Statistics for the Measurement Community

Steven Gilmour

Queen Mary, University of London

Statistics is the science of data collection, processing, analysis and interpretation. It is relevant to all fields of endeavour which give rise to numerical data. Most importantly, statistics is a way of thinking, rather than just a collection of techniques. Although nobody can be a master of all statistical techniques, statistical thinking should be for everyone and should inform all our research. This session will cover some of the basic concepts of statistical thinking in the context of network measurements. We will also discuss some practical questions, such as how to talk to statisticians and some pitfalls to be avoided.

In order to interpret data sets of any size, we must consider the structure of our data and the objectives of our research. Clarity about which variables have been measured and what are the observational units will help in describing the structure carefully. Any known or unknown relationships among variables should be considered. The objectives of a data analysis can be very specific or very vague and the appropriate technique to use will depend on these objectives, as well as the structure in the data.

The first step in an analysis should always be to plot or tabulate the data. This should be done in a way which respects the structure in the data and illustrates the questions of interest.

Questions about how to collect data are even more important than questions of how to analyse them. We have to consider how to collect data which will be able to answer the objectives of our study. If we are using data which have been collected by others, we need to know as much as possible about how these data were collected. We must always realise that the answer to question “Can these data answer this question?” is not always “Yes”. It depends on the structure of the data, as well as the amount of data.

When attempting to engage the statistical community, it is important to realise that not all statisticians have the same interests. They range from those who are interested only in organising and tabulating the data, to those who do research in statistical mathematics. Some specialise in particular areas of application, while others specialise in particular techniques.

Reference

Hand, D.J. (2004) Measurement Theory and Practice: the World through Quantification. Edward Arnold.