Chapter 3 Methodology
3.4 Research design
This research study has been conducted within the pragmatic paradigm in order to benefit from the merits of both qualitative and quantitative data collection when addressing the research questions articulated above (Biesta, 2010).
Using a mixed-method approach benefits the current study in different ways. First, since this study sought to investigate perceptions and attitudes, both quantitative and qualitative data have been used to avoid any bias that could affect the validity of results by using only a single-method approach. As suggested by certain theo- rists (e.g. Creswell & Clark, 2011; Cohen,et al., 2011), the use of a mixed-method approach facilitates an in-depth conceptualization of perceptions and attitudes. Fur- thermore, it is often applied to surmount the challenges resulting from using one data type. In this study, relying solely on the quantitative data would not endow the research with sufficient data or the analytical subtlety required to wholly address the research questions. As Creswell and Clark assert, adopting a mixed-method ap- proach enables a more sophisticated understanding of the same phase of the study. A further advantage concerns triangulation, which enables the researcher to ‘test’ the
consistency of results (Cohen et al., 2007) through comparison and by drawing on the greater detail offered by two data set types: quantitative and qualitative. In the case of the current study, a sequential explanatory mixed-method design (Creswell and Clark, 2007) was adopted, with the quantitative results obtained from the online survey conducted in Phase One (see Figure 3.2).
Figure 3.2: Research Design for the Current Research Study (Adapted from Creswell,
2009)
The design of the mixed-method procedure selected for this study was influenced by four aspects, namely timing, weighting, mixing and theorizing (Creswell, 2009). Regarding the timing, as shown in Figure 3.2, the design used in the study is sequential, with the QUAN phase preceding the QUAL phase. The decision on weighting and mixing was based on the research goals and practical considerations around data collection. One of the drawbacks of the sequential explanatory design is that data collection can become a lengthy process; it was decided, therefore, to prioritise one form of data and consider it as the primary data set. In this study, therefore, the QUAN data was treated as the primary data set and the qualitative data collected via open-ended items in Phase One was quantified for the final data analysis. The qualitative data from Phase Two, while not being quantified, was used to provide any further explanation for the findings based on the quantitative data in Phase One. Finally, theorizing or transforming perspectives/lens is a factor in mixed method approach that determines ‘whether a larger theoretical perspective guides the entire design’ (Creswell, 2009; p.208). These transforming lenses could be ‘theories, frameworks and hunches [researchers bring] to their enquiries’ and help shaping ‘the types of questions asked, who participates in the study, how data are collected, and the implications made from the study’ (ibid). Such theorizing, Creswell adds, can remain implicit or can be made explicit. As for this study, the transforming lens−the programme evaluation framework proposed by Lynch (1996) −is made explicit.
One of the strengths of using this design is expanding the findings from Phase One by providing a more in-depth investigation of the issues in Phase Two (Morse, 1991,
as cited in Creswell, 2009). As previously stated, it was determined that a wholly quantitative or qualitative study would not adequately address the two main foci of this research, namely characteristics and evaluation of in-sessional academic En- glish provision. In the case of the evaluation of in-sessional for instance, only using QUAL research approach would make the study too ‘context-specific’ and resulting in employing unrepresentative participant (D¨ornyei, 2007, P.45). Using a mixed method approach, however, helped cancelling the sampling bias by informing the QUAL sampling by an initial representative survey (ibid). In addition, considering the complexity of the concept of effectiveness, ‘converging numeric trends from the quantitative data [obtained from the survey in Phase One] and specific details from qualitative data [obtained from follow up interviews in Phase Two]’ allow for ob- taining a better understanding of the phenomenon under investigation in this study (ibid). The use of a mixed-method design thus strengthened the research study by allowing more secure inferences to be made, while providing opportunities for the collection of a greater quantity of more diverse data (Tashakkori and Teddlie, 2003). Regarding the type of mixed-method design used in this research, the separate phases that make up the sequential explanatory design render the research descrip- tion, implementation and report more straightforward. Moreover, since the current study is conducted by a single researcher rather than a research team, the sequen- tial two-phase structure makes the data collection more feasible as the simultaneous collection of the quantitative and qualitative data requires the involvement of dif- ferent researchers and or research assistants who would then be responsible for the collection of each data set (Creswell and Clarks, 2011).
As mentioned above, since the two types of data are collected independently in a sequential mixed-method design, concerns may arise in terms of the length of time needed for data collection. One way to control such a potential limitation was to prioritise one data type. This meant that the quantitative data collected in Phase One informed the design of Phase Two, including the sampling and design of the interview items. Since the sampling for Phase Two was informed by the results of the data analysis in Phase One, purposive sampling techniques were used to identify participants for the interviews. According to this method, the selection criteria for the sample population was informed by the results obtained from the analysis of the quantitative data set. It should be noted that, according to purposive sampling techniques, the sampling criteria are more important than the sample size and, therefore, in the case of this study the same principles applied when determining the size of the sample population for Phase Two. The sample selection was, then, based on returning participants and the available resources for this study (e.g. time and
having a solo researcher). Further information on how the sampling was ultimately conducted for Phase Two is provided in section 3.7.1. Figure 3.3 summarises the data collection procedures for both Phase One and Phase Two and the way in which the two data sets were mixed, with a view to discussing and analysing the findings.
Figure 3.3: Data Collection Procedure Study