Preparing for your appointment with the Statistical Support Service
Ideally statistical advice should be sought at the research project planning stage. Grant applicants may obtain an estimate of the costs of later planned statistical analysis, to include in their proposals. More importantly initial advice on choosing the best study design, defining analysis methods most suited to the design and determining the sample size needed to ensure reasonable power of finding the desired effect/s if present can be provided. Thus high quality quantitative research procedures can be fully explored.
Bring to your first appointment with the consultant: the protocol or methodology you plan to use, a description of the data to be collected, or secondary data analysed, and relevant references.
When you are at the data analysis stage, remember to prepare the data for analysis as appropriate, for example by entering your data onto an Excel spreadsheet, by checking thoroughly for data entry mistakes and checking that data has not been excluded. For some good practice guidelines on using Excel to prepare your data for statistical analysis see http://www.reading.ac.uk/ssc/publications/guides/topsde.html.
If you have already done some data exploration and analysis bring this too. All original and correct data observed or obtained should be brought to the appointment. Concerns about unusual data should be fully discussed at your appointment, and nothing should be removed or adjusted from the collected data.
When you are writing up your results or summary and drawing conclusions or making predictions as a result of your statistical work, remember to bring to your appointment all previous work on methodology and data, as well as your written draft so they can be properly co-ordinated. Correct interpretation of results and statistical reasoning will affect other aspects of the research findings.
For additional help the UCLA consulting unit web site provides many good resources in a number of areas
- choosing a method for analysis
- courses and seminars they have run (some with movies)
- how to run many different types of analyses using different software packages.