Choosing relevant variables

Scholars concerned with such(big) wars almost never examine events that threatened to become big wars but did not escalate beyond low levels of dispute. There is surprisingly little research about wars with dramatic consequences that were not themselves big events.

We cannot understand the causes of big wars without examining many crisis that had the potential to become big wars but were averted by reaching a negotiated settlement beforehand......... there is little historical research where war did not take place, while the threat was imminent, fear of just such a war prompted parties involved to grant concessions that might otherwise not have been given....ln such examples the selection effect is of a different type. Researchers are sampling on the dependent variable, the thing to be explained, and so are unwittingly biasing their inferences by failing to identify whether relationships are spurious or meaningful. That is, they are picking only cases where a particular outcome has occured, as opposed to choosing cases where a particular set of variables are relevant.


Arguments about bipolarity and multipolarity suffer from just such a selection bias, as does much writing on the rise or the decline of great powers.

Some researchers argue that decision makers are risk acceptant in the face of uncertainty and are more risk averse as uncertainty diminishes. Certain others argue that multipolar systems are stable and bipolar systems are unstable, implicitly asuming that decision makers are risk averse in the face of uncertainty, but become more risk prone as uncertainty diminishes. Historical and statistical research produces very mixed results, lending no strong support to neither hypothesis. .....Again, by turning a variable (response to uncertainty) into a constant, the accompanying investigations have selected out the cases that belie their hypotheses. That is, only cases in which policy makers have chosen in a particular way are examined.


Ngaire Woods, (1996) Explaining lnternational Relations Since 1945, OUP



Thin and thick narrative analysis: On the question of defining and analyzing political narratives


To explore how we can define the concept of political narrative, looking at implications in terms of analyzing political discourse. The examination of the various strategies used to define narrative, leads to the suggestion that, at least in the context of political narrative analysis, we need structural definitions that stress the barest minimum for terming a message a narrative. Basing on the proposed strategy to define narrative, it is suggested that narrative analysis should operate on two levels: the “thin” level and the “thick” level. The thin level relates to events and situations described in a discourse and their order of appearance in the text. “Thick level” of analysis, relates to everything included in the “narration” and the relation between the components of the thin narrative.

Source: Shenhav, Shaul R.1
http://www.ingentaconnect.com/content/jbp/nari
/2005/00000015/00000001/art00005