Extract of a table for processing the collected data for week 2 of the participant 7 

Context in source publication

Context 1

... (to identify their situations of vulnerability), camera instrumentation (to observe the riders’ behaviour in the identified situation and allows conducting interviews) and interviews (to describe the subjective part of the riders’ activity and identify their problems). These tools were chosen because they permit to fill the questions of research of the present study and they are referred by Baldanzini et al. (2009) as fruitful method to conduct a naturalistic riding study. Throughout the experiment, the novice motorcyclists filled in a diary which was specifically designed for this study. In it they were asked to note down the situations they had experienced during their journeys which were or could have been problematic from the motorcyclist’s point of view. A considerable amount of work was done with the motorcyclists beforehand in order to explain the type of situations to be included in the diary (i.e. not just accident or near-accident situations but all the riding situations where they think that their performance, health, comfort, development or pleasure is or could be impaired) and reach agreement about the amount of detail in which these situations should be described. The participants were asked to record when these situations occurred (date, time, journey) and describe them briefly in words and if possible with a diagram (see Figure 1). Each participant was also asked to make a diary entry every day describing the situations encountered in as much detail as possible. Each motorcyclist was provided with a diary for each week the experiment ran. The main purpose of the diary was to reveal the situations of vulnerability encountered by the motorcyclists during the study period. The situations that were described provided a basis for the interviews that were conducted at the end of each week of monitoring. Audiovisual recordings were made with four cameras that were mounted on each motorcycle 3 . Two cameras were mounted on each side of the motorcycle’s rear top case, covering about 160 degrees of the front visual field (Figure 2). In order to mount the other two cameras, a wind deflector was specially developed for the study which was 10cm wider than the standard models. On this were mounted one camera pointing towards the scene to the front and another pointing towards the driver’s face. The data was stored on SD memory cards. The recorder was housed in the top case of each motorcycle (Figure 3). At the end of each week of the experiment, a face-to- face interview that lasted between 20 and 30 minutes depending on the motorcyclist’s availability was conducted at the participant’s home or place of work. This interview was in two parts: (1) an interview that focused on the description and accuracy of the situations of vulnerability reported in the diary, and (2) a self- confrontation interview based on the video footage of the identified situations. The interview also provided an opportunity to recover the data storage cards and replace them with empty ones. The term “self-confrontation” is used to describe a very wide variety of practices. The technique presented here was developed for CACD. This method provides a way of documenting in detail the subjective part of the participants’ experience and their immediate understanding of their behaviour when shown audio, video or other types of recordings of their actions. In our study, this technique provided a way of obtaining a step-by-step description of the motorcyclist’s actions while driving. The interview consisted of asking the motorcyclists to express their emotions and sensations, share their main concerns and interpretations, and explain their statements and actions during the driving situation when faced with the audiovisual recording of it. The interviewer’s questions focused on action in order for the motorcyclists to be able to put themselves back in the dynamic context of the experienced situation. The researcher’s input is systematically related to what the participant has just said or done during the interview or the situation on the screen and playback of the video footage is interrupted from time to time to give the participant time to speak. The researcher tried to intervene as little as possible so as to avoid triggering thought processes among the participants, so that the only focus was what was significant for the motorcyclist in the situation when the events occurred (Theureau, 2003). The interview was recorded with the same audiovisual device that was installed on the participant’s motorcycle, but only three cameras were used for the interview (Figure 4). The collected data was then loaded into processing tables that were based on the verbal protocols developed by Theureau (2003). The aim was to combine the different levels of data in order to obtain the most detailed description possible of the situation of vulnerability described by the motorcyclists. For the novice rider’s study, the tables consisted of three sections (Figure 5). The first contained a full retranscription of the data collected using the diaries (texts and diagrams). The second listed the remarks made by the researchers based on the video data that showed the context of the situation (behaviour of participants, infrastructure, traffic...). The third section contained verbatim retranscriptions of the participant’s verbalizations during the face-to-face interview that included the interview based on the diary and the self- confrontation interview). Data analysis consists in a macro-analysis to characterize the identified situations of vulnerability, and a micro-analysis to make an in-depth analysis of the driver’s activity in each situation. The analysis categories for a corpus of data of this type may be determined by applying a “ top-down ” model (deductive reasoning) and a “ bottom-up ” model (inductive reasoning). The categories are the outcome of a combination of research questions and a preliminary analysis of the collected data. Immersion in the empirical data provides a starting point for the development of analysis categories and a way of conserving an evidential link with the field data (Glaser and Strauss, 1967). The identification of the analysis categories and the corpus analysis for the novices involved a six-stage process that was conducted by two researchers: 1. Construction of a first version of the analysis categories based on an initial examination of the data and the research issues (4 major categories, 18 sub-categories). 2. Further work in order to develop a practical definition for each category and sub-category after a rigorous examination of a series of extracts from the corpus of data and identify a prototypical example drawn from the data. 3. Discussion and decisions about the validity of several of the categories. 4. Drawing up of the final version of the classification which contained three main categories (“the journey as a whole”, “the context of the situation”, “the motorcyclist’s internal dynamic”) and 15 sub-categories (for example in the case of the “context of the situation”: “Infrastructure”, “Types of interaction with another road users”, “State of the pavement”, “Meteorological conditions”, “Driving alone or with a passenger”, “Driving in a group”). 5. Coding of the corpus of data using the analysis categories by the two researchers working independently of each other. 6. Comparison of the results of the two coding operations and taking of a joint decision where there were differences of opinion. The goal is to analyze each of the identified situations of vulnerability accurately in “activity graphs” using the various types of data collected in order to have a comprehensive view of the activity of the riders in these situations. These “activity graph” present the motorcyclist’s actions during the observed situation on the basis of his verbalizations during the interviews, the rider’s actions observed on the videos (“Slows down” and “Takes bend”), photographs of the context and the time. The aim is to reconstruct the “film” of the situation on a step-by-step basis using all the different types of data that was collected. This way of presenting the data has the advantage of retaining the fundamental aspects of human activity: dynamic aspects (displaying the change in the situation over time, the sequence of actions performed by the motorcyclist, etc.) and integrated aspects (links between the different dimensions of the activity by combining different levels of data). This presentation highlights the constant links between observation of the context and actions, as well as between the driver’s actions and emotions. In addition, the “activity graphs” provide a comprehensive summary of the participant’s actions in the studied situation. With this kind of time-based modelling, we are able to understand the dynamics of activity and the difficulties experienced by the drivers without omitting the situational aspects. The first example we shall examine, involves a situation that occurred on 2 October 2011 at 18:22 during a home-to-work journey made by the novice riders number 3. This situation was identified from the following diary extract: The data given in Figure 6 gives us a better understanding of this situation. The second example involves a situation experienced by participant number 2 on 18 October 2011 at 18:57 during a leisure journey: The motorcyclist’s activity graph for this situation is shown in Figure 7. The approach described in this paper provides a way of (1) characterizing the situations of vulnerability encountered by drivers – especially by the macro- analysis, and (2) identifying the skills the actors lack in the situations in question in order to incorporate them into training – especially by the micro-analysis. Statistical processing can be carried out in order to obtain a clearer picture of the contexts in which the situations of vulnerability arose. This processing was able to identify the ...