VALIDATION: The key to experimental repeatability and a sound scientific publication
Whenever I write a paper, review a paper or edit a paper, a key item that I look for is whether the methodology and equipment has been validated, demonstrating the effectiveness (accuracy and reliability) of the research. Without validation, we risk presenting erroneous results, which could be detrimental to the interpretation.
My PhD supervisor drilled “validation” into me, to the extent that it features as a keyword in one of my early papers back in 2006! To be honest, it was not something I had considered before, and it was very disheartening to learn only then, following 15 years of data collection for a non-governmental conservation organisation. I suddenly realised that the data I had contributed to collecting, along with hundreds of other volunteers, was highly flawed, as the equipment, methods, and surveyors had never been validated, leaving huge gaps in data continuity and reliability.
If you use thermometers (or any measuring equipment), you must check how much each instrument differs to the standard at the start and end (even regularly in the middle if long-term) of an experiment. If you use data loggers, you must check the variation among loggers in use, as well as calibrate drift. Tape measures can stretch, so you must check their accuracy at the start and end of an experiment. If you “visually” assess something, it must be repeatable, so you must validate the same result can be consistently obtained by you. If more than one person collects field data, it is essential to validate, at regular intervals, that everyone is collecting data in the same way – otherwise, certain items are recorded more or less frequently as different people focus on providing more or less detail, respectively. A classic example from my past is recording sea turtle egg cases following hatching – over a 15 year period the method changed 3 times, with major variation in accuracy. If several people participate in a visual assessment, you must validate their ability to interpret an item in the same way, for instance inferred animal body length from underwater observations, the distance of an item, or taste and colour differences in fruit or other food stuff. If you decide to select certain parts of data, you must validate the criteria used, such as the start and end of an animal dive, or even the number of dive bouts that are left out on either side, leading into and out of a repeated behaviour.
I think you get my point – the list is endless. Here, I presented examples that should never be overlooked, but it extends to theoretical biology, experimental modelling, GIS bandwidths, etc. Thus, when compiling a paper, it is essential to take time to work through the methods content carefully, checking for any items that require some form of validation and ensuring that these items are addressed prior to journal submission. It is worth making this effort, as it strengthens the reliability of the method and, hence, interpretation of the results, and may even make the difference between a manuscript being accepted or rejected.
Learn more about ScienceDocs Editor Dr. Schofield
