Chapter 3 Literature Pertaining to the Methods
3.7 Data Analysis of Interview data
Data analysis is a process of organising and providing structure to and extracting meaning from research data (Polit and Beck, 2004). The purpose is the same whether the research data are qualitative or quantitative. However, the qualitative data analysis process is different from the quantitative and is very challenging. Firstly, the lack of universal rules for qualitative data analysis and presentation make it difficult to explain how to do data analysis and how to present findings (Polit and Beck, 2004). Secondly, qualitative data analysis requires the researcher to undertake a great amount of work because of the number of pages of data that need to be organised, integrated and interpreted. The last challenge is to condense the data for reporting the findings which should still maintain the richness and value of the original data (Polit and Beck, 2004). In addition, selection of data analysis methods is essential to establish the study rigor and it should be based on the data collection methods used in the study (Polit and Beck, 2004). Thus, it is crucial to select the most appropriate data analysis method and present the process of data analysis for establishing the study rigor. There are a variety of analysis procedures based on different traditions and paradigms (Burns and Grove, 2005). However, some studies are not based on a formal tradition or paradigm, or are not characterised and described in a particular
disciplinary framework (Polit and Beck, 2004). With these cases, the data analysis methods are often stated as content analysis or qualitative content analysis (Polit and Beck, 2004). Although the researcher decided to apply a case study approach, this study was difficult to characterise or describe as belonging to one tradition or discipline. In this study qualitative content data analysis was applied to the interview data to understand the meaning of the context. In the following paragraphs, the conceptual framework for qualitative content analysis is discussed.
3.7.1 Background of Content Analysis
Even though content analysis has appeared in English literature for only about 70 years, the history of content analysis can be traced back to in 17th century (Krippendorff, 2004). It has been widely used in health care studies, and more than 4,000 articles have been published with content analysis as a subject heading within the last 20 years (Hsieh and Shannon, 2005). Initially, researchers used content analysis as qualitative and quantitative research approaches in their studies. However, later on content analysis was used in more quantitative research such as media research and newspapers (Krippendorff, 2004). Within this quantitative content analysis, text data were coded in explicit categories and reported with statistics. This approach was often referred to as quantitative analysis of qualitative data. More recently, the position of content analysis as qualitative data analysis has been established especially in nursing studies.
The distinguishing feature of content analysis, both quantitative and qualitative, is the use of a consistent set of codes to classify data segments that contain similar meanings and the goal of content analysis is to provide knowledge and understanding of the phenomenon (Morgan, 1993). Although some researchers regard content analysis as a quantitative analysis approach because it contains counting, many qualitative researchers who conduct studies which claim no particular disciplinary tradition, simply state the data analysis approach method as qualitative content analysis or content analysis because of its flexibility in study design and content sensitive methods (Elo and Kyngas, 2008). For these reasons, content analysis has been widely accepted and established in health research and nursing studies (Elo and Kyngas, 2008). However, it is often argued the lack of firm definition and procedures make the application of content analysis limited (Graneheim and Lundman, 2004; Hsieh and Shannon, 2005). For example, qualitative data can be interpreted in various ways and the understanding of the phenomenon can be dependent on the subjective inference (Graneheim and Lundman, 2004). In the next paragraph, the
concept of content analysis will be discussed to avoid any confusion or ambiguity for the purpose of understanding the data analysis procedure used in this study.
3.7.2 Different Approaches to Content Analysis
As previously stated, the common feature of quantitative and qualitative content analysis is that both approaches use a consistent set of codes to classify text data which contain similar meanings or concepts. The biggest differences between quantitative and qualitative content analysis are the coding procedure and the use of counting (Morgan, 1993). While generating codes in qualitative content analysis the researcher applies a more inductive way. The data were used as a source of their codes. When applying codes, the qualitative analysis is unlikely to use search algorithms or a pre-existing coding scheme which apply code automatically rather than relying on carefully reading original data and subjective coding. How to use counting is more important as a difference between qualitative and quantitative content analysis. In quantitative content analysis, the count of codes presents only a numeric summary of the data and further analytic steps are usually not taken forward, whereas, in qualitative content analysis, counting codes is regarded as just an initial step of interpreting the data and understanding the new contexts revealed by a coding and counting process. In qualitative content analysis, counting codes is a guide to direct the researcher to further interpretation of the data, and for the quantitative content analysis counting codes is usually treated as all that is needed to be known about the data. The following table shows the difference between qualitative and quantitative approaches in content analysis (Table 3-1) (Burla, Knierim, Barth, Liewald, Duetz and Abel, 2008).
Table 3-1: Quantitative and Qualitative Approaches in Content Analysis
1. Quantitative Content Analysis Qualitative Content Analysis Systematic and quantitative description of
text data
Testing hypothesis by statistical inference Focusing on the manifest contents
Mainly deductive category application
Systematic and rule-guided classification and description of text material
Qualitative and quantitative presentation for results
Focusing on manifest and latent contents Inductive category application
As mentioned before, there is no clear guide or universal rule to match a qualitative data analysis technique to a particular type of data or research goal. Thus, how to select an appropriate analytic technique and how to address the choice of them are key issues when analysing qualitative data. Morgan (1993) argued that qualitative content analysis is appropriate when the available data and the research purpose match the advantage of content analysis, which are describing and interpreting the context of the data (Morgan,
1993). For example, a study comparing the views or perspectives of two different samples is well matched to the strength of qualitative content analysis in terms of presenting the explicit answer to the question about what the differences are and also why the differences exist. Thus, in this study qualitative content analysis was applied.
3.7.3 Process of Qualitative Content Analysis
The lack of literature on meaning, concept, procedure and interpretation of qualitative content analysis has been addressed as a key issue (Graneheim and Lundman, 2004; Hsieh and Shannon, 2005; Elo and Kyngas, 2008). Hsieh and Shannon (2005) described 11 steps for conventional content analysis:
1. Reading all data repeatedly to obtain the sense of whole.
2. Highlight the exact word from the data that capture key thoughts or concepts.
3. Making notes of initial impressions or thoughts of the text.
4. Labels that are reflective more than one key thought will emerge.
5. Developing initial coding scheme from the labels emerged.
6. Code should be sorted into categories.
7. Grouping categories into meaningful clusters.
8. Organising subcategories into a bigger category.
9. Developing the definition of these categories, subcategories and codes.
10. Identifying the relationship between categories and subcategories if needed.
11. Addressing the relevant theory or other research finding if it is appropriate.
Although there are several approaches to qualitative content analysis, some of them take similar steps and share similar concepts. Elo and Kyngas described the process of qualitative content analysis with two different approaches; inductive and deductive. They
divided the process into three phases; preparation, organising and reporting. The initial step for the preparation phase is identifying the unit of analysis. The unit of analysis can be decided based on the purpose of the research and the source of data. Deciding on the depth of analysis is also important when identifying the unit of analysis, such as manifest or latent. In this phase, researchers are required to be familiar with the data and need to make sense of the data. Without this, data cannot be interpreted and no insights or theory will emerge. The next step for inductive content analysis is open coding, creating categories and abstractions, taking similar steps as conventional content analysis. Open coding is making notes and headings while you are reading the original data and creating categories with these headings and notes. During the process of creating categories, it has to be noted that creating categories is not simply bringing similar or related notes or headings together. They have to be classified as belonging to the particular category. Decisions have to be made, as well as some degree of interpretation, to which the unit of coding is then put in to the same category. After creating categories and grouping them in the bigger category, a general description of the research topic should be developed, and this process is called abstraction. Deductive content analysis shares the same concept as directed content analysis in the purpose of the data analysis and using the predetermined coding scheme developed from the previous literature or existing theory.