After the data have been cleaned, data analysis can begin. Depending on the method, specific procedures for identifying and explaining relationships between the study’s key variables should be outlined at the level of detail that would allow other researchers to replicate this study’s design (Swales & Feak, 2012). If there is more than one method to be employed, each can be presented under a subheading. If a method described in other studies is to be used, reference should be made to the relevant source(s), together with any necessary information on how this method will be modified to fit the current study.
In quantitative and mixed methods research, data analysis will most often involve the use of mathematical and statistical tools to establish and explain correlations between variables and/ or identify the patterns of association between these variables. Since statistical analysis of data (e.g., correlation and regression analyses) is typically done with the help of software statistical packages and other analytical tools (e.g., STATA, SPSS, SAS, etc.), these are often specified, and information on procedures for measuring the values assigned to given variables is provided. Specific ways to determine the statistical significance of the results are also explained. What is even more important, the rationale for employing these procedures to ensure the validity and reliability of the results should be given in connection to the study’s research questions or hypothesis. The level of detail at which this is done depends on the field and the method.
In qualitative research, the data analysis involves the researcher's subjective interpretation of concepts which are difficult to measure and evaluate (for example, attitudes, beliefs, and feelings about something). This kind of analysis is based on information which is in most cases presented as text - spoken or written (for example, interview transcripts, recollections, memos, etc.) and thus requires a careful definition and interpretation of the meaning of text. To ensure the reliability of these interpretations, procedures for analyzing qualitative data may have to be explained at greater length (Partridge & Starfield, 2007). This also implies that the units of data for coding are identified (for example, an artifact, a word, a group of words, such as a participant’s quote; the test scores of an individual student or a whole class of students, etc.) and explained. There can in fact be several units of analysis depending on the type of analyses employed in a single study.
Here are two unadapted extracts from the Methods section of two articles (not research proposals!) in the field of education. Note what kind of language the authors use to explain what the chosen data analysis methods and procedures are and to justify their use in each study.
| Openness to Theory and its Importance for Pre-Service Teachers’ Self-Efficacy, Emotions, and Classroom Behaviour in the Teaching Practicum |
|
Descriptive statistics were carried out with SPSS 20. The model illustrated in Fig. 2 was analysed by means of structural equation modelling (SEM) using Mplus 7.1 (Muthén and Muthén, 1998–2012). SEM modelling entails the concurrent calculation of confirmatory factor analysis (modelling latent variables) and path analysis. Thus, it provides us with both a measurement model and a structural model at the same time (Kline, 2005). SEM modelling was chosen because it allows complex models to be tested in one step, including the simultaneous testing of direct and mediating effects. Missing data were imputed through the FIML procedure (Geiser, 2010). The data's goodness of fit was evaluated through various fit indexes: ... Source: Hascher, T., & Hagenauer, G. (2016). Openness to theory and its importance for pre-service teachers’ self-efficacy, emotions, and classroom behaviour in the teaching practicum. International Journal of Educational Research, 77, 15–25. |
|
Developing Pre-Service Teachers’ Global Understanding Through Computer-Mediated Communication Technology |
|
Data analysis emphasizes holistic descriptions and interpretation of the social phenomenon created by online discussions and reflections. The primary goal was to provide as broad a presentation of participants’ understanding as possible. I employed the constant comparative method (Strauss & Corbin, 1990) to identify and develop patterns and themes in preservice teachers’ understanding of global education. At the first level of analysis, generally known as ‘‘open coding’’ stage, I read and coded all participants’ online messages and reflection essays to identify major categories of preservice teachers’ thinking and compared data for similarities and differences. Strauss and Corbin (1990) cautioned that during this early stage of data analysis, the research ‘‘may come up with dozens, even hundreds of conceptual labels,’’ thus requiring another stage of data analysis that leads to the discovery and naming of new categories (p. 65). The categorization process involved the reexamination of the data that I had coded and labeled as I collapsed categories into themes. Source: Zong, G. (2009). Developing preservice teachers’ global understanding through computer-mediated communication technology. Teaching and Teacher Education, 25, 617–625. |
In research articles the Past Simple is predominantly used when describing data collection and analysis methods. However, Present and Future Simple/ Would + Infinitive are more appropriate when describing proposed methods in research proposals. Both active and passive voice can be used, as in the examples below:
The use of the passive voice is especially common in descriptions of procedures (e.g., steps to be completed in a certain order, under specific conditions, etc.). Note that the Present Simple Passive is common when conventional procedures established in a research area are described (especially in hard sciences, medicine, psychology, etc.):
It is advisable to switch to Future Simple when you want to show what you are planning to do in your study to separate this from the standard procedures applied in your area:
On the other hand, although passive voice is a typical feature of academic writing, continuous use of the passive voice may result in writing in which the reader loses sight of who the agents are or may perceive the text as overly monotonous. To avoid this, the active voice can be used, especially when one’s own commentary on different aspects of their research is provided.
The use of the active voice is in fact quite frequent. It can alternate with the passive voice to ensure a better “flow,” as in this extract from a psychology article (note that Past Simple is used to describe what was done in the study).
| The Emergent Effects of Psychological Intervention |
| During the administration of the intervention, students received sealed envelopes with writing materials from their teachers, who were blind to intervention condition and the research hypotheses. In silence, students in the interventioncondition wrote about their most important values, such as friendships or artistic ability, while students in the control condition wrote about their least important values (see the supplemental material in Cohen et al., 2006, for details). After approximately 15 min of writing, students placed the materials back in their envelopes, and the materials were returned to the researchers. Students believed they were participating in a regular classroom writing exercise and were unaware that the exercise was a psychological intervention or part of a research study (Psychology). |
| Source: Powers, J., Cook, J., Purdie-Vaughns, V., Garcia, J., Apfel, N., & Cohen, G. (2016). The Emergent effects of psychological intervention, Psychological Science, 27(2), 150-160. |
Note that if there is a choice between I/ we pronouns when the active voice is used, “we” tends to sound more appropriate when one writes in collaboration with other authors rather than as a single author. The pronoun «I» may figure in published academic writing (especially in mathematics, psychology, etc.), however, it is safer to avoid the use of «I» and «my» so as not to sound too assertive and individualistic unless this is a norm in your field.
A variety of linking words are commonly employed to discuss data analysis and collection methods, and specific procedures.
Prior to/ before + noun/ V-ing, next, during, after (that), on + noun/ V-ing (on obtaining the samples)
As a result, consequently, because, because of/ due to smth