428
Assessing Human Processes in Society-Environment Interactions
and macrolevels, but they cannot be inferred from
macrolevel conditions. Rather an explicit theoretical-conceptual framework is required to explicate
them.
A recent study of policy making for natural resource management over time in Rwanda illustrates
the ecological fallacy (Brown et al., 1996). In the
early 1980s, national-level information indicated
that the Rwandan population was growing rapidly
and that deforestation and land degradation were accelerating. The association between them was interpreted as causal-the driving force of environmental deterioration was rapid population growth in the
context of limited land resources. The government
established policies on land reclamation, resettlement, and family planning. These continued into the
1990s, when policies to diversify rural livelihoods
through the off-farm-nonfarm sector were added.
However, higher-resolution studies using air
photography and household surveys reached a different conclusion. The analysis of air photography
showed that government-sponsored tree cutting to
establish a buffer zone around national forests and
tea plantations accounted for a great deal of the deforestation. In areas under cultivation, the information from farm households demonstrated that,
while the contiguous tree cover seen on satellite
imagery and air photographs might have decreased
over time, the number of trees had actually increased as a result of planting by poor farmers (Olson, 1994; Olson et al., 1995).2
Government policy was based on national-level
information and assumed relationships among variables that had been imputed to the constituents
making up the national-level information. Failure
to respect the ecological fallacy resulted in inaccurate representation of rural conditions and in
inappropriate policy. Understanding the ecological
fallacy is critical to effective analysis of peopleenvironment interaction. Without it, facile interpretations of environmental issues are possible, based on
2 An association between population growth and indices of
enviro~ental deterioration is often found in global, lowresolution data. Many have adopted a position that this
lends support ~o .Malthusian perspectives, that population
gro~ IS a drivmg force of envlfonmental deterioration.
Empmc~ studies at a finer re~olution have challenged this,
postula~mg a far more comphcated relationship. There are
~creasll~g num~ers of e~am~les where population growth
IS aSSOCIated WIth amelioration of environmental conditions. There is a growing literature on increased forest
cover. in agri~ultural area~ where growth in population is
aSSOCIated WIth tree planting as rural populations seek to
produce much of their own wood for fuel, building, fencmg, and the like (Dewees, 1993; Tiffen et al., 1994; Castro, 1995; Olson et al., 1995).
statis~cal association among generalized variables,
resultmg in remedial policies based on incomplete
understanding of the circumstances. In these situations, both science and policy are poorly served.
28.6.3 Dynamics of Scale and Time
The dynamics of society-environment interactions
and the scalar structure of information and its implications for interpretation call for explicit attention to be paid to the interacting concepts of scale
and time in the analysis and modeling of such interactions. These interactions fall within the do?Iain of a number of. different disciplines, and the
Issues of scale and tIme are complex in these interdisciplinary settings.
Defming the appropriate spatial resolution for
analysis of environmental issues is one of the most
diffi~ult problems facing environmental analysts
(Robmson, 1950; Harvey, 1968; Blaikie and Brookfield, 1987; Meentemeyer and Box, 1987; Hall et al.,
1988; Turner et al., 1989; Turner, 1990; Campbell
and Olson, 1991; Izac and Swift, 1994; Klijn and de
Haes, 1994). Different processes have different dominant scales of activity. For example, government
economic policy is national in its scope, cultural
characteristics and rainfall patterns are regional, and
soil conditions and agriCUltural practices are at the
scale of the community, household, or field. These
interact in determining land-use options, but the
available information mayor may not permit the interactions to be depicted. It is important therefore
that the gaps in information and knowledge that exist in linking across scales be examined and identified using methods such as description, statistical relationships, and simulation models.
The issues have been concisely summarized for
land-use/cover change:
It is well documented that as questions about landuse/cover change are addressed at different space-time
scales, so the answers vary significantly. More importantly, these answers cannot be adequately linked. For
~x~ple, the "cause" factors of land-use/cover change
Identified at the global scale cannot be integrated well
with thos~ found at the local level. As a result, coupling
local, regIOnal and global models is extremely difficult.
Moreover, resolving or making significant advancement
on the scales issue will go a long way towards resolving
ideological and polemical debates over the causes of
land-use/cover change. (IGBP, 1994, p. 83)
Scale
There is widespread interest among scientists to better understand the links between and among micro-,
Assessing Human Processes in Society-Environment Interactions
and macrolevels, but they cannot be inferred from
macrolevel conditions. Rather an explicit theoretical-conceptual framework is required to explicate
them.
A recent study of policy making for natural resource management over time in Rwanda illustrates
the ecological fallacy (Brown et al., 1996). In the
early 1980s, national-level information indicated
that the Rwandan population was growing rapidly
and that deforestation and land degradation were accelerating. The association between them was interpreted as causal-the driving force of environmental deterioration was rapid population growth in the
context of limited land resources. The government
established policies on land reclamation, resettlement, and family planning. These continued into the
1990s, when policies to diversify rural livelihoods
through the off-farm-nonfarm sector were added.
However, higher-resolution studies using air
photography and household surveys reached a different conclusion. The analysis of air photography
showed that government-sponsored tree cutting to
establish a buffer zone around national forests and
tea plantations accounted for a great deal of the deforestation. In areas under cultivation, the information from farm households demonstrated that,
while the contiguous tree cover seen on satellite
imagery and air photographs might have decreased
over time, the number of trees had actually increased as a result of planting by poor farmers (Olson, 1994; Olson et al., 1995).2
Government policy was based on national-level
information and assumed relationships among variables that had been imputed to the constituents
making up the national-level information. Failure
to respect the ecological fallacy resulted in inaccurate representation of rural conditions and in
inappropriate policy. Understanding the ecological
fallacy is critical to effective analysis of peopleenvironment interaction. Without it, facile interpretations of environmental issues are possible, based on
2 An association between population growth and indices of
enviro~ental deterioration is often found in global, lowresolution data. Many have adopted a position that this
lends support ~o .Malthusian perspectives, that population
gro~ IS a drivmg force of envlfonmental deterioration.
Empmc~ studies at a finer re~olution have challenged this,
postula~mg a far more comphcated relationship. There are
~creasll~g num~ers of e~am~les where population growth
IS aSSOCIated WIth amelioration of environmental conditions. There is a growing literature on increased forest
cover. in agri~ultural area~ where growth in population is
aSSOCIated WIth tree planting as rural populations seek to
produce much of their own wood for fuel, building, fencmg, and the like (Dewees, 1993; Tiffen et al., 1994; Castro, 1995; Olson et al., 1995).
statis~cal association among generalized variables,
resultmg in remedial policies based on incomplete
understanding of the circumstances. In these situations, both science and policy are poorly served.
28.6.3 Dynamics of Scale and Time
The dynamics of society-environment interactions
and the scalar structure of information and its implications for interpretation call for explicit attention to be paid to the interacting concepts of scale
and time in the analysis and modeling of such interactions. These interactions fall within the do?Iain of a number of. different disciplines, and the
Issues of scale and tIme are complex in these interdisciplinary settings.
Defming the appropriate spatial resolution for
analysis of environmental issues is one of the most
diffi~ult problems facing environmental analysts
(Robmson, 1950; Harvey, 1968; Blaikie and Brookfield, 1987; Meentemeyer and Box, 1987; Hall et al.,
1988; Turner et al., 1989; Turner, 1990; Campbell
and Olson, 1991; Izac and Swift, 1994; Klijn and de
Haes, 1994). Different processes have different dominant scales of activity. For example, government
economic policy is national in its scope, cultural
characteristics and rainfall patterns are regional, and
soil conditions and agriCUltural practices are at the
scale of the community, household, or field. These
interact in determining land-use options, but the
available information mayor may not permit the interactions to be depicted. It is important therefore
that the gaps in information and knowledge that exist in linking across scales be examined and identified using methods such as description, statistical relationships, and simulation models.
The issues have been concisely summarized for
land-use/cover change:
It is well documented that as questions about landuse/cover change are addressed at different space-time
scales, so the answers vary significantly. More importantly, these answers cannot be adequately linked. For
~x~ple, the "cause" factors of land-use/cover change
Identified at the global scale cannot be integrated well
with thos~ found at the local level. As a result, coupling
local, regIOnal and global models is extremely difficult.
Moreover, resolving or making significant advancement
on the scales issue will go a long way towards resolving
ideological and polemical debates over the causes of
land-use/cover change. (IGBP, 1994, p. 83)
Scale
There is widespread interest among scientists to better understand the links between and among micro-,