6. RESEARCH DESIGN
6.7 Data analysis methods
Like other activities of the research, data analysis was also done in accordance with the multi-staged sequential mixed method design. The data collected through the first phase of the quantitative survey was analysed first before proceeding further as it guided the operationalization of the second phase. Later, the findings of the qualitative interviews were also analysed and subsequently, the findings of both components were integrated to present the conclusions of this study. The following sections describe the details of the analysis methods used for each component.
6.7.1 Quantitative survey
All the data collected through online survey was directly exported to the softwares of SPSS and MS-Excel as the LimeSurvey software provides compatibility mode features with these data analysis softwares. The very first step before doing any kind of analysis was data editing and processing to check for any mistakes which might be due to misinterpretation of any questions, maintaining uniformity in the data and preparing the data for analysis purposes.
5 Moving Picture Layer-3 audio file format
Chapter 6 Research Design
84 Data editing and processing
The very first thing to start with for data editing and processing was to look for responses by non-qualifying respondents. The questionnaire design ensured that the non-qualifying respondents were excluded from the questionnaire on the basis of unmatched migration background with the established qualification criteria. But the questionnaire was unable to filter out those respondents who entered Germany after 2008. I noted eleven such cases in the data and hence, disqualified them and excluded their entries from any analysis. Further, there were 50 partial data entries where respondents interrupted the process of responding and hence, they were also excluded from the analysis part.
Many of the questions of the online questionnaire were designed in simple ‘yes/no’
questions, ‘multiple-choice’ questions and in a ‘list’ format where respondents were required to select the relevant options according to their situations. The entries against these questions did not contain any errors except the issue of non-responsiveness and resulted in the problem of missing values for some cases. Only a few of the questions from the questionnaire were mandatory, while all other questions were optional. Due to this, the issue of non-responsiveness and consequently missing values was pertinent for all such questions. A few errors and inconsistencies were also noted with questions where respondents were required to make text entries. This happened in some cases because the respondents either misinterpreted such questions or they did not follow the given guidelines about the format of data entry. These issues required data editing and labelling for correction before doing actual analyses in the SPSS environment. All the missing values and wrong entries in some cases were properly labelled as ‘no response’ and ‘irrelevant response’. Further, there were many loops which were designed in the questionnaire which meant that not every question was applicable to every respondent. To differentiate with the
‘no response’ issue, all such missing values were labelled as ‘not applicable’. After this was all the structured quantitative data edited and processed in the SPSS environment.
The online questionnaire also included an unstructured part which generated qualitative data. This was a part which inquired the respondent’s experiences of the time spent in Germany during the global economic crisis period. The respondents were given a
‘huge free text’ space where they were able to write as much as they wished to share their experiences. But this part was kept optional and consequently only 115 out of total 188 eligible respondents replied. These unstructured responses were ‘coded’ in MS-Excel to categorize mentioned aspects of affectedness faced by them during the crisis time, if there were any.
Data analysis
The structured and semi-structured datasets were statistically analysed by simple frequency distribution and descriptive measures. The codified unstructured data, after summarization gave an impression of either affectedness or unaffectedness to the respondents. Further, the respondents who were found affected by the outcomes of the economic crisis, a scale for affectedness i.e. ‘level of affectedness’ was also assigned to them, keeping in view the severity of affectedness mentioned by the respondents and by an established criterion (Table 7.3) . These findings of qualitative data were then inserted into the SPSS environment by adding the variables of ‘affectedness’ and ‘level of affectedness’ for participating respondents. This dataset was then integrated with the data of return migrations intentions
85 of the respondents and the reasons mentioned by them for those developed return migration intentions. That integrated data was then used to develop a typology for the respondents. Lastly, a ‘Chi-square test’ was performed to establish a relationship between the affectedness of the respondents and the developed return migration intentions of the respondents (details are in the Chapter 8).
6.7.2 Qualitative interviews
As mentioned earlier, the qualitative inquiry was a multi-purpose part of this research. The analysis of data generated through this activity also required a scientific procedure to follow to reach at some conclusions and to keep the reliability component intact. For these reasons, qualitative interviews were first transcribed before making actual analysis.
Data transcription
In simple terms, transcription is the transferring of an audio or video recording file into a textual form (Dresing et al., 2012: 14). A transcript is usually generated by manually typing the recorded material (ibid.). The main aim of transcription is to keep a record of verbal conversation during interviews, converting it in written form to make analysis easier and faster, and to bring transparency and reliability to the research (ibid.). The main goal of transcription is to grasp the recorded situation and the verbal conversation as well as possible so that a reader could interpret the presented data correctly (ibid.: 15). For that reason, there is always a need to have some ‘transcription rules’ to guide the readers. The transcription rules depend on how one wants to transcribe the data i.e. simple transcription or detailed transcription. The focus of simple transcription lies with readability while the detailed transcription is required where the focus is not on surface content of a conversation (ibid.: 16-17). I selected simple transcription method as it was suitable with the aim of this content oriented research and the available time. So I prepared ‘simple transcription rules’ in accordance with the transcription guidelines provided by Dresing et al.
(op. cit.: 20-25) for transcribing the conducted interviews (see Appendix 10). I followed the transcription rules in order to maintain consistency in the transcripts and to disseminate the verbal conversation of interviews in a coherent manner. Sample transcripts for every broad identified type of the interviewed respondents are placed at Appendix 11.
According to Dresing and Pehl (2010: 76, as referred to by Dresing et al., 2012: 31), the average time required for transcribing one hour of an interview is five to ten hours with simple transcription rules means a ratio of 1:5-10 (interview time: transcription time). But for this research, it was not the simple transcription, it also required translation into the English language at the same time. For me, that ratio worked out as 1:12-15 depending on the volume of the interview content. Further, Dresing et al. (op. cit.: 33) also noted that a transcriber can only work efficiently 4-6 hours per day. Appendix 7 provides details of the time spent on various data collection and analysis activities including the transcription activity.
Data analysis
The findings of the first confirmatory part were used to triangulate the established broad and detailed typologies of the selected sub-sample of the respondents. There were only two
Chapter 6 Research Design
86 cases (out of 20) that required a readjustment of the respondents’ type against the established broad typology. Though more variations across detailed typology were revealed, it did not affect the relationship between the affectedness and the developed return migration intentions as it was established across the broad typology of the respondents. In this way, both established typologies were readjusted in light with the findings of the qualitative part.
The second part inquired about the performed translocal practices of the Pakistani immigrants. This part was comprised of questions which were arranged in a checklist format. All the possible practices in the selected markets/sectors were asked one by one for which the respondents were required to respond in either ‘yes’ or ‘no’. This data was analysed by simple frequencies against each translocal practice performed by the two identified categories of the Pakistani immigrants, i.e. those who had return migration intentions during the economic crisis time and those who had not such intentions. I compared the performed practices with each other across the two identified categories of the respondents to find the differences. There was little difference found in the performed practices by each category, except labour market interactions which were explicitly exercised by those respondents who had return migration intentions during the global economic crisis period.
The next part dealt with an unstructured inquiry which asked about the motivations behind the performed practices. Every respondent was asked about the reasons for performing various activities which were disclosed by them in the previous section.
Analysing this part was a challenge. Though respondents mentioned some specific reasons for their performed practices, it became clear that in many cases, there was not a single reason for performing a practice. Also, it was a difficult task to differentiate between motivational reasons behind various performed practices. I developed a criterion for fixing the main reason behind the performed practices (details are in Chapter nine). I mainly looked for those practices which were explicitly performed under return migration intentions during the global economic crisis period. The data analysis enabled me to find some of the practices which were explicitly performed under return migration intentions due to the affectedness by the outcomes of the global economic crisis. Those were the practices which could not be explained by the theories of livelihood strategies and transnationality alone.
The last part inquired about the implications of the performed practices for the urban development sector. The main challenge was to address the varying scale at which the practices were being performed. Also, another problem was to address the contradictory implicational aspects of the performed practices and their impact on urban development. I analysed the implicational aspects of the carried out practices in accordance with the UN-Habitat urban indicators and devised an Urban Development Index (UDI) to assess the collective impacts of the carried out practices on the urban development in the origin places. Again, I looked for the implications of particularly those practices which were performed under return migration intentions and found that those practices constituted a significant share of the impact produced by all the performed practices (details are given in Chapter ten).