Paul Pierson, Politics in Time (2004), Chapter 3.

Summary:
Temporal structure must be taken into account when studying political behaviour or structures [81]. Analyses in political science should take into account longer term structural forces. Ultimately, the time frame depends on the goal of analysis, however, methodological problems will arise if you fail to take into account the appropriate time length.
Key Definitions:
Time Horizon [Abbott 1988]: The period over which meaningful changes occur.
Three Paths:
*Cumulative Causes’ –> Change in a variable is continuous but very gradual [DV’s with long time horizons]
*Threshold or Inertial Effects –> Path dependency arguments are based on threshold effects, but PD arguments are not necessarily based on non-linear change (i.e. thresholds) [*IV’s and DV’s with long time horizons]
* Causal Chains –> X initiates a, b, c, which leads to y; i.e. a sequence of developments  leads to an outcome (Collier and Collier 1991; Carpenter 2001) [*IV’s with long time horizons]
These types of arguments face the infinite regress problem (i.e. why not search for the previous political choice – maybe *that* [was the important instigating point) [89]
•     Problem: Maybe people break the chain at “critical junctures” (Collier & Collier)
•     Problem: Maybe they break the chain where causes become hard to pinpoint
•     Problem: how do we delimit the casual chain based on the author’s theoretical interests?