Chapter 4 Methodology
4.8 Data Analysis
The following section presents the mixed-data analysis procedures used for the qualitative and quantitative data, including ethical considerations and validity.
4.8.1 Quantitative Data Analysis
The full raw survey results could be downloaded at any time from a secure and licensed account held at the Survey Monkey Website, https://www.surveymonkey.com/s/C6RF3FH. Once I had accessed and downloaded the survey results, ranging from the survey’s initial ‘live’ release in February, 2013 to the most recent in January, 2014, I filed and organised the .xls spreadsheets in my computer. I filed the digital survey data in various forms, organised by months, depending on how the information was cleaned for various inquiries. I analysed the statistical results from the 252 teacher educators who completed the Teacher Educator Survey using SPSS and EXCEL software. Chapters 5–6 detail the results and validity of the statistics in full.
One of the members of staff of the School of Education at the University of Glasgow, Dr Muir Huston, and I were the only two people to access the survey data. The results were described in two ways: an evidence description and a description based on a comparative analysis between the population sample analysis and the analysis of three samples (University-based, College-based and Teacher training school-based) from the population sample. I also focused on the key research questions and their correlative factors in the questionnaire. Some assumptions were made and tested using SPSS tests (e. g. T-test, One- Way ANOVA). For example, I wanted to find out if that the choices for professional learning of teacher educators in Shanghai had been strongly influenced by their length of time as a teacher educator. In addition, correlation analysis was used to understand the relationships between the two variables; for example, the relationships between professional attraction and retention, and gender, age and educational background.
102
4.8.2 Qualitative Data Analysis
After digitally audio-recording and transcribing the interviews, I sorted, coded, and categorised the qualitative data in a ‘systematic and meaningful way’ (Brantlinger et al., 2005). This section outlines the analysis procedures in detail.
I recorded all fourteen interviews on a handheld, Sony Digital Voice Recorder ICD- PX312M device. Each file was saved on the device as an mp3 file. I downloaded the audio files onto my laptop. This laptop was stored in a code case with my own password-lock. I edited the files as needed. At this point, each interview file was ready for transcription.
I listened to each audio file through Windows Media Player, which was installed onto my laptop, using headphones, and typed the interviews into a Microsoft Word document in Chinese initially. I transcribed each interview almost word for word, but ignored all the ‘ums’ and background noises. If an interviewee mentioned any personal information or proper names/locations, I coded it with an ‘XX’ and timestamp. I referred to the interviewees by their codes (e.g. FG01a) and myself as ‘Q’ throughout the transcriptions. After the completion of all transcriptions, I listened to the audio files again, and amended the transcripts that I had made, ensuring that I had not missed anything from the audio files (Interview transcripts can be seen in Appendix M, see p.241).
I translated all the transcripts from Chinese into English, again word for word11. To get
them right, I discussed my interview transcriptions with a native speaker during our regular English conversations that had lasted more than a year. To further make sure, I employed a member of the Society of Editors and Proofreaders in Glasgow to proofread the completed translations, and discussed the interview transcripts through emails to ensure as accurate a translation of meaning as possible although the limitations of translation from Chinese to English are acknowledged.
I kept two data files for each participant: a digital file folder and hardcopy version of those files. All email correspondence related to this research was kept in, and accessed from, my laptop account. The digital file folders were stored and frequently backed-up on the following password protected devices: my personal laptop computer, a portable 16 GB Sony thumb drive, one 60 GB DELL external hard drive locked in a code case located at
103
my home. An example of the digital file folder for Participant II01 would contain the following: II01_interview.mp3; II01_transcript (in Chinese).doc; II01_transcript (in English).doc; the relevant file folders for the participants would contain the following: signed consent form, notes about times and dates for interviews, maps, and other travel documents. This information was stored securely at my home or my office.
In terms of coding of the data, at first, I carefully read each of the three policymaker interview transcripts and pencilled notes comprehensively on each, ‘to obtain a general sense of the information and to reflect on its overall meaning’ (Creswell, 2003: 191). Secondly, I opened a doc file for each interviewee and systematically coded my notes, selected quoted material into categories and labelled those categories. At times, I used terminology based on the language of the participant, known as in vivo terms. At the same time, I wrote memos when I had some thoughts on specific points. Next, I began the final process of theme identification to identify recurring topics or characteristics of professional roles and responsibilities and professional development of teacher educators (An example of coded transcript is attached as Appendix L, see p.239). In general, I followed the Miles and Huberman (1994: 10) comparative analysis matrix for analysis of the data. They define three ‘flows of activity’ in data analysis as: ‘data reduction, data display, and conclusion- drawing/verification’. At the ‘data reduction’ stage, the interview transcripts were reduced and organised by coding, writing summaries, and discarding irrelevant data – but I still ensured that I had access to it later if required to re-examine some unexpected findings that might have been previously considered unnecessary. At the ‘data display’ stage, I drew conclusions from the data in the form of tables, charts and networks. At the ‘conclusion drawing/verification’ stage, my analysis allowed me to begin to develop conclusions regarding my study. These initial conclusions were then verified, that is, their validity was examined through reference to my existing field notes.
4.8.3 Analytical Synthesis and Integration
The results of the independent analyses discussed above were combined at the interpretation stage of the research. The integrated data analysis strategies involved using analytic techniques for integrating the results and for assessing whether the results from the two databases were congruent or divergent (Creswell and Clark, 2011a: 223), or complementary (Pearson et al., 2014). For example, I had a quantitative database, which
104
was produced through a survey on the platform Survey Monkey, and a qualitative database which was collected by means of both focus group interviews and in-depth interviews. I analysed the findings produced from one of these two databases and then compared them with the corresponding findings on the other database to determine whether they were divergent, congruent or complimentary. If they were divergent, I would further analyse the data from either Survey Monkey or the interviews. In this study, the strategies for integrating the two databases mainly involved a comparison of the results through two methods, namely: (a) side-by-side comparisons in the discussion, and (b) data transformation in the results (transforming one type of data into the other type of data).
The first strategy for integrating data is a side-by-side comparison of both the quantitative and the qualitative findings in order to allow for synthesised data analysis which involves presenting the two types of results together in the form of a discussion so that they can be easily compared. The presentation then becomes the means for conveying the merged findings (Creswell and Clark, 2011a: 223). In this study, I first presented the quantitative results from the interviews followed by the qualitative findings from the survey in the form of quotes in the discussion section. My comments then followed and specified how the qualitative quotes either agreed or disagreed with the quantitative results.
This study first and foremost adopted the first strategy (side-by-side comparison) to integrate the quantitative and qualitative databases and complemented the integration through the means of the second strategy, namely data transformation merged analysis. In this sort of integration, the researcher transforms one type of data into the other type so that both databases can be compared (Creswell and Clark, 2011a: 224). In my study, in some cases, I transformed the qualitative results into quantitative results which involved reducing the themes to numeric information. For example, I counted the frequencies of the comparative adjectives that were used by the teacher educators in relation to the degree of professional knowledge they needed in comparison to school teachers (See p.119).