Pearson’s correlation coefficient is relevant for this project as it informs the researcher about the ‘strength and direction of the correlation’ (Burnham et al. 2004: 163). Once the value has been obtained for each IV, including the intervening variables, I will be able to chart the factors which have the highest effect on the percentage of women in parliament and which can be disregarded. More sophisticated analysis can then occur in form of multiple regression.
This is a suitable tool of analysis for the following reasons. Firstly, it charts the depth of the relationship between a dependent variable (number of women seats) and a variety of independent variables. Therefore, the independent variables are assessed as a group and not individually therefore ‘maximising the levels of explained variance, whilst simultaneously including only terms which are statistically significant’ (Castles 1998: 19). It shows which of the IV are strongly influential and which have little effect on the DV. It is also useful as it shows if there is any overlap between the IV’s and if they interact in any way.
However, multiple regression is only successful in estimating the relationship between the dependent and independent variable if the relationship is linear, and initial observations have shown my variables’ relationship to be linear in nature. Also, although multiple regression allows for exploration of the effect of more than one independent variable, it is also quite awkward to use. The best models provide only 3 or 4 independent variables. I intend to decide which IV to use once I have gathered my data in order to create the most significant and contributory model.
Essentially, multiple regression is an extended version of single linear regression that means prediction or estimation of one variable from others is possible, although the reliability of this figure is dubious when outside the range of original data. Therefore, when a regression line is drawn, it is possible to use this information to predict the value of the dependent variable using the model created whilst also allowing calculation of the total variance as a result of the combination of variables.
I believe that my two-step approach to data analysis will prove very effective in focusing the study and will help me to make more sophisticated causal inferences. I will ensure that my research reaches these organisations by posting and e-mailing the project to them with a covering letter briefly outlining my initial aims. Implications of this research are wide reaching and it is my belief that the findings could influence future policy. The UK, although not the worst offender in levels of female representation, could learn from its Nordic neighbours and, if it is shown that certain increases in education or childcare do affect the levels of female seats in parliament, steps could easily be taken to change the situation in this country.
My approach to this research is wholly quantitative-centred and, despite there being a large body of work on the more qualitative aspects of female representation, citing inherent psychological and behavioural differences between the sexes, I believe there to be more need for a re-examination of the social and economic factors. However, my research may result in my research question being rendered completely void and irrelevant.
My methods are simple, time-effective and parsimonious. They involve no surveying or fieldwork and rely solely on data collected by professional bodies and organisations. I consider the design of this research project to be effective and potentially successful in the reaching of conclusions that will have a positive effect on the circumstances regarding the level of women in parliament.