Journal reflections
4.3 Collaging quantitative and qualitative data
Extraction of data from the app while in use was explored (questions and tracking, for example) and participants completed questionnaires containing qualitative and quantitative questions immediately after the app walk. Paper questionnaires were used both for
convenience in the field and because of the physical pre-cognitive response needed by marking a line with a pen to answer some of the questions. Informal conversation and
semi-98 structured interviews at various stages of the test cycles were used to capture additional information that the questionnaire perhaps hadn’t elicited in order to probe deeper into underlying feelings. Using various methods of communication (local newspapers and
pamphlets, local BBC radio, email networks, social media such as Facebook, word of mouth), as wide a range of participants as feasible was drawn including those from the local
community, artists, academics, heritage volunteers, experts, students and the general public, to encourage a mix of experience, perspectives and feedback.
Figure 16: Social media and traditional press helped attract volunteers to try the app. The articles are from The Cornishman newspaper based in Penzance and the St. Ives/ Hayle Times and Echo.
There are existing overlaps between quantitative and qualitative methods. In questionnaires, when responding to qualitative questions, answers often contain
quantitative statements such as ‘I liked it a lot’. If questions are rephrased, these answers become measureable and can provide quantitative data. An advantage of using quantitative methods is how well they lend themselves to statistical analysis, which is used to evaluate whether findings have come about by pure chance or to identify when there is underlying systematic evidence. New insights may emerge from statistically supported significant differences. When I compared my app to an MP3 walking tour, for example, I looked at the differences between evaluations from the app and the MP3 walk to see if there was a systematic difference between them.
A level of chance in how a question was answered needs to be accounted for. The level of chance is called probability, but referred to when discussing findings in this research as p-value. To be statistically significant the probability has to be lower than 5%, which means that chance couldn’t have interfered with the result more than 5%. In this thesis, the 5%
probability will be referred to as .05. In order to be significant, the p-value needs to be equal or smaller than .05.
The graphic rating scale (Stone et al 1974) is the quantitative data-gathering method used in this research. To answer a question in the questionnaire, the participant marks a line that uses words at the extremes of the scale rather than numbers, for example:
Figure 17: The graphic rating scale uses words rather than a scale of numbers. The participant marks a line between the extremes of the scale spontaneously without wondering whether, in this case, the level of
ease of use was 7,8 or 9.
100 By marking the line spontaneously, rather than having to think about whether the level of enjoyment is a 6 or 7, appeals to intuition. The spontaneous response, I believe, is more suited to quantitative research in the arts, as it does not require cognitive interference but, instead, a feeling. Numbers do not appear on the line but it measures 10 centimetres, or 101 units, as marks are measured to the millimetre. Once the form has been returned the researcher measures the placement of each mark on the line (for example 6.3), which is logged and then analysed using SPSS (Statistical Package for the Social Sciences, IBM) software. The graphic rating scale was also used in relevant locative media empirical
research on Riot! 1831. Using elements of the same quantitative data collection and analysis method could, if needed, make comparison between the two studies easier. Comparison with this particular study may be relevant because it’s the only other study on locative media nonlinear narrative with significant empirical evidence and a focus on immersion.
Collecting quantitative data is rare in practice-based research (Nelson 2013: 99) and the arts in general and seems to be increasingly rare in cultural geographies’ research, which leans more and more towards creative work as a research output (Hawkins 2015: 17). Collaging of quantitative and qualitative data is another way of describing a mixed-method approach.
Good practice complements or challenges quantitative results with qualitative data and vice versa. In this research, significant results in quantitative data are illustrated and deepened through insight gained from qualitative answers and interview quotes and vice versa.
Quantitative data are used here as a check against subjectivity, since my position within the practice-based research is not neutral but has an agenda (Minchinton 2002: 4). As the app creator, a member of the community under scrutiny and the researcher, inclusion of quantitative data retains objectivity and others’ perspectives, thereby decreasing any influence of my subjective interpretation and what I want from the research. Whether collaging personal reflections from practice, subjective comments of others and deep systematic analysis of quantitative data will reveal new insight, or simply add weight to the conclusions, is of interest.
Numbers from quantitative data, once analysed, are turned into a coherent story that is disseminated through words. By overlapping both types of data-gathering the method of collecting evaluations, as well as my own question design and approach, can be
interrogated, especially if disparities appear. Did I ask the right questions? Does the quantitative data support my interpretation of qualitative answers?
4.3.1 Managing data
Upon completion of the questionnaires all marks on the graphic rating scales were measured and qualitative answers written and entered into an Excel spreadsheet – one document for each test cycle. Names were coded and anonymised in a way that indicates gender (M1 and F1 for example). The key to the anonymising code is kept on my personal laptop. The scores of the marks on the graphic rating scale were processed using SPSS (Statistical Package for the Social Sciences, IBM) to examine correlations between answers.
ANOVA software packages were used to analyse variance to explore differences using scale data. A scale of 0-100 was used, which includes looking at more than one variable at a time, allowing for more powerful analysis than nominal data (8 out of 10 men prefer….) and ordinal data (Group A prefers musicals….). Through analysis of variance, answers were grouped by how highly they were rated by participants (usually on the scale between ‘not at all’ and ‘ very much’) while showing whether the participants were in close agreement with each other in each answer.
Qualitative answers in each of the three evaluations were sorted into themes identified in the contextual review: qualities of locative media, immersion, embodiment, and the combination of immersion and embodiment in landscape. Other answers were gathered into emerging themes. Anonymous data gathered through the app, such as comments or tracking coordinates, were collated and compared to emerging themes.
Analysis of and reflection on quantitative and qualitative data from the evaluations
informed changes to the locative media work, the app. These changes ranged from practical changes to help usability (for example, moving GPS points or adding features to the
interface such as a replay button) as well as creative changes (for example, editing down stories), and adaptations to draw out more data using the conceptual framework or to explore emerging themes further in the next test (for example, designing in mild immersion disruptions).
Throughout the research process participants could test the app and ask to engage with the evaluation and results. Emerging data and insights have been and are still discussed and
102 disseminated at transdisciplinary conferences, seminars and practical workshops on
landscape art, landscape research, heritage, digital futures, archive and digital media, geography, geopoetics, walking arts, Practice as Research, digital narrative, site-specific work, writing communities and so on. My AAG (Association of American Geographers) paper within the ‘Reanimating Region’ strand (Florida, April 2014) will be published as a book chapter (co-authored by Erik Geelhoed and Misha Myers) in December 2016 by Routledge, edited by James Riding and Martin Jones.