A semi-structured interview is one of the most effective tools for systematically gathering qualitative and quantitative data. This is a method which allows you to ask predetermined questions, determined, perhaps, by the theoretical framework or theory of change underpinning the project, or by your research hypothesis. It is also one which keeps questions open-ended, to gain a comprehensive view of surrounding information.

Before the semi-structured interview

Selecting an appropriate sample size, or deciding whether to interview the complete treatment group are all important aspects of research design. To do this, you need to consider the characteristics and size of the entire treatment group, and then determine whether a representative sample would be best selected according to some of the attributes, or within sub-groups, or whether taking a randomized sample is best.

Interviewers should be trained, and understand the challenges with asking open ended questions; how to adhere to the research questions and seek specific information, while also not leading respondents to answer a specific way. Remaining engaged while taking notes, and establishing rapport are also important aspects of a good interview technique.

After the semi-structured interview

Once the interviews have been conducted, knowing what to do with this data is key. There are a range of different analytic methods which could be applied to qualitative data, and choosing the right one will depend on the research you are conducting. By definition, qualitative data is categorical, thus, you should already have some idea of the different categories you are working with, and what “strongly agree” or “strongly disagree” might mean in the context of the theory for a given question. Coding and sorting your data to give it meaning is key to a sound analysis. As you designed the questionnaire, you will already have an idea of the value you place on different responses, and what constitutes an improvement in your ordinal data. You will also want to keep track of some of the classification variables which might hold explanatory power in the differences in outcomes, such as gender or age.

Clean the data

You will likely need to spend some time cleaning the data, particularly the ‘open-ended’ questions where the answers may be long, and differ from person to person. If you have had a range of interviewers, you may want to spend some time just reading through all the data to try and gain a consistent view of the picture being painted by the data before beginning with the analysis. Never underestimate the power of ‘eye-balling’ your data. Give it a good read, and long think before commencing with any number-crunching.

Define clear questions

For your more defined questions, it may be simple to represent the findings. Say, for example, you are trying to ascertain the level of resources, and their use across a group of early childhood development centers, some of your questions will ask directly the number of books, toys or games available. This is quantitative information easily shown. You could graph the number of ECD centers which have between 0 – 5 resources, 5 – 10, or more than 10. If you then want to comment on the quality of these resources, you may have your interviewer observe their quality across ranked categories such as ‘new and in excellent condition’; ‘well-used but still with useful life’; illegible, parts missing, very old’. This might be done in conversation with the interviewee, the ECD center worker to ascertain their understanding of the importance of quality and engaging resources. From this data you can easily determine the overall resource need by determining how many of the centers require resources supplementation, but you may wish to apply a deeper analysis to how these resources are perceived by ECD workers. Then there may be questions on use; the frequency and level of engagement. Where frequency may be easy to map (once again you could categorize frequencies and group centers accordingly), you may want to conduct some type of thematic or narrative analysis on the comments regarding how well the children engage with the resources. There will be layers of information. A thematic analysis will allow you to begin to gather themes across a range of comments. You might find themes around reasons why the young children are not engaging adequately such as feeling hungry or tired, or themes around how the significance of engagement is perceived by the ECD practitioners. There are softwares for conducting rigorous thematic analysis through word identification (narrative analysis), or in a small enough sample, you can identify the frequency of themes yourself.

Methods Map

Sage Publishers have a really useful guide, the ‘Methods Map’, for defining and mapping various research methods, which begins to guide you in the most appropriate one for your purpose. The map surrounding qualitative data analysis includes a range of analytic methods. From this resource, you can delve into the broader academic literature to understand how various methods may be implemented. Even when using deductive approaches, such as action research where you are not testing a theory, but rather gathering information to form one; approaches where your questions will be more open-ended, and adhere less to a stricture, you will want to systematically analyze the data from these interviews to develop a strong base of information.

If you have a good idea of the overall implementation and research questions you hope to answer as you design the interview, it will then be much easier to understand what to do with your data.  This is why framing your impact and researching methodologies for reaching this impact is such critical groundwork. Researching what ‘good’ looks like in the space that you’re working will assist you in developing appropriate categories for data collection, which are context relevant, and empirically proven. For example, if an empirical study finds that exercise 3 times a week improved well-being, and that frequency of exercise means certain things about mental wellness, and you’re running a program in community wellness, design your categories accordingly.

Share.

About Author

Angela Biden is a consulting strategist and M&E consultant. She has worked across a range of development, and business contexts. She holds a Masters in Economics and Philosophy, and has worked in the nexus of M&E and social impact; to help those doing good do more of it; for some 15 years. From policy board rooms, to Tech start-ups, to grass roots NGOs working in the face of the world’s most abject challenges; Angela is focused on conducting relevant and meaningful M&E: fit for purpose, realistic, and useful for stakeholders creating positive change.

Leave A Reply