Geography Reference
In-Depth Information
an individual listed on neither patent has collaborated with members of both
i
's and
j
's
teams, the distance is one, and so forth. If no path connects members of the two teams,
the distance is ∞. See Singh (2005) for a complete description of his approach.
Based on the distance measure, we construct three indicator variables for each
dyad:
11
Close collaboration
ij
= 1 if the distance between patents
i
and
j
is less than four; 0
otherwise.
Far collaboration
ij
= 1 if the distance between
i
and
j
is four or greater but less than ∞;
0 otherwise.
Unconnected
ij
= 1 if no path connects
i
and
j
.
The shorter the path between
i
and
j
, the better the access to the template enjoyed by the
team involved in patent
j
. Our core hypothesis is that this superior access translates into
a higher probability of citation especially when the components of patent
i
display inter-
mediate interdependence. Accordingly, we expect the gap in citation probability between
a close and a far inventor - the probability that a close inventor cites a focal patent minus
the probability that a far inventor cites the patent - to peak at an intermediate level of
k
.
Although our collaborative distance measure provides direct evidence of access
and we believe that it captures many of the important connections between inventors,
inventors also have many other types of relations that might also facilitate access. For
example, a potential recipient might be a friend of the source even if they have never
collaborated. Attempting to identify all of the potential relationships existing in any
population of individuals is not feasible, but we can examine two factors - geographic
proximity and joint organizational membership - that tend to structure social rela-
tionships and therefore may proxy for unobserved social paths between our source-
recipient dyads. As McPherson et al. (1992, p. 154) note: 'Homophily structures the
l ow of information and other social resources through the network so that the dimen-
sions themselves stand as proxies for the number of intervening steps in transmissions
through the system.'
Geographic proximity
Space represents one important dimension that structures social
interaction. Indeed, some of the earliest literature on social networks emphasized the
dramatic decline in the likelihood of a social relation as two parties became increasingly
distant (Bossard, 1932; Park, 1926). Accordingly, we develop a measure of geographic
proximity for each patent dyad:
Geographic proximity
ij
= the natural log of the distance in miles between the
i rst inventors listed on patents
i
and
j
multiplied by negative one (so that
larger values indicate greater proximity).
12
As with our direct measure of social proximity, we expect geographic proximity to have
the greatest impact on citation likelihood when the potentially cited patent displays
moderate interdependence.
