|Module 3: Interpreting Data|
2. Using data to support an argument
To test a theory or answer a question a study is designed, sampling is conducted and the data is collected. The process of descriptive statistics then involves presenting the data in tables and graphs. The data may seem to indicate a clear conclusion about the population which has been sampled. But how strongly do the data support that conclusion? Is there strong evidence for the link between data and conclusion? How can you be sure that the effect observed is due to the experimental treatment and is not just an accidental result?
Deciding on the strength of the link between data - and making conclusions about the population - involves interpretation. The basis of how to make interpretation lies in another statistical process called inferential statistics. Inferential statistics involves the use of statistical methods and models to make measurable claims about populations (and population parameters) on the basis of samples (and sample statistics).
Usually researchers do not know the value of population parameters - they have to estimate them. But they do have measurements made on a sample -these are sample statistics. Researchers also realise that if they used a different sample from the same population to produce more data, the new sample statistics would be different to the first ones.
Inference uses probability to account for this sample variability. However, to make inferences you need to have designed a reliable, unbiased study so that the data that are produced are accurate and valid. Therefore, in order to make useful interpretations about data, or to assess the appropriateness of other interpretations, you need to first ask about how the data were produced and presented.
This page last updated 31 August 2009