Research Design

Abishek S Narayan
8 min readApr 27, 2020

--

How to design research that involves critical thinking and scientific rigor.

Good research design can predict, prevent and sort most problems that happen during research.

Research is simply an investigation of facts in order to arrive at a logical inference; thereby producing new knowledge. Such an investigation must be systematic, rigorous, and having critical thought for the conclusions to be reliable, and therefore useful.

Here, I will share few things that I learnt in the process of designing my PhD research that might be useful for researchers from most scientific disciplines — especially the ones in their early career. However, this might also be useful to anyone trying to develop critical thinking to “research” anything, right from figuring out what pollen you’re allergic to, to what the optimal quantity of dog food your dog, Milo needs. At the end of the day, all of it could be researched, provided, you have the funding. Whether it will be interesting to anybody else, of course is an open question.

Research design is a strategy that has to be carefully chosen and planned, in order for the research study to be systematic and clear while producing the most accurate answers to the “research question”. It’s like the blueprint for a building you want to construct. If that isn’t well made, obviously the building will collapse or slant, on it’s own (i.e., self contradictions) or through external factors (i.e., opposing theories).

You can’t build without a blueprint ! Image © Adobe/Pexels

It’s like the blueprint for a building you want to construct.

Lets follow the same analogy of research as a building under construction. Then, the foundation is the fundamentals of the field you’re in, and that’s built through a Literature Analysis (read more on it here). The more familiar you are with the available literature and data on the field, the stronger your building’s foundation is. Further on, the basement, is the research questions you identify, and the main floors are data collection, the top floors are perhaps the data analysis. Finally the building’s radio mast would depict the conclusions of the research itself.

So a research design would begin after the literature analysis leading to the research question, data collection, and data analysis including the interpretation lens and inference mechanism. These words need to be unpacked surely. For that we need to first understand the steps in a generic research study, given below.

An interpretation of a generic research design process

Research Design Process Explained Step-wise

  1. Literature Analysis
    Although, I have explained it at length in my previous article here, it’s worth repeating — familiarizing the existing work in the field of research prevents us from re-inventing the wheel, and understanding what worked, and what didn’t.
  2. Research Gap
    Based on the review and analysis of the literature in the field, we can start to identify what is the specific gap that’s present in the field. This means, that scholars might have researched on why A causes B, and what causes A, and B itself, but perhaps how A causes B might be a question that is left unanswered. These are in few cases, explicitly written at the end of publications as scope for further research, or might need to be implicitly understood, when compiling and reviewing literature. No matter, what the field, there will always be a research gap — if it is a well researched field the gap might be quite small.
  3. Research Questions
    Formulating the research question is one of the most significant steps in the entire process, as the questions decide the boundary, direction and nature of research itself. A research question is with which you aim to close the research gap, upon answering. The more specific you are with the research question, the easier it is to answer. Typical research questions begin with What, How, or Why — aiming to understand a fundamental phenomenon, it’s mechanism, and cause respectively. Who and When are less often used, but this completely depends upon the research gap identified. e.g., “Who are the people within a charity most likely to donate to another charity?” or “When do people get motivated to make a donation in a calendar year?”.
  4. Hypothesis
    The use of a hypothesis is dictated by the nature of research, whether it is inductive or deductive (see figure below). A hypothesis is a proposed explanation with limited information available at this stage of research. In a deductive research, based on the observation and existing knowledge a an explanation or ‘theory’ is hypothised, which is then tested through data from experiments or models. Whereas, in an inductive research, based on an observation, data is first collected, and then a theory which fits the data is then proposed. Both forms of research are well accepted, and is highly dependent on the field, and the research questions.
  5. Methodology Selection
    Research design and research methodology are often confused with each other. While a research design concerns with the overall research and includes several steps mentioned, a methodology is only relevant to data collection, and data analysis. Of course the selection of methodology affects the nature of research, it is still only a tool for conducting the research. However, in many cases, due to inclination or knowledge of particular methodologies, the research questions themselves are adapted. This is not good practice- “To a person with a hammer, everything they see is a nail”. A methodology could be quantitative, qualitative, or mixed methods. And under mixed-methods, there are several variations of quantitative and qualitative that occur sequentially, or concurrently — that could be explanatory, exploratory or transformative. These are details, which are well covered in Creswell’s 2009 book on Research design. There will be a separate article on Methodology for research coming soon. In general, the selection of methodology is another critical aspect of any research. Any method, would come with it’s own advantages, challenges and limitations. They go hand in hand with a theoretical lens, meaning a specific perspective of looking at the problem in hand. For eg., understanding Why a national Government chose to close borders during the COVID19 situation, could be viewed either through an epidemiological lens or a political-economy one, giving vastly different answers, however accurate they may individually be.
  6. Data Collection
    In order to answer the research questions, a methodology is applied in the form of data collection. This could be experimental for material science, observational for animal behavior, case study based for governance, or model based for climate change. Regardless of the field and nature of data, the collection must be done in a systematic way that is free of biases, and spill over effects. The data collection must be done keeping in mind the errors that could creep in, making the final results unreliable. Most established methodologies come with plenty of documented experiences that warn, and provide mitigation strategies to minimize such biases and errors.
  7. Data Analysis
    This is the part where the collected data is inspected, cleaned, organised, and interpreted. It is here that the data is often fit within a theoretical framework to standardize the information for making an analysis transparent and systematic. Depending on the nature and volume of data, it could be analysed through software that exist for each of these fields. The number of such tools that are being developed for specific analysis is ever on the increase. It is important to keep personal biases such as confirmation bias at a check to carry out a neutral analysis. Double blind checks with colleagues is particularly useful at this stage. Data visualization is complementary at this stage which will make the interpretation easier.
  8. Results and Discussion
    It is in this section of the research, that the results are compiled and presented in a way that is useful for the community. Here, you document clearly what happened in your experiments, and how the results could be interpreted. The discussion section mainly allows the researcher to base their findings in a wider context and what that means for this specific research. The limitations of the methods and narrowness of the research is also owned up in this section.
  9. Conclusion
    The shortest section of most research articles is the conclusion, and that basically answers the research question that was set out to be answered, in the most concise and precise way possible. The scope for further research, and the smaller gap that now exists could be pointed out.
  10. Publication
    The most exciting phase for any research project is not the publication. It could be, but it probably is not. It is the phase where all the work carried out in the past weeks, months or years have to be summarized succinctly in less than 10,000 words. The aim most of the time is to publish in an internationally recognized journal, which is indexed in the important databases, has a wide readership, and of course given an impact factor. In this process, the peers in your specific scientific domain, who are well aware of what is good and bad research practices, will review your work and endorse it for publication. This is probably the gold standard of ensuring scientific rigor in the research community — Making a fellow expert buy your argument.
Inductive vs Deductive type of research

A Research Plan is a document, where all of this could be put together with clear reasons as to why this this research would be interesting — as relevance to science and society, research objectives and the whole research design that is proposed. It could also have few other important sections — the timeline (as a ghant chart), a budget, risks and mitigation. Such a document, (the preparation of which itself would be a good exercise) could be maintained as a live document and serve as the guiding big picture and the agreement between all those involved in the research — students, supervisors, colleagues, interns and more.

Finally, few overarching points for a good research design —

  1. Focus on Causation and not merely Correlation.
  2. The research must be deep enough that it is comprehensive and well analysed, but not narrow enough that it is not as useful to anybody else. For example, the case study is scope for carrying out the research, and must not drive the research itself — then it would be consultancy and not research itself.
  3. Research could be both fundamental inquiry or application driven — they both are necessary. What is fundamental now, would be applied later. However, if you are keen to strike a balance between relevance for science and society, check out Transdiciplinary research.
  4. There is a wholly different approach to social science research based on Grounded Theory, where it is aimed at developing a higher understanding that is ‘grounded’ in, or derived from a systematic analysis of data. The best resource to check out is Kathy Charmaz’s book from 2014 which has branched out from the original publication by Glazer and Strauss 1967.

At the end of the day, research design is like your building. It needs a good foundation, floors of data, a well recognizable mast which is our conclusion, and when viewed by outside, it needs to glass transparent and clearly well designed for it to make it to the Magazine cover page !

About the Author:
Abishek S Narayan is a PhD researcher at ETH Zurich and the Swiss Federal Institute for Water (Eawag), where he works on Water, Sanitation and Solid Waste for Development.

You can follow him on Twitter and LinkedIn.

--

--

Abishek S Narayan

From my times as ETH Zurich PhD, Oxford MSc and work in India/Africa and travels elsewhere. Often on Life and Water-Waste Research. Sometimes on both.