Analyse

Get them most out of your data by optimising your data analysis! The following article aims at giving you an overview over important aspects to consider when analysing behavioural experiments.

If you are planning to conduct experiments in UUBF's facilities, please make sure to carefully read through the important information section on this website.

Caption "Analyse". Icon depicting the outline of a person with a thinking bubble containing an oscillogram.

Analyse experiments - Gain insights

1. Make sure your raw data is saved

Before you start with your analysis, make sure to create a copy of your raw data that will not be worked with during the analysis. That way, if anything unforeseen happens during the analysis process, you will always have an unaltered version of your raw data at hand to verify your analysis.

The same applies to experimental journals if you are planning to make manual annotations in these during the analysis process. In that case, make sure you are using copies or otherwise ensure that the original journal is unaltered and an additional, annotated copy is created.

2. Preprocessing of your data

This step can involve a large number of different individual measures depending on your particular experiment. It may involve:

Acquisition of your experiment in a tracking software
Manual scoring of behavioural experiments
Merging different data sources (for example time protocols of interventions with video recordings)
Bringing the raw data of different sources into the right format for your analysis
Annotating raw data with subject information
Ensuring blinding also during the analysis

3. Statistical analysis

The majority of studies will involve some type of statistics to be done. The details of which should be defined already during your planning stage to ensure that you can achieve meaningful results from the experiments you performed.

The details of your statistics will largely vary depending on your experimental design and hypothesis. If you need specific help with your statistical analysis, feel free to write an email to UUBF.

4. Document your analysis and findings

Make sure you are not only performing your statistical analysis, but also documenting the conclusions you are drawing from each test and possible steps leading up to each and following from each individual test performed. By doing this, you make reporting and controlling your analysis process and results easier.

Depending on the software you are using for performing your statistics, you have several options for documenting your analysis process, the conclusions you are drawing from it and your motivation for proceeding to each consecutive step.

Document your thoughts and process in your lab book or ELN and make sure to create a trace between your documentation and the actual analysis.
If you are using R for your analysis, you can use literate programming with RMarkdown or LaTeX and the packages Sweave and Knitr to create and design reports containing both information regarding your analysis as well as your data analysis scripts.

5. Additional experiments and repetitions

Depending on the results of your experiment, you might need to perform additional experiments or alter the setup of the experiment to gain additional insights. Furthermore, you might need to increase the number of animals used if the assumptions underlying the sample size calculation were not completely accurate. In several cases, this will warrant going back to the design and perform stages of your research to make the necessary alterations.