China
Africa
Research design
This page is part of Taylor & Francis and Sense about Science's research integrity toolkit.
Research design refers to the methodology or framework for conducting research, including how data is collected, analyzed, and interpreted.
Poorly designed studies:
- Result in inaccurate conclusions
- Undermine the reliability of research
- Waste resources
- Damage reputations
- Damage future research by others
As an early career researcher, you may face constraints such as limited experience, inadequate training, or pressure to produce results quickly, which can lead to poor design choices.
Reproducibility and replicability
Although "reproducibility" and "replicability" are sometimes used interchangeably, reproducibility means obtaining consistent results using the same original data.
Replicability means obtaining consistent results across studies aimed at answering the same scientific question, each of which has obtained its own data.
Factors such as lack of open research good practice and poorly described or flawed methods can prevent researchers from being able to reproduce or replicate a study.
Making research ethical and replicable
Research must be ethical and replicable, although in some disciplines (humanities and social sciences), strict replication or reproducibility can be challenging.
To achieve this:
- The design should be appropriate for the research question. For example, if the question is "Does this intervention decrease the risk of family breakdown?" the study design must have the numbers and representativeness required to answer it. A non-randomized study involving interviews with 20 families would be limited to answering a different question, such as "What is the experience of some families of this intervention?"
- The study design must also try to address confounding variables or potential biases that could affect results, such as estimating the effect of excluding sites that are hard to sample or restricting the role of a commercial funder in the analysis. Critical Appraisal Skill Programme (CASP) has some good explainers about the different types of research bias.
- The ethical implications must be identified and managed. You should ensure adequate planning time for replicability and ethical considerations in the design phase, to ensure that everything that could compromise the conduct of the research, and the validity, reliability, and quality of findings, is discussed and addressed. When you have your proposed design, you should map ethical considerations to it so you can consider whether you require more time and resources to manage them well.
The EQUATOR network offers freely available guidance to support research design, including a central hub for guidelines for different study types.
Use of AI in Research
The use of Artificial Intelligence (AI) in research is rapidly increasing. It offers exciting possibilities for all stages of a research project as well as aspects of the research process.
However, it gives rise to a number of challenges:
- Potential for bias in AI algorithms, caused by unrepresentative or flawed underlying data, which can skew results
- Concerns over accountability and transparency, where researchers do not understand or cannot explain how the AI is making decisions
- Over-reliance on AI-generated conclusions or the uncritical acceptance of results
- Ethical concerns regarding the use of AI in sensitive research areas, such as privacy violations or the replacement of human decision-making
There are also wider concerns about the ethics of AI and its impact on society and the environment.
Given the fast-moving and technologically complex nature of this field, researchers should seek advice from their institutions on how best to approach these issues.
To ensure that AI is used appropriately, researchers need to be proactive in mitigating risks to research integrity by:
- Critically evaluating AI tools to understand the data that they have been trained on
- Ensuring appropriate data handling
- Being transparent about when and how AI has been used
If you have used AI responsibly as part of your research, you should detail this in the methods or acknowledgments section.
The U.K. Research Integrity Office (UKRIO) has produced guidance on Embracing AI with Integrity, and the European Commission has Living Guidelines on the Responsible Use of Generative AI in Research.
Go to the next section of the toolkit, browse all sections below, or download the toolkit as a PDF
Built with Shorthand