Environmental Engineering Reference
In-Depth Information
Key elements of the experimental design were:
Four Amerindian groups were visited in Bolivia by two separate researchers,
who administered the same household survey. Inter-observer reliability was
tested in a pilot study.
Four hundred and forty-three households were surveyed overall, in 42
villages. The sample proportion averaged 4.3% of the households over the
four ethnic groups.
The male and the female heads of the households were surveyed separately.
The analyses were carried out for each of these, but the results were similar,
and so the male head of household's results were used in the main analyses.
Researchers used one-week recall methods, asking the respondent to list all
the fish and bushmeat consumed in the week before the interview. This was
the dependent variable.
They also obtained information on explanatory variables, including income
from the farm, wage labour and other activities. Wealth was measured by ask-
ing whether the household owned each of 12 physical assets. Each of these
assets was assigned a monetary value based on its current price in the village
being surveyed.
Other explanatory variables included household size, education level of the
household head, village and ethnic group of the respondent, and the price of
fish and domestic meat in the village where the household lived.
The data analysis included:
A broad understanding was gained of differences between the four ethnic
groups. These included levels of education, degree of isolation from commer-
cial markets, ability to speak Spanish, level of contact with and threat from
outsiders. This contextual understanding is important in explaining differ-
ences between the groups highlighted in the statistical analyses.
Six linear regressions were carried out, three for fish and three for bushmeat.
These included a regression of consumption on explanatory factors for the sam-
ple as a whole and for the top and bottom halves of the income continuum.
Dividing the sample into these two income groups helped to elucidate any sub-
stantial differences in consumption patterns between richer and poorer people.
The regressions were carried out using the natural logarithms of all the vari-
ables. This meant that the coefficients of the variables were interpretable as
elasticities (i.e. the percentage change in consumption given a unit change in
price or income). The effects of ethnic group and village were also tested. Only
the price and income elasticities were reported.
The results showed that:
About 60% of households ate neither fish nor bushmeat. This led to weaker
inferences than if the entire sample had eaten these goods.
Fish was an inferior good (i.e. the higher the household income, the less fish
they ate). This effect was strongest for the lower income group.
Search WWH ::




Custom Search