Descriptive Statistics - Agricultural Statistical Data Analysis Using Stata - page 61

deal with exploratory analyses. This may be the beginning of a research
project to tease out associations, differences, or trends. In addition,
statistics are used to analyze large datasets of information for specific
details that may be present. For example, census data may contain a
great deal of information that is not immediately apparent. Statistics
can help identify such information. In both of these cases, there is no
formal experimental plan; however, there may be specific methods to
acquire these data to ensure unbiased results. Much of agricultural
research deals with planned experiments with carefully planned designs
and treatment selection, which will be the emphasis of this topic.
Biological systems by their nature will vary from one individual
to another. This makes it difficult to determine whether treatment
effects are real or just an artifact of these intrinsic differences. Even
if two populations are treated exactly the same, they will differ when
measured. For example, two plots of onions grown under the exact
same conditions, when harvested will have different yields. These
yields under these conditions obviously reflect no real difference.
When an experiment is conducted and inevitably there are measured
differences, are these differences real or do they reflect the intrinsic
differences between individuals in a population? Various statistical
procedures have been developed that give us a means of measuring
and determining if these differences are real.
Data collected in experiments can be of a wide range of types.
Numeric data can be parametric which means that it consists of a con-
tinuous range of numbers. An example would be the weight of experi-
mental animals or the yield from vegetable plots. Other data may be
nonparametric, which would include categorical data. These data would
include things like sex (male or female). These two different types of
data would use different statistical approaches for analysis. These differ-
ent data types also are often referred to as continuous or discrete.
Ordinal data are yet another type of nonparametric or discrete data
that use ranks. Ranked data would use yet another type of statistical
approach for analysis. Data may include counts as well. Count data
would be analyzed in a different method from continuous data.
In some cases, nonparametric or discrete data can be transformed
to meet the criteria that would be used with parametric data (see
Chapter 11). Or specific tests that were developed for nonparametric
data can be used (see Chapter 12).