Statistical data analysis

Do you have a dataset? And do you need assistance with the statistical data analysis of it? Epi Result offers statistical support with the analysis of epidemiological data, using different techniques, and gives guidance in the interpretation of research findings.

Epi Result can either do the statistical data analysis for you, or assist you in the process by giving guidance and commenting on your output. Either way, we will go through the following steps together, in order to systematically analyse the data:

1  Familiarization with the dataset

In order for me to get to know the data, I need to know the meaning of each variable (e.g. it’s link to questionnaire items) and the type (e.g. string, continuous, categorical). In case of structured questions also the meaning of the answer options must be clear. A codebook – which you preferably should have developed in order to facilitate data-entry – will guide me in this process.

2  Data-cleaning

In order to check whether the data has been correctly entered into the dataset, some checks need to be conducted e.g. by making frequency tables and conducting cross-tabulations. That way you can check whether there are any values that are missing, out of range (e.g. age of 208), or otherwise not correct (e.g. date of birth after date of interview; male participant indicating to be pregnant at the moment). If needed, mistakes should be checked with the original data.

3  Questions to be answered

The following step is to clarify the questions you want answered with the dataset. Do you really have the correct data available? If so, a plan for the actual data-analysis is developed. Otherwise, we might need to revise the questions.

4  Actual data-analysis

The statistical data analysis can either be only descriptive (frequency distributions, measures of central tendency e.g. mean and variability e.g. standard deviation) or analytical as well (e.g. measures of association like relative risk (RR) and odds ratio (OR), t-test, chi-square test, ANOVA, logistic and linear regression). Analyses will be conducted in e.g. SPSS.

5  Reporting on the results

The actual results of the analyses, as given by the statistical programme, will be reported (e.g. p-value of a test, RR/OR) and in addition a description will be given of what this actually means in the context of the study. Furthermore, relevant tables and graphs/charts can be compiled. This information forms the basis of the Results section of a report/article. If needed Epi Result can assist with the writing of your publication.

If I conducted the statistical data analysis for you, you will get the final dataset and the syntax when the analyses have been finalized. Please contact me to discuss your own needs for statistical data analysis.

Recent projects

• Analysing data on diarrhoea incidence in rural communities with or without water interventions in Limpopo Province.
  Customer: Tshwane University of Technology
  Reference: Paul Jagals
• Analysing data to validate verbal autopsies in the Agincourt Demographic and Health Surveillance system.
  Customer: Wits University
  Reference: Kathleen Kahn

Please contact me to discuss your own needs regarding statistical data analysis.