The Dictionary of Human Geography

institutions and their role in the making of
space and place. But its most significant im-
pact has undoubtedly been in economic geog-
raphy, where it has forged a path between
mainstream approaches and radical political
economy (seeinstitutional economics). aa

instrumental variables A statistical and
econometric technique to estimatemodelsof
theregressiontype that have problems of
endogeneity. In the standard multiple regres-
sion model of the general form:

Y¼b 0 þb 1 X 1 þb 2 X 2 þe,

whereYis the dependent variable,X 1 andX 2
are the independent variables andeis the ran-
dom error term, the assumption is that the two
independent variables are genuinely exogen-
ous. Suppose, however, thatX 1 is endogen-
ous, itself determined by Y and other
exogenous variables such asX 3 andX 4 .An
example might be ifYrepresents regional em-
ployment growth in ‘hi-tech’ services andX 1 is
regional income:Yis influenced byX 1 , butX 1
is itself influenced byY, implying thatX 1 is
correlated withe, a violation of the assump-
tions of regressions.
The technique of instrumental variables
works by finding a set of ‘instruments’, vari-
ables that are themselves exogenous and are
good at predictingX 1. Such a set would
include X 3 ,X 4 (andX 2 for the version of in-
strumental variables known as two-stage least-
squares), otherXvariables and perhaps spatial
or temporal lagged values of X 3 and X 4.
Denoting this set of instruments asZ, one
regressesX 1 againstZ, obtaining the predicted
or fitted estimates ofX 1 , denoted asX^ 1. The
estimateX^ 1 depends only onZ, and so is
uncorrelated withe, and the original regres-
sion forYis now estimated, replacingX 1 by
X^ 1. This procedure, which thus requires two
regressions, eliminates the endogeneity bias.
The limitation of instrumental variables lies
in the ability to find a suitable set of instru-
ments that is both exogenous and good
at predicting X 1 – weak instruments will
circumvent the endogeneity but provide
poor estimates of thebcoefficients (for a full
discussion, see Bowden and Turkington,
1990).
An interesting recent geographical example,
widely publicized through The Economist
magazine, is a study by James Feyrer and
Bruce Sacerdote of the economic prosperity
of 80 former-colonial islands around the
globe. They argue that the current prosperity

is directly related to length of colonization, but

recognize that fertile and promising islands

may have attracted colonists; that is, endo-

geneity. They then use instrumental variables

with measures of wind speed and direction

as instruments: before steam-ships, such vari-

ables may well have influenced the date of

colonization, but are strictly exogenous. lwh

Suggested reading

Kennedy (2003). See also The Economist,2

November 2006.

instrumentalism A philosophy of science

concentrating on end results such that science

becomes an instrument judged in terms of

practical utility rather than the truth or falsity

of theory. Gregory (1978a) identified instru-

mentalism as a feature of geography’squanti-

tative revolution, informed by a particular

reading of positivism, the generation of

models being welcomed for their potential to

shape policy. Such work entailed a narrow

conception ofrelevance, subsequently cri-

tiqued. Concerns over instrumentalism have

informed debates overgeographic informa-

tion systems, with calls for GIS to adopt a

more critical and reflexive perspective on the

conditions of its production and utilization

(Pickles, 1995a). dmat

Suggested reading

Gregory (1978).

integration The creation and maintenance

of intense and diverse patterns of interaction

and control between formerly more or less

separate social spaces. Integration involves

the bringing together of different systems of

meaning and action founded in different sets

of social relations. It takes place in different

registers – economic, political and cultural –

and so is an inherently uneven process. This is

compounded because integration takes place

through – not merely over –timeandspace:

without the formation of new times and

spaces, the social relations embedded in inte-

gration could not be constituted. As such, it is

profoundly affected by technical change.

Innovations in the movement of people and

commodities, and most recently in the elec-

tronic transmission of ideas, images, infor-

mation and cultural forms, have made possible

new modes of interaction and new ways of

being ‘present’ in other places (seetime–space

compression;time–space distanciation).

Places have never been closed, cellular sys-

tems; people have always been caught up in

Gregory / The Dictionary of Human Geography 9781405132879_4_I Final Proof page 387 31.3.2009 7:05pm

INTEGRATION