Geography Reference
In-Depth Information
in the rest of the topic. Material of this nature is sometimes placed, for good reason, in
the appendices of introductory topics. In this topic, such material is presented within
the main text so that there is a direct transition from standard aspatial statistics (i.e.
methods that do not take into account the spatial location of observations) to their
spatial equivalents. h is chapter deals exclusively with standard aspatial statistical
methods. Methods which do take spatial location into account are the subject of later
sections and Section 4.8 deals with the principles of one statistical approach to charac-
terizing spatial variation (i.e. geographical patterning) in the property of interest.
Before proceeding, it is useful to consider how the kinds of data we have to work
with may dif er. Data may be divided into four types, which contain dif erent amounts
of information. h ese data types are:
Nominal
An arbitrary naming scheme, for example ethnic group (White, Caribbean,
African).
Ordinal
Values are ordered, but there is no information on the relative magnitude
of values, for example small, medium, large.
Interval
h e intervals between measurements are meaningful, but there is no natu-
ral zero point, for example temperature. Dif erences between adjacent values are equal,
i.e. 28-27 is the same as 99-98. Temperature (where zero is arbitrary, e.g. the freezing
point of water for degrees Celsius) is ot en cited as an example of an interval variable
(see, for example, Ebdon, 1985; O'Sullivan and Unwin, 2002). For two temperatures in
degrees Celsius (e.g. 10°C and 20°C) and degrees Fahrenheit (50°F and 68°F), the ratios
between the two sets of values are dif erent:
10°C
/
20°C
= 0.5 and
50°F
/
68°F
= 0.74.
Ratio
Values with a natural zero point, ratios as well as intervals, are meaningful.
For example, the ratio of 25 to 50 mm is the same as the ratio of the same measure-
ments in inches (i.e.
25 cm
/
50 cm
= 0.5 and
9.8 in
/
19.7 in
= 0.5).
h e main concern in this chapter is with the analysis of interval and ratio data and this
is also the main focus in Chapters 8, 9, and 10.
Univariate statistics
3.2
h e principal focus in this section is on what are termed 'descriptive statistics'—that
is, methods to summarize or describe observations (measurements of some prop-
erty). Summarizing an individual variable (e.g. precipitation amount) is done with
reference to its distribution. h e distribution of a variable refers to the set of values
ordered from the smallest to the largest. Ot en, identical or similar values are grouped
together, for example values 0-10 may be grouped, then values 11-20 and so on. In
this way, we can refer to the frequency of values. For example, are most values very
small with only a few large values or is there an even proportion of small and large
Search WWH ::
Custom Search