William Cross, "Democratic Norms and Party Candidate Selection: Taking Contextual Factors into Account," Party Politics, 14 (September, 2008), 596-619.

First paragraph:
It has long been settled that candidate selection is one of the central functions of political parties. Scholars such as Sartori (1976: 64) have observed that the selection of candidates is the core activity that universally distinguishes parties from other political organizations. It is not surprising then that students of party organization and party democracy place considerable importance on norms of candidate selection when determining where power lies within a party. What has been considerably less studied, however, is the relative importance of party candidate selection among parties and across jurisdictions. In this article, I argue that the relative significance of party candidate selection varies depending upon a limited and identifiable set of contextual variables. While extremely influential in determining representational and policy outcomes in some states, candidate selection may have a marginal relationship with these in others. It follows from this that the strength of the normative argument for democratic organization of these processes is contextual rather than absolute, and thus varies across jurisdictions.

Figures and Tables:

Table 1. Factors influencing the relative importance of party candidate selection to representational and policy outcomes

Next to last paragraph:
This article suggests that, under certain circumstances, party candidate selection processes may be equally or more determinative of who ends up in the legislature than are general elections. These outcomes might occur, for example, when a single dominant party exists or when an electoral system provides general election voters with very limited choice. In these cases the arguments for democratically organized nomination contests have significant merit as these contests play an important role in determining democratic outcomes. Alternatively, in those systems where general election voters have significant choice, or where legislators have little influence over policy outcomes, there is considerable merit to the argument that other democratic interests (such as party-building) might best be advanced by permitting the parties to select their candidates in the method they believe most appropriate. The key point is that the full consideration of these arguments requires the ability to measure the relative significance of party candidate nomination within a particular political party and state.