28.6 Methodological Implications
28.6 Methodological Implications
Analytical methods for examining people-environment interactions should facilitate the inclusion
of the components listed in Table 28.2. This implies that analysts should make them explicit at
every stage of study, in conceptualizing research
questions, defining data needs, and sampling
frames, in analysis and interpretation of information, and in drawing conclusions.
The majority of these issues is common to approaches from the social science perspective, such
as political ecology, and that of ecological scientists, such as landscape ecology. These approaches
share many common elements, and the differences
between them represent clear opportunities for exploring more effective ways to examine the interactions between societal and biophysical processes
(Campbell, 1998). However, a comprehensive understanding of the complex interactions between
biophysical and socioeconomic processes, particularly of the issues of scale and time, remains incomplete. In fact, some scientists are calling for a
completely new approach, arguing that the integrating of biophysical and socioeconomic approaches "implies that the politico-social-ecological system be treated as a complex system, that an
understanding of ecosocio relations is truly a new
field that requires new methodologies, and those
methodologies need not have anything to do with
the standard methodologies of either ecology or the
social sciences" (Vandermeer, 1994, p. 105).
28.6.1 Data, Measurement,
and Verification 1
In modeling patterns of human activity and their
interactions with ecological systems, we are seldom
in a position to specify all the variables and interactions in a particular setting. Model building calls
for selection of some and omission of other variables and linkages. This selection should be based
on a clear theoretical-conceptual framework or on
empirical evidence that the variables and interactions omitted would not alter or add to the likely
interpretation. Because the model is selective in the
components that are included and excluded, some
linkages will be measured empirically, while others will be assessed qualitatively.
The inability to measure or completely specify
the system and the requirements of cost efficiency
and ease of representation often result in indirect
IThis section draws on Campbell (1996, 1998).
427
measurement and simplified representation of complex processes and situations. Pragmatic concerns
such as cost and time may constrain the ability of
an analyst to collect the specific information
needed to reflect a particular process or pattern. Recourse is often made to available data, often spoken of as "the best available information." This is
particularly true of macro-level data whose
measurement is for the most part opportunistic and atheoretical. Lacking the necessary resources to collect their
own data, investigators have usually searched for available indicators without adequately spelling out the assumed connections with the variables they are intended
to measure. Very frequently, macro-level researchers are
forced to rely on derived measurements that have been
based on decisions made by persons whose purposes are
very different from those of the investigators. Therefore,
it becomes essential to examine very carefully just how
these derived measures have been constructed, what homogeneity assumptions have been made, and how they
are to be linked to the important constructs in macrotheories. (Blalock, 1982, p. 237)
Blalock's concern with theory has to be complemented by empirical and experimental activity to
try to identify causal processes because
the demonstration that a specific mechanism can in theory give rise to a range of observed patterns is not proof
that that mechanism is indeed responsible for those patterns . . . in general, there will be many conceivable
mechanisms that could give rise to any set of patterns.
All that theory alone can do is to create a catalogue of
possible mechanisms; experiments are then needed to
distinguish among the candidate mechanisms. (Levin,
1992, p. 1948)
The patterns observed over broad areas represent
the outcomes of complex processes acting over subareas. In many cases, information on variables gathered over large spatial units, for example, a province
or nation, reflects a summary of complex events and
processes that occur within that unit. These variables
are therefore surrogates of this complexity. Their
meaning and interpretation depends on a clear understanding of the complexity and what elements of
it are reflected in such surrogate variables.
28.6.2 The Ecological Fallacy
Since the publication of Robinson's (1950) article,
we have been on notice of the ecological fallacy
whereby characteristics of individuals cannot be inferred from group characteristics. Aggregation can
yield summary measures with a descriptive utility,
but relationships between aggregated variables cannot be implied to individuals that compose the aggregate. There are interactions between the micro-
28.6 Methodological Implications
Analytical methods for examining people-environment interactions should facilitate the inclusion
of the components listed in Table 28.2. This implies that analysts should make them explicit at
every stage of study, in conceptualizing research
questions, defining data needs, and sampling
frames, in analysis and interpretation of information, and in drawing conclusions.
The majority of these issues is common to approaches from the social science perspective, such
as political ecology, and that of ecological scientists, such as landscape ecology. These approaches
share many common elements, and the differences
between them represent clear opportunities for exploring more effective ways to examine the interactions between societal and biophysical processes
(Campbell, 1998). However, a comprehensive understanding of the complex interactions between
biophysical and socioeconomic processes, particularly of the issues of scale and time, remains incomplete. In fact, some scientists are calling for a
completely new approach, arguing that the integrating of biophysical and socioeconomic approaches "implies that the politico-social-ecological system be treated as a complex system, that an
understanding of ecosocio relations is truly a new
field that requires new methodologies, and those
methodologies need not have anything to do with
the standard methodologies of either ecology or the
social sciences" (Vandermeer, 1994, p. 105).
28.6.1 Data, Measurement,
and Verification 1
In modeling patterns of human activity and their
interactions with ecological systems, we are seldom
in a position to specify all the variables and interactions in a particular setting. Model building calls
for selection of some and omission of other variables and linkages. This selection should be based
on a clear theoretical-conceptual framework or on
empirical evidence that the variables and interactions omitted would not alter or add to the likely
interpretation. Because the model is selective in the
components that are included and excluded, some
linkages will be measured empirically, while others will be assessed qualitatively.
The inability to measure or completely specify
the system and the requirements of cost efficiency
and ease of representation often result in indirect
IThis section draws on Campbell (1996, 1998).
427
measurement and simplified representation of complex processes and situations. Pragmatic concerns
such as cost and time may constrain the ability of
an analyst to collect the specific information
needed to reflect a particular process or pattern. Recourse is often made to available data, often spoken of as "the best available information." This is
particularly true of macro-level data whose
measurement is for the most part opportunistic and atheoretical. Lacking the necessary resources to collect their
own data, investigators have usually searched for available indicators without adequately spelling out the assumed connections with the variables they are intended
to measure. Very frequently, macro-level researchers are
forced to rely on derived measurements that have been
based on decisions made by persons whose purposes are
very different from those of the investigators. Therefore,
it becomes essential to examine very carefully just how
these derived measures have been constructed, what homogeneity assumptions have been made, and how they
are to be linked to the important constructs in macrotheories. (Blalock, 1982, p. 237)
Blalock's concern with theory has to be complemented by empirical and experimental activity to
try to identify causal processes because
the demonstration that a specific mechanism can in theory give rise to a range of observed patterns is not proof
that that mechanism is indeed responsible for those patterns . . . in general, there will be many conceivable
mechanisms that could give rise to any set of patterns.
All that theory alone can do is to create a catalogue of
possible mechanisms; experiments are then needed to
distinguish among the candidate mechanisms. (Levin,
1992, p. 1948)
The patterns observed over broad areas represent
the outcomes of complex processes acting over subareas. In many cases, information on variables gathered over large spatial units, for example, a province
or nation, reflects a summary of complex events and
processes that occur within that unit. These variables
are therefore surrogates of this complexity. Their
meaning and interpretation depends on a clear understanding of the complexity and what elements of
it are reflected in such surrogate variables.
28.6.2 The Ecological Fallacy
Since the publication of Robinson's (1950) article,
we have been on notice of the ecological fallacy
whereby characteristics of individuals cannot be inferred from group characteristics. Aggregation can
yield summary measures with a descriptive utility,
but relationships between aggregated variables cannot be implied to individuals that compose the aggregate. There are interactions between the micro-