Video analysis of peer relationships: rich data, a lot of work

Video makes it possible to perform very precise conversational and behavioural analyses. As technology keeps getting better and easier to use, more and more research is done this way. The WORP meets every three months to exchange experiences. Hoeben: 'We are a multidisciplinary working group with educationalists, psychologists, criminologists and sociologists from - currently - five different research institutes (EUR, UU, UvA, VU and NSCR). Other areas of expertise are of course very welcome.'

What is the advantage of video analysis, for example in your research?

'I research how adolescents encourage and persuade each other to engage in risky behaviour, such as substance use and crime. I do this on the basis of, among other things, video recordings made specifically for this project in secondary schools. During video analysis you can see and hear exactly what is said or done and what happens next. As a result, you know what the immediate cause is for certain behaviour. A questionnaire, for example, is always completed retrospectively. And afterwards people cannot remember word-for-word how a conversation went. They are also usually unaware of non-verbal signals from conversation partners, which can influence their behaviour. But there are also other reasons to use video analysis: for example, two colleague workgroup members are conducting research into pre-school children, which you can hardly submit a questionnaire to.'

The camera often makes people self-aware, how do you ensure that the material is still authentic?

'The realization that the camera is rotating usually diminishes quickly. After five minutes, children often forget that it is there, and that is actually the case for most adults. Sometimes young people involve the camera in their conversation, which is nice to see. Then they dance for the camera or stick their tongue out. At the same time, their mutual behaviour - which is what matters to me - remains the same. They are focused on each other and possibly on the camera, instead of a third person who is observing.'

An enormous amount of data must be available from video material.

'Video analysis does indeed produce very rich data. The question is: how do you do justice to all that data? You are dealing with a chain of mutual reactions and behaviour. You have to code that properly in order to really be able to use it. In the working group, we are investigating how we can make full use of that data. We share experiences with drawing up coding schemes or we discuss the advantages and disadvantages of the different analysis methods. Sometimes we also exchange very practical information: how long does it take to train a new coder? Or we share literature about a particular method or analysis. We are still figuring out what we can learn from each other.'

Is video analysis the future?

'Certainly. But it still takes a lot of time. First you have to design a coding scheme and then you have to properly familiarize the coders. This is an important part, because all coders must interpret the same behaviour in the same way. All in all, this can take weeks to months and only then can you start the actual analysis of the data. I recently heard from a colleague who spent a day and a half coding a ten-minute video. Only for the non-verbal behaviour! I am also involved in research into the development of automatic detection software. This shows that in practice it is already difficult to recognize a person in a video, let alone that an algorithm can grasp complex communication patterns. Behaviour is difficult to capture in automated coding.'

Do you specialize in peer relationships (children, young people and adults) and do you conduct research using video analysis? Then you can register for the Workgroup Observational Research on Peers. Send an email to EHoeben@nscr.nl or visit LinkedIn.