Hypothesis Generation: Deductive Testing Leading to Generalization
The capacity to test hypotheses is a central feature of deductive social science. Hypotheses are “links in a theoretical causal chain” that aid in determining the “direction and strength of a relationship between two variables” (Neuman, 2000, p. 131). Appendix A presents a comprehensive display of the range of characteristics and variables discussed throughout this book.
Within a given study, any combination of variables may be viewed as independent, dependent, or intervening. In Table 11.1 we introduce a set of potential independent and dependent variables. We distinguish between four types of potential hypothesis:
- ? Actor characteristics impacting actor behaviors (actor-to-actor relations)
- ? Actor characteristics impacting network-wide characteristics and behaviors (actor-to-network interactions)
- ? Network-wide characteristics impacting actor characteristics and behaviors (network-to-actor interactions)
- ? Network-wide characteristics to network-wide behaviors (network-to-net- work relations)
Table 11.1 outlines some of the ways that hypotheses may be generated across these types.
A much more comprehensive review of the range of hypotheses that have been generated and tested across the literature is needed. For some of these variable parings we drew on the hypothesis laid out by Michael Provan and Patrick Kenis based on their network governance model (2007).
Hypothesis testing requires that independent and dependent variables be quantifiable. In some instances, variables are naturally reported and conceived of numerically (as in the case of money and other financial resources). Others require systemic attempts to define an “operationalizable” that adequately matches abstract constructs with empirically quantified values.
Table 11.1 Range of Potential Hypotheses to Be Generated from a Governance Network Taxonomy
|
Levels of Interaction |
Independent Variable(s) |
Dependent Variable(s) |
Example of Generic Hypothesis |
|
Actor to actor |
Individual network actor's contribution of capital resources |
Network actor's roles; functions; ties within the network |
Network actors contributing higher levels of capital resources to the network are positioned as principals within the network |
|
Social sector of network actors |
Network actor's roles; functions; ties within the network |
Public sector network actors are more likely to play a lead organization role than nongovernmental actors under x circumstances Private sector actors are less likely to contribute financial resources to a governance network under y circumstances |
|
|
Actor to network |
Actor perceptions of trust of other network actors |
Governance structures |
Where trust between network actors is high, a shared governance structure is more apt to form Where trust is intermediate, a network administrative organization is more apt to form Where trust is low, a lead organization structure is more apt to form (Extrapolated from Provan and Kenis, 2007) |
Table 11.1 Range of Potential Hypotheses to Be Generated from a Governance Network Taxonomy (Continued)
|
Levels of Interaction |
Independent Variable(s) |
Dependent Variable(s) |
Example of Generic Hypothesis |
|
Actor perceptions of trust of other network actors |
Performances outcomes |
Those networks that exhibit high levels of trust between network actors are more apt to perform than those networks with low levels of trust |
|
|
Citizen, elected representatives, interest groups, courts' interest in governance network performance |
Performance outcomes |
The existence of democratic accountabilities ensures more effective network performance |
|
|
Actor social sector Geographical scale |
Network governance structures; network functions |
The more homogenous actors' sector and geographic scale, the more likely that ... |
|
|
Network to actor |
Network performance |
Actor acquisition of capital resources |
Networks performing at a high level infuse network actors with Z kinds of capital resources |
|
Network governance structure |
Actor perceptions of trust in other members |
Shared governance structures generate higher degree of trust between actors than either lead organization or network administrative organizations (extrapolated from Provan and Kenis, 2007) |
Table 11.1 Range of Potential Hypotheses to Be Generated from a Governance Network Taxonomy (Continued)
|
Levels of Interaction |
Independent Variable(s) |
Dependent Variable(s) |
Example of Generic Hypothesis |
|
Network governance structures; network functions |
Actor social sector; geographic scale |
Actor participation in governance network of certain structures and functions contributes to sector blurring, or changes in their scope of geographic focus |
|
|
Network to network |
Network governance structures |
Network policy function; network policy domain function |
Certain governance structures are more amenable to the pursuit of particular policy functions and policy domain functions than others |
|
Network policy function; network policy domain function |
Network governance structure |
X governance network functions lead to A network structures |
|
|
Network governance structure |
Network performance |
Certain governance structures lead to better network performance than others within a particular policy domain |
|
|
Network functions |
Network performance |
Governance networks that take on Y functions are more apt to perform better than those networks that do not take on those functions |
|
|
Network performance indicators and performance management systems |
Network performance outcomes |
Quality of performance management systems lead to better network performance |
(continued)
Table 11.1 Range of Potential Hypotheses to Be Generated from a Governance Network Taxonomy (Continued)
|
Levels of Interaction |
Independent Variable(s) |
Dependent Variable(s) |
Example of Generic Hypothesis |
|
Network governance structure |
Network change and adaptation |
Networks of B network structure are more apt to change and adapt than those of C network structure |
