Humboldt-Universität zu Berlin - Musikwissenschaft

Structured observation: method application in the real world context.

Structured observation: method application in the real world context.

In order to study the behaviour unfolding as social interaction between two users of the Sentire system, structured observation will be applied as a primary research method. 

Structured observation is highly structured non-participatory observation used as a means for measuring behaviour through categories devised at the early stage of the research process. The observational categories constitute a coding scheme according to which the events recorded in the audio-video material under study can be detected and described in a systematic way.


If the categories fulfil some logical conditions in accordance with the representational measurement theory, structured observation can be counted as a measurement [1]. For this, they have to satisfy the requirements of a nominal scale. The Nominal scale is a measurement scale allowing to categorise different variables by labelling them. This requires a strict one-to-one correlation between a unit of behaviour that is identified and the label assigned to it. The categories also have to be mutually exclusive and exhaustive, meaning that only one code can be applied to each event and that all possibilities are covered, so it is always possible to attribute code to an event. In this context, it is appropriate to characterise structured observation as a quasi-quantitative research method. However, structured observations are preceded by a qualitative exploratory phase that aims at developing the coding scheme and contributes to informing the research question(s).


In the structured observation method, observers operate as measurement devices. Different from a qualitative approach, where the study fully depends on the observer’s interpretation, structured observation doesn’t rely on the observer and his or her qualities. However, the application of the method depends crucially on the skills of the observers. Hence, it is required that observers are trained in the method. Also, they have to be “calibrated”, which means that their agreement and reliability have to be measured. Cohen's kappa which corrects for chance is used for this purpose. 

In order to examine the effectiveness of the Sentire system in the applied field of non-clinical therapy and personal experience of those using it, real-world research will be conducted. This implies the engagement of various collaborators - experts in music and movement therapy among others - who will use Sentire at their own facilities with their clients. Through workshops and a user manual that will be developed along with the study, experts will obtain knowledge about the system and strategies for the interaction with the clients. The social interaction between expert and client mediated by sound feedback is the object of the study. The research question aims at investigating whether and how the system facilitates social interaction. Accordingly, the project team will study when and under what circumstances related patterns of behaviour occur.

Real world situations in which social interactions unfold are seen as an appropriate framework for tackling relevant problems and issues directly.  This non-experimental context that involves people with their problems and issues arising along with their daily experiences and interactions, implies multiple factors that might affect the course of the study. Due to this, it necessarily requires to pay careful attention to the study design that will guide all subsequent work while implementing the design. In these terms, structured observation with its fixed design and thorough systematic procedures is seen as an appropriate research method.

The first phase of the study which ends up using structured observation will be exploratory and the exploration will occur prior to the choice of a coding scheme [2]. Thus it is planned to start with the informal observation that includes note-taking and general gathering of information while observing the recordings. Elan software will be used as a tool for assigning annotations to the video content. 


Both subjects - the expert and the client - will be observed independently one after another. As a first step, observers will identify events within empirical (or physical) domain, e.g. “touch”, “upper body movement”. Second, the functional (or social) domain will be observed aiming at identifying the events at the level of the interaction, e.g. “initiation” or “synchrony”. Thus, the data gathered as empirical categories will be interpreted also as functional categories.  

[1] Seifert, Uwe, Kim, Jin Hyun (2008). Towards a Conceptual Framework and an Empirical Methodology in Research on Artistic Human-Computer and Human-Robot Interaction, In Ioannis Pavlidis (Ed.), Human Computer Interaction (pp. 177-194). Vienna: In-Tech 2008, p.188
[2] Robson, Colin 2002. Real World Research – A Resource for Social Scientists and Practitioner-Researchers. 2nd edition. Oxford: Blackwell, p.335