**Main Argument**: Cross-national research is subject to Galton’s problem: autocorrelation of data between two countries owing to diffusion. There are two ways to get rid of Galton’s problem: get rid of it/embrace it. This can be done by operationalising diffusion and by including both diffusion and functional processes as an explanation for political instability.

**Key Definitions:**

*Galton’s problem*: problem of drawing inferences from cross-cultural data due autocorrelation

== Notes: ==

Naroll and others have devised +9 solutions to Galton’s problem for use in cross-cultural studies:

* Sampling sol – most solutions focus on case selection in a way that min’s effects of direct borrowing/diffusion among units in sample (4)

* Diffusion seems to require a “sampling” solution because researchers consider diffusion as somehow getting in the way of functional or “real” explanations (ie, focus on geo prox to remove diffusion) however, diffusion isn’t a problem to be ‘removed’ and doesn’t get in way of functional or ‘real’ explanations

* Vermeulen & de Ruijter contend that functional explanations are nomothetic, while diffusion or historically based explanations are ideographic

* Prz and Teune pt out that challenge is to replace historical particularities and place names w/gen variables they rep – thus, functional explanations emphasize intra-sys processes and diffusion explanations stress inter-sys processes

* Vermeulen & de Ruijter suggest to add additional variables (5) to analysis to give diffusion and function some explanatory status

**Problems:**

* Rare to find geographical isolation anymore

* Reliance on dichotomous data we use continuous, or at least ordinal, data [9]

* Many solutions involve samples that include only a small proportion of the total population of cultural units

**Solutions:**

* All cross-national researchers need to build into their designs ways of testing for the effects of both diffusion and function

* Diffusion and function are not incompatible, but do represent two different sorts of processes in a political system [20]

* Importance of including both external (diffusion) and internal (function) variables in cross-national studies

* Use P&T’s most different systems design [23]; maximise variance in DV; by including only units from different areas in sample, effects of diffusion = limited/eliminated; any resulting relationship = more easily interpreted in terms of intra-systemic effects

* Most diff systems sampling design = inferior to including additional variables which uses partial correlation and regression [25]

* Even with MDSD, cannot be entirely sure you’ve ruled out diffusion effects

* Important not merely to rule out diffusion, but to integrate both function and diffusion into a single explanation

* Also suggests measuring diffusion in several directions at once through use of a matrix rather than through pairing a country with only one other