Page 709 - Environment: The Science Behind the Stories

Graph Type: Scatter Plot Each data point is
independent of others;
A scatter plot is often used when data are not sequential and 8 note Syria and India have
when a given x-axis value could have multiple y-axis values. A nearly the same x-value
scatter plot allows us to visualize a broad positive or negative 7 Ethiopia
correlation between variables. FIGURE B.8 shows a negative cor- 6
relation (that is, one value goes up while the other goes down):
Nations with higher rates of school enrollment for girls tend 5 Cambodia Guatemala
to have lower fertility rates. Jamaica has high enrollment and Total fertility rate (1995–2000) 4 Kenya Syria Egypt South
low fertility, whereas Ethiopia has low enrollment and high Africa
Peru
Colombia
fertility. 3 2 India Vietnam Jamaica

0 1
0 10 20 30 40 50 60 70 80 90 100
Female secondary school enrollment rate (%)
FIGURE B.8 Fertility rate and female education. (Figure 8.19, p. 224)

10 Key Concept: Statistical Uncertainty
Most
variation Most data sets involve some degree of uncertainty. When a
Woody fuels (Mg/ha) 6 Least urements, the researcher may want to show the degree to
8
graphed value represents the mean (average) of many meas-
which the raw data vary around this mean. Mathematical tech-
niques are used to obtain statistically precise degrees of vari-
4
variation
ation around a mean. Results from such statistical analyses
2
may be expressed in a number of ways, and the two graphs in
this section show methods used in this book.
0
Not Burned Burned & In a bar chart (FIGURE B.9), thin black lines called error
burned only logged bars may be shown extending above and below the tops of the
bars, or simply above them. In this example of woody debris
FIGURE B.9 Fine-scale woody debris left after
treatments in a salvage logging study. (Figure 2a, remaining after salvage logging, error bars show the most vari-
p. 343) ation in measurements occurred at "Burned and logged" sites.

Sometimes shading is used to express variation around a mean.
The black and red data lines in FIGURE B.10 show mean global 250
sea level readings since 1870. The data line is surrounded by 200 Gray shading
gray shading indicating statistical variation. Note how the 150 shows variation
around mean.
amount of statistical uncertainty is exceeded by the sheer scale
of the sea level rise. This gives us confidence that sea level is Sea level rise (mm) 100
truly rising, despite the statistical uncertainty we find around 50
mean values each year. Note also how the amount of uncer- Tide gauge data
tainty has decreased through time. This reflects improvements 0 Satellite data
in technology enabling more accurate measurements. –50 Statistical uncertainty
The statistical analysis of data is critically important in 1870 1900 1950 2000
science. In this book, we provide a broad and streamlined Year
introduction to many topics, so we often omit error bars from
our graphs and details of statistical significance from our dis- FIGURE B.10 Change in global sea level, measured since
cussions. Bear in mind that this is for clarity of presentation 1870. (Figure 18.14, p. 517)
only; the research we discuss analyzes its data in far more
depth than any textbook could possibly cover.
B-4

Z02_WITH7428_05_SE_AppB.indd 4 13/12/14 10:51 AM