Interpreting Data

1. Introduction and review

1.1 Review

In Modules 1 and 2, you learned about ways of producing and representing data. In this module you will learn to critically examine the way that these data can be interpreted.

For interpretations to be meaningful, it is necessary to know

• how the variables were defined and
  
• the method used to produce the data.

This is because it is only within a context that numbers become data. As you will remember, data are presented in tables and graphs because these are ways of clearly presenting important features and of highlighting patterns that exist in the data. Tables and graphs are effective vehicles for presenting the data to support an argument.

So far most of the statistical ideas and tools we have studied in Module 1 and Module 2 can be referred to as descriptive statistics. That is, we considered how to plan for a study and collect data (almost always based on samples), and then how to describe (present) the results.

What we now need is to understand how to interpret our sample results and how to apply these to the population from which we sampled. This is a second part of statistics - inferential statistics - we infer from the sample what the data suggests is true of the population.

1.2 Module objectives

1. interpret data in tables and graphs with a special emphasis on data involving two variables (bivariate data);
   
2. use skills developed in earlier modules to assess the quality of data;
   
3. demonstrate an understanding of the role of hypothesis development in deciding whether or not an effect is present in a population;
   
4. demonstrate an understanding of the concepts of association and causation;
   
5. assess the production, representation and interpretation of data.

1.3 Follow the link to read the details of this case study on data and its use. Impact of ambient air pollution on birth weight in Sydney, Australia (Robotham 2005) [1]

"Babies at risk from pollution."

In the Sydney Morning Herald of 28 July 2005, the results of a study on all 138,000 Sydney babies born from 1998 to 2000 were quoted. Dr. Vicky Sheppeard, from the NSW State Health Department's environmental health branch, said babies exposed to the highest pollution levels before birth were on average about 12 grams lighter than those at the lowest end of the pollution range. Commenting on this study, Professor Bruce Armstrong (Head of the School of Public Health at the University of Sydney), said it was "...hard to dismiss the body of evidence that indicates air pollution is associated with low birthweight ... one has to consider this is probably cause and effect ... It re-emphasises the point that we should be keeping air pollution as low as we possibly can ... "

From this article the main points emerge as:

1. A study was planned.
   
2. The study was done to examine an hypothesis - namely that pollution affects birthweights of babies.
   
3. A sampling method was planned (in this case it was all babies in the three years 1998 - 2000 inclusive).
   
4. An argument is made: a connection is drawn between the two variables which were measured - it is stated the sample shows increasing pollution causes decrease in birthweight.
   
5. An argument is made: the sample in the study is generalized to the population of all babies born in Sydney, with further reference made to global evidence.
   
6. Implications are made from the results of the study for public health policy and government spending/priorities - which in this case is the need to reduce pollution levels in Sydney.

This article is a good example of how statistics can be used - even in the popular press.

Next - Menu