Data analysis workshop held to improve knowledge on statistical packages

August 19th, 2016

A data analysis training workshop was held in Nairobi from 15-18 August 2016. The training was facilitated by Prof. Julius Oyugi, the lead of the virology component, and was attended by the TDR/IDRC Team B’s four PhD students, one MSc. student, the data manager and five trainers (see Table 1).

The data produced by Team B is generating new insights into malaria and Rift Valley fever in Kenya. Having a thorough knowledge of statistical tests, and the software available, is therefore a highly important skill. The workshop aimed to do this – to help the students develop in-depth knowledge of statistical data analysis and gain insights in the application of different statistical packages in analysing their own data sets.

These highly interactive sessions involved individual and group activities, with discussions on individual research studies. This provided a platform for students to discuss any challenges they were facing, and for the specialists to share their knowledge and advice with the students. The data analysis training was tailored to suit the specific needs of the individual students. This was to ensure that the training covered all the necessary material specific to the varied data sets that had been generated within the different project components. Aspects of data analysis that were handled per component were as follows:

Eco-hydrology component

• Time series analysis of environmental and climatic factors
• Cross correlation between Normalized Difference Vegetation Index (NDVI) and climatic variables (rainfall and temperatures)
• Trend analysis using Mann Kendall and Seasonal Mann Kendall tests
• Modelling, by relating environmental factors (NDVI) with climatic variables (rainfall and temperatures) while taking into consideration seasonal effects (Dummy variables)
• Spatial and temporal analysis – creating contour plots of monthly climatic and environmental factors using Matlab and R

Entomology component

• Data cleaning, merging and coding using MS. Excel and SPSS
• Overview of Biodiversity analysis – diversity ordering techniques
• Analysis of adult mosquito house type preferences and comparison of indoor densities per house type (ANOVA with post hoc test)
• Modelling (GLM): Relating mosquito densities as a linear combination of climatic variables, environmental factors (NDVI), house type and eave sizes

Anthropology/Virology component

• Reviewed qualitative data analysis (FGDs/KII)
• Introduction to MaxQDA – a qualitative data management and analysis software
• PhD student together with the trainer briefed the Virology Masters student on how to analyse qualitative data using NVivo software

Modelling component

• Sample cross correlation between RVF vector abundance and climatic variables
• Time series regression analysis (dyn package in R): RVF vector abundance against climatic variables
• Binary logistic regression modelling; presence or absence of RVF against Lake Baringo water levels (2006-2007)
• Early Warning System (EWS) development

In conclusion, the training was successfully completed and, in general, the students gained valuable insights in:

• Descriptive statistics
• Inferential statistics
• Statistical tests (parametric and non-parametric)
• Modelling (LM, GLM, GLMM)
• Spatial and temporal analysis
• Statistical software – Matlab, maxQDA and SPSS
• Identifying good journals for publication – EDANZ search engines, impact factors, etc.

Table 1. Trainers, facilitators and students who attended the Data Analysis Training Workshop

UoN-University of Nairobi; JOOUST – Jaramogi Oginga Odinga University of Science and Technology; ICIPE – International Centre of Insect Physiology and Ecology; AKFEA – Aga-Khan Foundation East Africa