 0.1 **p value 0.05 *** p value 0.01
To start with the logistic regression results, in model 1, the effect of women’s educational level on the odds of being in paid work is investigated by controlling for age and age squared. The model reveals the fact that the more educated women are statistically significantly more likely to be in paid work with log odds of 2.15 and 4.66 for medium and highly educated women compared to the reference category of low educated women. Age and age squared have a positive and negative effect
respectively (although it is non-significant in this model), suggesting a diminishing positive effect of age on the probability of women to work.
In model 2, variables of marital status categories are added to see how being in different marital status affects the probability of being in labour force for women of working age. The model shows that compared to married women, the log odds of never married women to be in paid work is 2.06 (exp 7.8) greater, as for no longer married women it is 1.41 (exp 4.1) greater. Adding the marital status variable slightly increases the coefficient for high education and slightly decreases the coefficient for medium education.
In model 3, the effect of having children at home is included. Having children at home decreases the probability of being in paid work by -.82 (exp .44). Adding this variable decreases the effect of education on women’s log odds to work, compared to model one.
Model 4 adds the gender role attitude variable (whether the respondent agrees that women should cut down on paid work for the sake of the family), which is a dominant determiner for women’s
probability to work, even when they attain higher education. The results of the logistic regressions of this study show that women who are less traditional towards gender roles (i.e. disagreeing with women to cut down on work for the sake of family) have a log odds of working that is 1.05 (exp. 2.86) times greater compared to women with more traditional attitudes. In addition, adding this variable decreases the effect of education on women’s log odds to work compared to model one.
In model 5, the categories of husband’s education are added. Husbands with medium education have a negative effect on women’s likelihood to be in paid work, as do husbands with high education (although this last coefficient is not significant at the 0.05 level). Adding husband’s education to the model increases the effect of women’s education on the likelihood of being in paid work.
In model 6, regions of Turkey are included to see how the effect of being from different regions, and accordingly from different backgrounds affects women’s likelihood to be in labour force. Compared to Istanbul, women from East and South parts of the Turkey are -1.32 and -1.1 log odds (exp. 3.74 and exp 3.00) less likely to work. The coefficients for the other regions are not statistically significant. In model 7, the full model of variables is presented.
Table 7 (interaction terms) presents the logistic regression analysis results, in which the log odd of
being in paid work rather than in housework is the dependent variable and women’s education is interacted with the other independent variables.
Table 7: Log odds of women being in paid work versus housework, interaction effects (standard error in brackets)
Model 1 Model 2 Model 3 Model 4 Model 5
Educational level (low = ref)
Medium Education 2.86** (1.21) 1.41*** (.37) 2.90*** (.60) 1.25*** (.38) 1.90*** (.61) High education 6.63*** (2.34) 4.36*** (.57) 4.90*** (.92) 3.61*** (.65) 2.65*** (.93) Age .11 (.13) .41*** (.15) .21* (.13) 13 (.12) .37** (.18)
Age Square -.002 (.002) -.006*** (.002) -.003* (.002) -.002 (.002) -.005** (.002)
Marital Status (married = ref)
Never married .75 (.81) No longer married .40 (.78) Children at home -.06 (.60)
Women to cut down on work for
The sake of family (agree = ref)
Disagree -1.05
(1.04)
Neither/Nor Dropped
Husband’s education (low = ref) Medium education -1.65* (.1.04) High education 1.22* (.82)
Interactions with women’s education Medium education X Age -.02 (.04) High education X Age -.06 (.06) Med.education X Never married 2.12 (.92)** Med.education X No longer married 2.27 (1.17)** High education X Never married Dropped High education X No longer married Dropped Med.education X Children at home -1.22* (.71) High education X Children at home -.56 (1.15) Med.education X Disagree 2.82*** (1.12) Medium education X Neither/Nor Dropped High education X Disagree 2.87* (1.61) High education X Neither/Nor Dropped Med.education X Medium edu.(hus) .88 (1.23)
Med.education X High edu.(hus) -2.46** (1.23) High education X Medium edu.(hus) Dropped High education X High edu.(hus) Dropped Constant -4.48** (2.31) -9.80*** (2.61) -6.11*** (2.11) -4.67** (2.11) -9.32*** (3.26) G(df) 159.90(6) 167.50(8) 165.19(7) 163.83(9) 76.31(8)
N=1029(women only) *p value 0.1 **p value 0.05 *** p value 0.01
In the first model, the variables of women’s educational level are interacted with age. For medium educated and highly educated women, the effect of age on the probability to work is not statistically significantly different compared to the effect of age for low educated women.
The second interaction model presents the effect of education given marital status. As the results show, marital status makes no significant difference to the log odds of working for low educated women; however there is a strong positive effect of being never married rather than married for medium educated women. (The interaction of highly educated women with never married and no longer married categories are dropped because of the small sampling in these particular
combinations.)
In the third model, the interaction of education and having children at home is presented. For medium educated women having children at home has a statistically significantly negative effect. However, for highly educated women and low educated women (the reference category) having children at home has no statistically significant effect on the log odds of being in paid work.
In the fourth interaction model the gender role attitudes of women are interacted with educational level. According to the results, having more liberal attitudes has no significant effect on the likelihood of working for low educated; however, the effect is strongly positive and significant for highly educated women.
The fifth model presents the interaction terms of women’s and husband’s educational level. For both low and medium educated women, having a medium rather than low educated husband makes them less likely to work. For low educated women, however, having a highly educated husband makes them more likely to work. This result is surprising, although it should be remembered from the cross tabulations that low educated women are rarely married to highly educated men. More consistent with expectations, medium educated women are less likely to work if their husband is highly educated.
Discussions
The results of the cross tabulations and logistic regressions suggest that rising educational attainment for Turkish women makes it more likely for them to be in paid work. However, there are other variables affecting women’s probability to be in labour force, some of which weaken the positive effect of women’s education. The conclusions section will be organized according to the two hypotheses of the research. In the first part, the results will be evaluated considering the effect of marital status and educational level of women. The second part will interpret the educational homogamy patterns and effect of husband’s education on the probability of a woman to work.
*The effect of being married and traditional domestic attitudes attached to married women considering different educational levels
Logistic regression results indicate that married women are the least likely to work compared to women of non-married status. It is also seen that equal percentages of highly educated women agree and disagree with the attitude that women should cut down on paid work for the sake of family, is at 36.36%. Since disagreeing with this statement is linked to lower odds of paid work, it could be
argued that highly educated women have a relatively low odds of working because they are not dramatically less traditional compared to less educated women in Turkey.
To make more detailed interpretation of the results for the women’s education and marital status effect on the probability to work, logistic regression results will be discussed. Logistic regression results also suggest the increasing probability with rising educational level for women to be in paid work. Also, never married women and no longer married women are statistically significantly more likely to be in the labour market than married women. This could be interpreted due to the fact that married women are more attached to their domestic responsibilities. Having children at home also has a negative effect for women to work, which is another domestic burden preventing women to be in the labour market. For Turkish women, traditional gender role labels decrease the effect of educational level as seen from the interaction terms. In other words, the results suggest an overshadowing effect of traditional labels such as being married, having children on the probability to be in paid work. As being more educated means being more likely to be in paid work, this effect gets pulled back, even if not cancelled, when a woman is married. Also, although there is no significant effect for low educated women to have children at home, highly educated women’s probability to be in paid work is negatively effected on a significant level by this variable. In addition, having more traditional attitudes towards women’s work life does not have a dramatic effect for low educated women, but significantly affects the odds of working for better educated women.
Considering the offset effect of traditional labels on the odds of a woman to work, the results of the main effects and interaction effects show that the positive effect of women’s education might only, or at least primarily, be for the single and relatively younger women of working age. Since these women may be expected to be married in future, this may cause the currently positive effect of higher education to weaken with the effect of being married, particularly if this is combined with children at home.
*Educational homogamy and the effect of husband’s educational level
This part of the discussion will evaluate the effects of educational homogamy and thus the effect of husband’s education on women’s probability to be in paid work. As cross tabulation results suggest, among all educational levels, there is a high degree of educational homogamy in Turkey. For highly educated women the rate of marrying highly educated men is 75 %, whereas for low educated women, prospectively because of the “time out of school” factor (Blossfeld and Timm, 2003), it is very unlikely with a level of 3.69%. This might support the contention that high education is a means, which almost guarantees a higher status spouse and a better provider for the future family.
Considering the logistic regression analysis, women with better educated spouses are less likely to be in paid work. One possible explanation for this is that as low educated men are expected to be low profile providers, their wives have the need to make an additional financial contribution to their families. Another (and parallel) possible explanation is that highly educated husbands are better providers to their families, which makes it easier for a woman to choose her domestic responsibilities and to abstain the double burden of being a working woman and a homemaker. Interaction results also show that the negative effect of husband’s education holds only for better educated women. In other words, because of high educational homogamy, some of the increase in the rate of paid work due to being a highly educated woman is soon offset to an extent by marrying a highly educated man. It is important to note that this hypothesis and the results are not implying that the primary aim of women for attaining higher education is precisely to have a better spouse. The aim is to emphasize that higher education makes women more eligible spouses and as one of the motivations for higher education is related to fulfilling the domestic requirements of society; a decent marriage and being a better spouse are part of these requirements. Moreover, as also stated before; a better spouse is a means for cultural capital for the future family, too, -as cultural capital for oneself can be obtained through education-, to increase the level of it for one’s future family can be obtained better through having an equally educated and culturally qualified spouse. This hypothesis could also be combined with the fact that due to the possibility that women might consider to cut down on paid work as a result of economic or domestic oriented reasons, a good provider for the family could be seen as a necessity by them. The above stated factors are reflected in the statistical results and noted in the literature by showing a trend of high educational homogamy for especially highly educated women in Turkey and it could be added that as education is “a way” of meeting spouses from the same level of education; it is also a “result” of being in the same place with people from similar backgrounds, also related to the same “time out of school” (Blossfeld and Timm, 2003).
Conclusion
This research has investigated the increasing levels of educational attainment and yet stable and even almost decreasing level of occupational participation on the part of Turkish women. Turkey is an interesting context for this research because the patterns in educational and occupational trends for Turkish women are dramatically contradictory and there are a lot of internal and external determinants for a woman when she is making a decision whether to be a part of the labour market or not. This study explored the phenomenon from two different dimensions: a) the domestic labels attached to married women and the effect of marital status on the probability to work; b) high educational homogamy and the effect of husband’s educational level on women’s working status. This
phenomenon is worth to be studied from other aspects such as regional differences, from economic aspects mainly unemployment or in more qualitative ways, in which women’s attitudes towards being a working woman are measured. However, this study focused primarily on the relationship between women’s and husbands’ education and their effect on women’s probability to work.
Results showed, while education has an increasing effect on the likelihood to work for Turkish women, the negative effect of being married partly offsets the positive effect of education. Also having children at home decreases the likelihood to be in paid work as it is a crucial domestic responsibility for Turkish women, which they generally have to carry out without any professional support because of the lack of facilities provided to working women with young children in the work place. Furthermore, when men are better providers, mostly accompanied by their higher educational level, women are more likely to choose their domestic responsibilities over paid work. Considering the fact that most highly educated women in Turkey are married to highly educated men, one of the main reasons for the incompatibility between increasing level of education (which results in meeting people with the same educational level) and decreasing occupational participation of Turkish women is explained by having spouses who are better providers, which in turn has a negative effect on the probability of women to be in paid work. As also cross tabulation results suggest, highly educated women are not particularly liberal in their attitudes towards women to take their work life as priority, which makes it easier for them to choose their domestic responsibilities over their roles in labour force.
The trend of a non-increasing rate of women working despite being university educated is an unfortunate picture for Turkey. However, the outlook may be hopeful and things may become more progressive in future. The latest Social Security Institution Law, which is in effect since October 2008, is enabling working women with children, who have insurance, to retire 2 years earlier per child, for a maximum of 2 children. This recent development supporting women’s working life bodes well for Turkish women’s future propensities to work. Turkey also does not have the worst gender inequality trends compared to other countries. This could be turned into an advantage together with more supportive government policies upholding women’s work life, work places providing women with younger children more acceptable facilities such as reasonable maternal leave time, child care facilities in the work place for women with babies/young children and changing unequal gender perceptions in society. Both for women to be in paid work and have their economical freedom; especially for those with higher education to contribute to the economy of the country in return of the facilities provided to them for their education and the efforts they have been given for their
educational and personal improvement; Turkish women need to be supported more by social, cultural and economic regulating factors. As the current common mentality is that a woman’s work life is incompatible with her domestic responsibilities and family life, in future, this could be changed by equalizing the division of labour in the private sphere as well as the public sphere.
Bibliography
1) Acar F., 1994. “Women and higher education in Turkey: A gold bracelet for women” In Suzanne Lie, Lynda Malik & Duncan Haris (Eds.) , The gender gap in higher education , pp 160-170. London: Kogan page.
2) Başlevent, C., Onaran, O., 2003. “Are Married Women in Turkey More Likely To Become Added Or Discouraged Workers?”. Labour 17, (3) pp. 439-458, Fondazione Giacamo Brodolini and Blackwell Publishing.
3) Becker, Gary S. Human Capital: A Theoretical and Empirical Analysis, with Special Reference to Education. 2nd ed. New York: Columbia University Press for NBER, 1975.; also available at: http://www.econlib.org/library/Enc/HumanCapital.html, accessed in March, 2009
4) Bener, Ö., 2008. “Women’s Opinions About The Necessity Of Women’s Education” Available
online: (Hacettepe Sosyolojik Araştırmalar e-dergisi)
http://64.233.183.104/search?q=cache:qLlPLoffTL8J:www.sdergi.hacettepe.edu.tr/bener.pdf+%C3 %B6zg%C3%BCn+bener+about+womens&hl=tr&ct=clnk&cd=2, accessed in September, 2008 5) Blossfeld, H.P., Timm, A. (Eds),2003. “Who Marries Whom? Educational Systems as Marriage
Markets in Modern Societies”, Dordrecht: Kluwer Academic Publishers.
6) Boyd, E. R., 2002. ” ‘Being There’: Mothers Who Stay At Home, Gender and Time” Women’s Studies International Forum, vol 25, no. 4, pp. 463-470
7) Bradley, K., 2000. “The Incorporation of Women into Higher Education: Paradoxical Outcomes?”. Sociology of Education, vol. 73 n1 pp 1-18.
8) DiMaggio, P., Mohr, J., 1985. “Cultural Capital, Educational Attainment, and Marital Selection”. The American Journal of Sociology, Vol. 90, No. 6, pp. 1231-1261. American Sociological Association
9) England, P., Li, S. 2006. “Desegregation Stalled: The Changing Gender Composition of College Majors 1971 -2002”. Gender & Society, Vol. 20, No. 5, 657-677, Sage Publication.
10) ESS (European Social Survey) / Second Round (2004/2005), for more information on ESS Second Round Data:
http://ess.nsd.uib.no/index.jsp?year=2005&module=main&country=
(access with registration), accessed in July, 2008. Also more information on European Social Survey is available online at: http://ess.nsd.uib.no/, accessed in July, 2008.
11) Gershuny, J., 2000. “Changing Times: Work and Leisure in Post-industrial Societies”. Oxford: Oxford University Press.
12) Gündüz-Hoşgör, A., Smits, J., (2008). „Variation in Labour Market Participation of Married Women in Turkey”. Women's Studies International Forum, 31, 104-117.
13) Household Labour Force Survey for the Period of November (October, November and December 2007). Statistical, Economic and Social Research Training Centre for Islamic Countries (SESRIC), Ankara Centre. (Information obtained from Turkish Institute of Statistics)Available online: http://www.sesrtcic.org/files/nso/320.pdf, accessed in July, 2008.
14) Hunt, T. C., 1940. “Occupational Status and Marriage Selection”. American Sociological Review, Vol. 5, No. 4, pp. 495-504. American Sociological Association
15) Jonsson, J.O., 1999. “Explaining Sex Differences in Educational Choice: An empirical assessment of a rational choice model”, in European Sociological Review, 15(4): 391-404
16) Kalmijn, M., 1998. “Intermarriage and Homogamy: Causes, Patterns, Trends”, Annual Review of Sociology, Vol. 24: 395-421.
17) Kızılırmak, B., 2005. “Labour market participation decisions of married women. evidence from Turkey”. Available online: http://www.politics.ankara.edu.tr/dosyalar/burca.pdf, accessed in July, 2008.
18) Mare, R., 1991. “Five decades of Educational Assortative Mating”. American Sociological Review, Vol. 56, No. 1 pp. 15-32. American Sociological Association.
19) OECD Economics Department, 2004. Female Labour Force Participation: Past Trends and Main
Determinants in OECD countries pp. 2, Figure 1, available online:
http://www.oecd.org/dataoecd/25/5/31743836.pdf, accessed in August, 2008.
20) OECD Labour Force Statistics 1985-2005: 2006 Edition: Employment , no. 6, pp. 1, 449. (Turkey pp. 334)
21) OECD Statistics, Education at a glance, Table A10.1a. Employment ratios and educational attainment (2002) Number of 25 to 64-year-olds in employment as a percentage of the population aged 25 to 64, by level of education attained and gender.
22) Özer, M., Biçerli, M.K., 2003. “Türkiye’de Kadın İşgücünün Panel Veri Analizi”, Anadolu Universitesi Sosyal Bilimler Dergisi, 3, 1.
23) Reskin, B. F., Hartmann, H. I.(eds.), 1986. “Women’s Work, Men’s Work: Sex Segregation on the Job”. Committee on Women's Employment and Related Social Issues, National Research Council. 24) Rubin, S.E., Wooten, H.R., 2007. “Highly educated stay at home mothers: a study of commitment
and conflict”. The Family Journal, 15, 336.
25) Scott, J., Dex, S., Joshi, H., 2008. “Women and Employment: Changing Lives and New Challenges”. Edward Elgar Publishing.
26) Sebastian, A., Navaneetham, K., 2005. “Gender, Education and Work: Determinants of Women's
Employment in Kerala”. Available online:
http://www.isical.ac.in/~wemp/Papers/PaperAliceSebastianAndKNavanitham.doc , accessed in July, 2008 (Seminar on Gender Issues and Empowerment of Women Platinum Jubilee of Indian Statistical Institute, 1-2 February 2007).
27) Smits, J., Ultee, W., Lammers, J., 1998. “Educational Homogamy in 65 Countries: An Explanation of Differences in Openness Using Country-Level Explanatory Variables”. American Sociological Review, Vol. 63, No. 2, pp. 264-285. American Sociological Association.
28) TISK Raporu, (The Turkish Confederation of Employer Associations Report): OECD Ulkelerinde Kadinlar ve Erkekler (Women and Men in OECD Countries), Is gucu Piyasasi (Labour Force
Market) 2008. Table 2. no 289. Available online:
http://www.keig.org/raporlar/T%C4%B0SK%20RAPORU.doc, accessed in August, 2008.
29) Turkish Statistical Institute (ed.): 1923-2004 Statistical Indicators, 2005. (Education and Culture) pp. 65 Available online: http://www.die.gov.tr/ENGLISH/kit_ing_1.pdf, accessed in August, 2008. 30) Turkish Statistical Institute (ed.): 1923-2004 Statistical Indicators, 2005. (Labour) pp. 153 Available
online: http://www.die.gov.tr/ENGLISH/kit_ing_1.pdf, accessed in August, 2008.
31) Turkish Statistical Institute, Labour Force Statistics, Periodic Results of Household Labour Force Survey (Turkey, Urban, Rural), Results of 1988-1999 Terms/ Turkey, Table 4: Reasons of Not Being In Labour Force By Years And Sex ; Results of 2000 - October 2007 / Turkey, Table 4: Reasons of Not Being In Labour Force By Years And Sex.
32) Turner, S.E., Bowen, W.G., 1999. “Choice of major: the changing (unchanging) gender gap”. Labour Relations Review, vol. 52, no. 2, pp 289-313.
33) Women and Men in OECD Countries, 2006. Percentage of men and women aged 25-64 with tertiary education 2004 or latest year available, pp 12. OECD publications. Available online: http://www.oecd.org/dataoecd/44/52/37962502.pdf, accessed in July, 2008.
