• Keine Ergebnisse gefunden

As the literature suggests, there are many factors that influence the status of women in a given country and there is an ongoing lively discussion on both the causal effect of human rights treaties in general and a multitude of other factors impacting women’s human rights situation (see Landman, 2003: Ch. 9).

Many scholars reason that human rights treaties might not have a positive effect at all, and that on the contrary, their ratification could lead to a deteriorating human rights sit-uation (Keith, 1999; Hathaway, 2002; Vreeland, 2008, Hafner-Burton and Tsutsui, 2007).

As these studies concentrate on government repression of physical integrity rights and not on the discrimination of women in particular, I do not expect the causality to be the same.

At the domestic level, it has been argued that democratically accountable institutions ex-plain the actual respect of human rights, irrespective of whether a state has signed a treaty

or not. Also, the general level of development is likely to have an effect on the standard of women’s rights, as especially many of the socio-economic rights are closely linked to the capacity a state has to provide basic necessities such as health care and education (Hunt, 1996; Steiner et al., 2008).

In previous literature, a state’s level of international interdependence is identified as an important factor influencing women’s general status in society. States that are highly integrated into the international community, with strong trade relations (Neumayer and De Soysa, 2007), a multitude of memberships in international governmental organisations (IGOs) and many international non-governmental organizations (INGOs) working within their boundaries will have higher costs to bear if they fail to keep to the agreed hu-man rights agreements (Hafner-Burton, 2005; Landhu-man, 2005; Neumayer, 2005). I shall take into account the most frequently mentioned alternative explanations in my statistical model.

4 Research Design

4.1 Data and Operationalization

4.1.1 Case Selection

To test my hypotheses, I have opted for a pooled time-series-cross-section design to account for the variation in compliance over both time and space, with the observation unit being country-year. My sample of countries consists of 152 countries for which data was available on both their status of ratification of CEDAW and their respect for women’s human rights10. The time period covers 20 years from 1981 (when the Convention came into force) to 2000.

4.1.2 Dependent Variables

The dependent variables that provide estimates for women’s status are women’s political (WOPOL), social (WOSOC) and economic (WECON) rights, taken from the Cingranelli and Richards (2008) Human Rights Database. WOPOL, WOSOC and WECON are an-nual cross-national observations11. Using the annual United States (US) Departments Country Reports on Human Rights Practices and Amnesty International (AI) Annual Re-ports12, discrete variables are coded that take on one of four ordinal values that document different levels of rights that are granted to women:

‘0’ for No rights: There are none of the given rights for women in law and systematic discrimination based on sex may be built into the law.

‘1’for De jure rights: Women have some of the given rights under law, but these rights are not effectively enforced.

‘2’ for De facto rights: Women have some of the given rights under law, and the gov-ernment effectively enforces these rights in practice while still allowing a low level of discrimination against women in certain matters.

‘3’for Full Respect: Nearly all of the given women’s human rights are guaranteed by law and the government fully and vigorously enforces these laws in practice.

10Especially concerning the dependent variables, the missing observations are most likely not randomly distributed, as the assessment of human rights is usually not possible in times of unrest, political instability or during total loss of government authority. Following previous research, these periods are, however, usually highly correlated with increased human rights abuses and repression (see Poe et al., 2006; Zanger, 2000). I thus expect the average level of documented respect for womens rights to be slightly higher than the actual value would be. Additionally, countries with a population of less than 500.000 were also excluded from this study (see Gleditsch and Ward, 1999).

11See APPENDIX for full list of countries as well as list of rights included in the three independent variables.

12The fact that the indicators are mixtures of these two different reports makes them more reliable, since they are both said to have slight biases. Whereas US reports tend to have a slight positive bias towards their allies in foreign policy and assigned tougher measures to leftist regimes, AI is said to be slightly biased towards states that disrespect human rights. Both arguments are rooted in the reports’ original purposes: whereas AI started out in documenting human rights abuses in the most repressive countries, the US reports were compiled so as to decide on the scope or denial of US foreign aid to be allocated (see Poe and Tate, 1994)

4.1.3 Explanatory Variables

Legal Commitment to CEDAW:The main explanatory variable is the ratification of CEDAW, provided by Landman (2005)13. The dichotomous variable is assigned 0 in every year a state has not ratified the Convention and 1 for every year it has ratified it14.

There have been theoretical discussions on whether the actual year of ratification is in fact the ‘turning point’ when governments commence with the full respect of the treatys articles. Some have argued that states, as rational and anticipating actors will adjust their level of compliance prior to the accession (see Neumayer, 2005) while others have employed measures of the years passed since ratification, assuming that with time, the level of compliance will increase in a linear way (e.g. Hathaway, 2002). My statistical model does not intend to detect the improvement of government respect towards women’s rights, as I am not trying to explain this phenomenon. My object of enquiry is the level of compliance that states display after signing the Convention, and whether this level varies across the categories of women’s rights.

Democracy: The configuration of domestic institutions is expected to have an affect on the audience costs and thus on the level of compliance a state will display towards women’s human rights. I have therefore opted for the POLITY2 variable on the degree of procedural democracy a country displays, taken from the Marshall and Jaggers (2006) Polity IV Dataset15. This indicator defines a country’s regime type on a dimension from

‘strongly democratic’ to ‘strongly autocratic’ focussing entirely on the institutional and procedural dimension.

Influence of Islamic Law and Tradition: Islamic law has proven to have a strong effect on the level of basic human rights granted to women. For example, Islamic family law assigns women a different status than men within a marriage and in many Muslim societies women’s economic value and position is not regarded as being equal to that of men’s (Neumayer and De Soysa, 2007: 1522). The influence of Islamic law is operational-ized through the variable MUSLIM, which captures the percentage of Muslims among the population, with data provided by La Porta et al. (1999).

13The variable is available at URL: http://privatewww.essex.ac.uk/ todd/interests.htm.

14Landman originally distinguishes three categories: states that have ratified, those that have only signed and those that have not signed the treaty. However, states are only legally obliged to incorporate treaty law into their domestic system, once they have ratified a Convention. For the purpose of this study it is thus only of interest whether a state has ratified or not ratified (i.e. not signed or only signed) a treaty. For this reason I recoded the original dataset to fit this dichotomous character.

15Wherever the original Polity dataset entailed a missing value, I replaced it with the value provided in Gledtisch’s (2007) alternate version of the Polity IV that contains additional estimates that are not in the original Polity IV data. The variable was also recoded to fit a continuous scale from ‘0’ to ‘20’, with the level of democracy rising with the value of the indicator.

Influence of Catholic Customs and Tradition: The influence of Catholic tradition is operationalized by the variable CATHO that captures the percentage of Catholics among the population, with data provided by La Porta et al. (1999).

4.1.4 Control Variables

Development:16 The level of economic development is included in order to control for varying levels of women’s rights due to different levels of overall wealth a country displays.

The indicator used measures the per capita gross domestic product (GDP) in 1995 US dollars and is logged so as to account for problems of skewness (LNPCGDP), which would lead to statistical bias (see Kohler and Kreuter, 2005: 226-7).

Globalization: One of the indicators for international interdependence is the extent to which a country is integrated in the global economy. To account for this phenomenon, the logged amount of trade as percentage of the total GDP is used (LNTRADE).

IGOs: The indicator for international governmental organizations IGO is constructed using the data of Pevehouse et al. (2004). For every year, the variable captures the total number of a countrys IGO memberships.

INGOs: The variable indicating the involvement of INGOs in a state draws from the Union of International Associations (UIA) and reflects the number of INGOs that have a registered office in the relevant country. To account for the distributional skewness, the logged value is employed in the statistical model (LNINGOS).

Time trend: To control for possible time trends, I add a variable counting the years since 1981 to my model.