6.4 Between group differences in experience with disorders
7.1.2 Questionnaire development
To move towards developing an ESM application for measuring fear of crime, for the very first digital pilot of the study, the online survey-building application EpiCollect was used (Aanensen et al.) to create two separate questionnaires. EpiCollect was discussed briefly as a survey tool in Chapter 3, Section 3.4. To review, it is a free-to-use mobile application that allows researchers to create and deploy custom questionnaires to anyone with the EpiCollect application installed on their mobile phone. One questionnaire would take the approach of using the questions about fear of crime from the Crime Survey for England and Wales (CSEW) (formerly British Crime Survey). In the CSEW, the question: ’How worried are you about....?’ is repeated for various crime types (Office for National Statistics, 2014). For this pilot, the questions were only asked for three selected crime types: ’Being insulted or pestered by a stranger’, ’Being physically attacked by a stranger’, and ’Being robbed or mugged’ (see Figure 7.2b). Besides the questions about fear of crime, demographic questions were also included, asking people about their familiarity with the area, their gender, a general evaluation of their feeling of safety in that moment, and a free-text option where they could mention anything that made them feel worried (Figure 7.2a).
The second questionnaire took a different approach and instead asked participants to rate their feeling of safety in that moment from 1 to 9, and then use a free-text field to justify their answer. Further, it asked whether the person was alone or with company (Figure 7.2c).
Both forms had to ask participants to enter a unique name or nickname so that repeated answers from the same person could be linked. Further, both surveys also collected coordinates from the GPS of the mobile device, and the time stamp when the survey was submitted. A group of 5 University College London (UCL) Engineering students were asked to trial both versions of the survey, in two rounds of guided walks around the university campus, and stop-
7.1. Fear of crime application prototype development 163 (a) EpiCollect pilot 1 page 1 (b) EpiCollect pilot 1 page 2 (c) EpiCollect pilot 2 Figure 7.2: Piloting app questions using EpiCollect
ping at pre-selected locations to complete the questionnaire. The questionnaire based on the CSEW questions (modified to change the scope from ’In the past 12 months’ to ’In this moment’) was trialled first, followed by the second ques- tionnaire with rating safety on a numeric scale. After the walkabout, the pilot participants were invited to a focus group meeting to discuss their experience. The discussion covered their own experience, as well as what they thought that future challenges could be.
Three main issues emerged from the discussion. The first was that the questionnaire using the CSEW questions was tedious to complete multiple times, due to the repeated questions with the different crime types. Since all the different crime types were listed each time the participants completed the application, if they felt worried about one thing, they would have to write down what they were worried about, but then they would also have to answer ’Not at all worried’ to all the other crime types, before being able to submit the report. Pilot participants found this repetitive, and assumed it might discourage future participants from continuing in the study for longer amounts of time. However, they preferred the wording of the questions from the CSEW in terms of re- sponding, as opposed to the second questionnaire which asked them to rate their safety on a numeric scale of 1 to 9. As a result, a decision was made to use the CSEW question and answer wording, but to not repeat this for each crime type. Instead, if a participant answered that they were very or fairly wor- ried, a list of the possible crime types (all taken from the CSEW, but modified to make clear that I was asking about the current place and time) would be presented to them, to then choose to indicate what they were worried about. This also allowed for measuring fear of more types of crime than just the three pre-selected ones, without increasing the workload for participants.
The second issue which emerged was to do with the possible danger of someone putting themselves or their property at risk by being required to com- plete the questionnaire at a time when they felt unsafe. Or if they reported it later, after having moved to a safe place, the GPS location sent by the mo- bile phone with the report could be incorrect. If someone felt unsafe on their journey home, they would not be inclined to stop, report it, and then continue,
7.1. Fear of crime application prototype development 165 instead, they would want to get home as soon as possible, put themselves out of harms way, and report it later. In order for participants to be able to do this, there needed to be a way for them to enter the location of the incident to link to the report, as well as the time of the incident, rather than always sending their current location and the current time. As a result of this, a decision was made to include a retrospective annotation option, where participants can use a map to assign a location to their report, rather than always have to rely on their current location. This allows people to prioritise removing themselves from a dangerous situation first, but still be able to report about this location later, with accurate spatial and temporal data attached.
The third issue was similar to the first one of the repetitiveness of the ques- tions, but in addition to causing participants to become frustrated and poten- tially increasing attrition rates in a longitudinal study, this issue can cause prob- lems even if participants are willing to repeatedly participate. This issue was with asking people to enter their name or nickname repeatedly manually. By allowing non-validated entry of a unique identifier by participants, it becomes possible that multiple people pick the same name (for example if two people named ’Nick’ participate, they might both choose ’Nick’ as their username), and this wass not controlled for by Epicollect at the time of this trial. Further, since people have to type in their username repeatedly, typos become possible, re- sulting in multiple reports from the same person being counted as coming from two different people. To avoid this, a decision was made to use the phone or an existing account on the phone to automatically assign a unique user ID for each participant. This ID can then be used to link all reports to the same par- ticipant. As it is automatically assigned, there is no change for typos or other user error, and as it is based on a user ID previously created with a client which uses validation, the identifier is certain to be unique as well.
After the focus group, an initial look at the data also resulted in some changes made for the purposes of the analysis. While the focus group revealed that repeatedly filling out the same question with the same answers becomes annoying for participants, looking into the possible analyses that I would like to carry out with the data, it became evident that I would like to collect things like
demographic information about the participants. Therefore, since the unique ID would be able to link reports together, a decision was made to create a separate database with the unique ID and the demographic questions, from which the reports can later be linked back to the demographic information. This would allow the participant to complete these questions only once, but using the unique ID, this info can be linked back to all related reports of fear of crime made by that specific participant.
Finally, the free text fields were deemed to not provide much additional value, and it emerged from the focus group that people felt obligated to fill them out, which resulted in them feeling like the questionnaire took more effort than they would be willing to give to it repeatedly over the course of a longer study period. Since this overall is an initial feasibility study of using such an app to collect spatial and temporal information about experiencing fear of crime, the qualitative detail coming from such free-text boxes was left for future iterations of the application.
Based on the findings from these pilot studies, it was decided that a be- spoke mobile application would have to be built, incorporating the question from the CSEW. However, the question needed to be altered to reflect that the par- ticipants should respond about their feelings in the current time, and to ensure participants avoid having to complete it multiple times for all crime types. Fur- ther, this application would incorporate the mapping methodology from the par- ticipatory sticker mapping exercise in Subsection 7.1.1, to provide participants with an option to report something they had experienced retrospectively.