Sampling Strategies

First, you need to read Chapter 2 in Fieldwork Techniques and Projects in Geography (Lenon and Cleves) and pages 159-162 in Geography: An Integrated Approach (Waugh 3rd Edition).

The three main sampling strategies are:

Systematic Sampling

In this method you take a sample at regular intervals, either along a transect (straight line), or on a grid system (point sampling).  You follow this method if you are expecting to find a change within an environment.  Many projects suit this sampling method and it leads to easier data representation techniques, since you can simply plot graphs of your data if you have used a transect.  An isoline map can be constructed from a grid of data.

Random sampling

This technique is used when you expect no variation in the data within an area and the sampling points can be located anywhere within a given area.  It is also likely that you need to try to ask questionnaires using this technique to avoid bias.  Strictly speaking you should use random number tables or generators to pick your sampling sites, but this can lead to clustering of data.  In this technique, your sampling sites are spread across an area, without order.  This tecnique can prove problematic when you need to represent your data; isoline maps, located proportional cirles or symbols will be needed.  You can refine your data, and give order to it at a later stage; for example by measuring the distance of each site from a given point so that you can construct a graph.

Stratified Sampling

This technique can be combined with either systematic, or random samping.  The idea here is that you choose the locations at which you are going to record a given number of samples, and then sample randomly or systematically in that place.  You can therefore select appropriate numbers of sites for given areas, and select sites that will give you the most meaningful results.  There are problems with bias associated with this method, and again, graphs might be more complicated to contruct.

The Importance of Sampling

For a project to be worthwhile, you need to collect enough data to construct meaningful graphs and to be able to perform a statistical test.  For a Spearman's Rank Correlation Coefficient Test to work, and be statistically significant, you will require a minimum of 15 sites.  A Chi Squared test will not work if any one of your categories contains a zero (0).  As a general rule, projects at Advanced Level should have at least 20 sampling points, with at least 50 questionnaire responses (if applicable).



Read the relevant literature, consider your topic, and choose a sampling strategy.  Justify that choice by discussing your hypothesis and geographical theory.  Write a paragraph describing your sampling strategy, with 3 bullet points to justify your selection.