‘There are always vacancies: there are always roads not taken, vistas not acknowledged. The search must be ongoing; the end can never be quite known’ (Greene, 2000, p. 15). 

Data analysis is an essential part of any research design. It needs to be closely linked to the theoretical approach and methodology as well as to the plans for disseminating the research results. Data analysis is the research activity that dives into the evidence, structures it and explains it. Data analysis and interpretation help us prepare the sharing of the research findings in a way that is rigorous, transparent and verifiable as well as appropriate to relevant audiences. 

Data analysis and interpretation is a complex process that involves various cyclical iterations of:

• planning 
• analysing 
• synthesising 
• interpreting.

When done with an EDI lens, data analysis and interpretation acknowledge the identity of the researchers and researched individuals as well as the contextual aspects of the research highlighting potential limitations. This is linked to the fact that the interpretation of the data needs to be honest, i.e. not hiding or altering intentionally any findings, misleading our knowledge of the phenomena under investigation. Researchers also need to be interested in including diverse settings and participants and be attentive to aspects of their intersectionality that may affect the reporting of the findings. The Research Governance Framework for Health and Social Care issued by the Department of Health (2005, p. 8) in the UK recommended the following inclusive approach to research: ‘Research, and those pursuing it, should respect the diversity of human society and conditions and the multi-cultural nature of society. Whenever relevant, it should take account of age, disability, gender, sexual orientation, race, culture and religion in its design, undertaking, and reporting. The body of research evidence available to policy makers should reflect the diversity of the population.’ This consideration can only be guaranteed at this stage of the research process if it has been adopted in all the previous phases. 

Research practices can be unintentionally discriminatory, inaccessible, and ignorant of systemic marginalisations. Planning how the data collected is going to be critically interrogated to provide EDI-sensitive answers to the research questions is an integral part of the methodological approach of the research design and needs to be congruent with the aims of the research and the methods of data gathering – whether quantitative or qualitative. 

Doing this well necessitates identifying and understanding predispositions in the data collected and analysed. For instance, in qualitative empirical research, the use of multiple coders to interpret the information independently, and then crosscheck their categorisations or thematic codes, is one strategy that can help to address limitations. In fact, when we are interested in EDI in research, we are also interested in the identities of the actual researchers and the protocols they follow. This involves being vigilant of biases by discussing them explicitly and suggesting how they may be mitigated.  

Mitigation may happen in several ways: involving key stakeholders in the process, providing an explicit point of view, reflecting on assumptions and/or disrupting hierarchies and knowledges. For instance, an equitable interpretation of the data can be ensured including interpretations by the communities studied or when researchers themselves are insiders to the issues under study. Researchers could consider how marginalised, minoritised or under researched groups can be involved in all stages of data analysis and interpretation. Jane Seale (2016) invited researchers to consider research with silenced groups, such as students with disabilities, as a political act. Academic activism of this kind suggests that, in fact, ‘researchers need to engage with diverse voices […] voices may be multi-tonal and we need to think carefully about what methods we can employ in order to be sensitive to each tone’ (Seale, 2016, p. 160). Researchers could also consider how to share ownership of the data collectively analysed and interpreted. This situation has been exemplified in educational research by involving the collection and analysis of interview data online in a student-instigated study (Fox et al., 2022). Alternatively, researchers could reflect on their analysis and discuss as a limitation of the study not having included participants’ perspectives in such direct manner. Lucy Yardley (2017) suggested that researchers must show sensitivity to the data to achieve high standards in qualitative research and generate useful knowledge. She offers the example of ‘not simply imposing pre-conceived categories on the data but carefully considering the meanings generated by the participants’ (Yardley, 2017, p. 295).

In the video below, Inma Álvarez, Jennifer Agbaire and Grainne O’Connor discuss the importance of including marginalised, minoritised or under-researched groups in all stages of data analysis and interpretation.  

Transcript (156.7 KB) .

A possible strategy to the interpretation of data is framing our thinking using a range of theoretical approaches that aim to problematise discriminatory attitudes and practices. Critical Race Theory (CRT) challenges the presentation of objectivity and examines and combats oppressive structural racism, sexism, and classism. More recently, Quantitative Critical Race Theory (QuantCrit) is offering researchers a more nuanced and less biased interpretation of findings (García et al., 2018), for instance, by using confidence intervals in our interpretations. Also, researchers could utilise the social model of disability, designed by people living with a disability to highlight the social barriers and prejudice they face, combined with other theoretical frameworks that contribute to the emancipation of disabled individuals (Beauchamp-Pryor and Symeonidou, 2013).

Another aspect of EDI sensitive research is a consideration of the languages, tones and registers for the research. Again, this consideration is key at all stages of the research design from initial communication with participants and data collection in their preferred language to the dissemination of findings. The linguistic dimension in data analysis requires a consideration of whether it is appropriate to start with a careful preparation of translations of the data from the language used by the participants into the language of the researcher’s context when these are different. The question that emerges here is who is responsible for the translations and how to deal with mistranslations and misinterpretations (Viebrock, 2020). The alternative is to conduct the analysis and interpretation in the language of the data collected, an approach that has been found to produce cultural themes otherwise overlooked (Dolan et al., 2023). In any case, choice of terms and nuances in meaning could be negotiated by the communities under study even when the same language is shared. During the interpretation and explanation of the data, participants could also be invited to contribute to produce linguistic innovations as well as make suggestions on adequate language outputs which can be meaningful to the participants. Some initiatives offer support with different approaches to multilingual environments. For instance, ORB International has shared how they navigated the complex linguistic landscape to research in Nigeria using a collaborative approach with local partners.

Only by being sensitive to the implications of findings on the population studied can we ensure that the data analysis is relevant to the population researched. Making research accessible to them is also at the core of an EDI approach to our research. Accessibility could be achieved in different ways such as involving participants throughout the research process or enabling their access to the research findings in appropriate ways in order to widen and deepen knowledge possibilities that not only reflect more faithfully their realities but also have the potential to make a more relevant impact.