Humans as data sources!

I have recently begun collecting data for an experiment.  Data collection is the ‘bread and butter’ of science, without it there is no data, and therefore no results, conclusions or theories. While scientists can collect data from almost anything, as I am involved in behavioural science the data I require almost always comes from people; volunteers who agree to participate in an experiment. Using human participants (volunteers were previously referred to as ‘subjects’, but this term was dropped because it suggests that the volunteer is ‘subject’ to the experiment, rather than a willing participant) as your main data source produces additional (or at least different) problems to that presented from other data sources. I presume that in natural sciences, materials are ordered from a supplier, and therefore can be (hopefully) acquired to a predetermined timescale at a predictable cost. This is not the case with using participants, whose availability depends on the willingness of the local (normally student) population to submit to your study. Likewise whereas physical data-sources presumably perform reasonably consistently (i.e. putting the same quantity of lithium into the same quantity of water will always produce similar results, as long as other relevant variables are held constant) the same cannot be said for humans. The performance of two participants, tested under identical conditions, can vary drastically, even when the participants are from very similar backgrounds. Similarly an individual participant’s performance can vary widely during an experiment as concentration and motivation fluctuate. These factors produces a large amount of variance in the resulting data that is not due to the experimental manipulations the study is designed to investigate. The consequence of this is that the amount of data that needs to be collected in order to overcome such variance, and therefore provide a valid result, increases.

The variability in human performance also generates the further problem of generalisation. How can you be sure that the participants you have used in your study provide data that can be generalised to humans in general, given that individuals vary widely on how they perform the task? Larger samples (more data collection!) can make a sample more representative, but as undergraduates are usually the easiest source of data, inevitably most studies involving humans utilise samples that are non-representative of the general population to a greater or lesser extent. You could write an entire book on the issues around sampling and generalisation (indeed many have (1)) suffice to say that when you read any behavioural science research, especially that which is weighted towards the ‘social science’ end of the spectrum, it is worth considering the sort of people who may have participated in the research, and how that may effect the results that were found.

There are other, more basic problems with using humans as a data source.  Participants may fail to show up for the study, they may fail to understand what is required of them in ways that you couldn’t predict, they may even not take the experiment seriously, making little effort or deliberately producing nonsensical data. In physical science I suspect the main problem that can occur with an experiment is equipment failure. This is also a danger with behavioural experiments, but ‘participant failure’ is often a more pressing concern.

A final issue with using humans as a data source is that any study involving humans requires ethical approval, meaning that the research design is scrutinized by a committee prior to data collection for anything that might be deemed unacceptable. Ethical procedures are in place for a good reason, as in the past certain scientists were subjecting volunteers to all sorts of unpleasant and/or morally dubious procedures in the name of science (2). However perhaps inevitably ethical checks tend towards the cautious in terms of their application. While for many behavioural and social science research, ethical approval is merely a formality, it can restrict scientific enquiry for those of us that are interested in the facets of human behaviour that can only be evoked through manipulations of the participant’s emotional state or physical comfort.

So, given that I have just spent 700 words complaining about the problems of using humans as data sources,  why have I chosen a career path which relies so heavily on collecting data from humans? Well there are some advantages of performing research on humans. Most importantly humans are (to me at least) the most interesting subject in science. You can keep your chromatography, your mutagenesis and your particle accelerators, nothing they produce will ever be as interesting to me as investigations into human mind and behaviour. The variability in human performance which causes us so many problems is actually the main reason the subject of psychology is so interesting. A second advantage to behavioural research is that it allows you to meet a lot of different people who volunteer for your study for a variety of different reasons. The fact that certain people are prepared to give up their time and submit themselves to the often unpleasant or tedious tasks that make up your research project has helped reaffirm my faith in human nature after years of working in soul-destroying office jobs. Apart form anything else, the actual data collection part of a behavioural study certainly helps to break up a research process which would otherwise mainly consist of reading journal articles and staring at a matrix of numbers on a computer screen.

I’ll be coming to the end of the data collection process soon. I will then have weeks of grappling with the resultant data to look forward to!! As a final plea, if there are any men out there who fancy participating in my research then get in contact, as I still need a few human ‘data sources’ to complete my study!

(1) Rao (2000) Sampling methodologies with applications. Chapman & Hall
(2) See the early chapters of Naomi Klein’s book “The Shock Doctrine” (Penguin, 2008) for a description of some particularly unethical experiments performed in the US.

Rob Hoskin

Received a PhD from the Neuroscience Department of Sheffield University. Views expressed in blog posts do not necessarily represent the views of the Science Brainwaves organisation. https://twitter.com/Hoskin_R

Leave a Reply

Your email address will not be published. Required fields are marked *