19.01.2013 Views

Parties, Candidates and Citizens On-Line - Åbo Akademi

Parties, Candidates and Citizens On-Line - Åbo Akademi

Parties, Candidates and Citizens On-Line - Åbo Akademi

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Because the study is case-oriented with a low number of cases, the choice of<br />

suitable research methods is limited. Nonetheless, the design of the study is variable<br />

oriented <strong>and</strong> would require a research method which facilitates some form of variable<br />

testing. Given these restrictions, the method of Qualitative Comparative Analysis<br />

(henceforth abbreviated as QCA) is an appropriate solution. The method has the<br />

advantage of being able to retain the “wealth of case-oriented material <strong>and</strong> case-derived<br />

‘data’ in political science <strong>and</strong> convert them into a form suitable for variable based<br />

research” (Peters 1998: 163; Ragin 1994: 304). Peters (1998: 160-2) merits the technique<br />

as one of the tools for including qualitatively-derived data in meta-analyses.<br />

The QCA employs the technique of Boolean algebra which is basically a form of<br />

logical reasoning (for a detailed description of the technique, see Appendix A). The<br />

procedure follows a logic of ‘differences can’t explain similarities’ in order to find<br />

multiple, logical combinations of causal variables in which the dependent variable occurs<br />

(Coverdill & Finlay 1995: 458; Ragin 1994: 312). Each case is considered holistically; the<br />

effect of one independent variable is seen as different in each case depending on the<br />

values on the other independent variables (Miethe & Drass 1999: 9-10).<br />

The Boolean technique is not without its problems. Firstly, the method is<br />

dependent on theoretical reasoning <strong>and</strong> argumentation in determining which causal<br />

variables to include in the analysis (Peters 1998: 168). Secondly, the requirement of binary<br />

data is problematic as societal phenomena are often quite complex <strong>and</strong> difficult to<br />

simplify into a yes-no dichotomy (Jungerstam-Mulders 2003: 70-1; Peters 1998: 169-70).<br />

The third problem with the Boolean technique is the limited availability of data in existing<br />

cases. The outcomes of all possible combinations of variables are usually never tested<br />

(Jungerstam-Mulders 2003: 71). Finally, it is not uncommon to encounter contradicting<br />

cases where the same combination of values on the independent variables produces<br />

different outcomes. In part, this could be due to the selected variables not being able to<br />

discriminate between the cases. This problem is hard to overcome with QCA as the<br />

inclusion of too many variables in the analysis is difficult (ibid). Still, Lijphart (1971: 686)<br />

argues that deviant cases do weaken a hypothesis but are not sufficient to fully invalidate<br />

it unless they appear in frequent numbers. Therefore, a QCA may still be conducted<br />

including the settings which produce contradicting outcomes if one is aware of this<br />

throughout the process (Jungerstam-Mulders 2003: 82).<br />

Operationalizations<br />

The dichotomizations of the variables, both independent <strong>and</strong> dependent, are discussed in<br />

this section. As a rule of thumb, the conditions likelier to stimulate findings of<br />

42

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!