Research Ethics

Research ethics make up the first pillar of the DIME Research Standards compiled by DIME Analytics. Impact evaluations often directly involve human subjects, and the research team often collects sensitive or personal identifiable information (PII). There are several ethical concerns related to handling personal data, and publishing results from a field study. The research team must ensure complete confidentiality of respondents, and ensure that respondents are providing informed consent, which they can withdraw at any time.

Read First

• DIME Research Ethics Standards, compiled by DIME Analytics.
• Protecting human research subjects and informed consent.
• IRB approvals.
• The impact evaluation team (or research team) must get an ethics approval from an institutional review board (IRB), and an appropriate authority in the study location.
• All members of the research team who handle personally identifiable information (PII) must have up-to-date Human Subjects Research Certification.

Before Data Collection

Before the research team begins primary data collection, it is important to draft ethical standards that must be followed throughout the research study. The research team must follow the guidelines listed below to ensure ethical research:

• Pre-analysis plan. The research team must draft a detailed pre-analysis plan before starting the process of data collection. An ideal pre-analysis plan should discuss the following principles of study design in as much detail as possible:
  • Sample and level of randomization. The research team must decide on the process of sampling, as well as the level of randomization. For example, the research team might decide to perform stratified random sampling, and randomize on the level of households in a village.
  • Sources of data. The research team must also decide the main sources of data they will use during the study. For example, the research team can use administrative data for the study.

• • Key outcomes and primary hypothesis. The research team should also lay out their primary hypothesis, that is, the main question they want to answer using the study. This is often referred to as theory of change. They must also clearly lay out the outcomes they wish to achieve, and what indicators they will use to measure outcomes.

• Human subjects and PII. In most cases, impact evaluations involve directly interacting with, or collecting personally identifiable information (PII) about human subjects. In such a case, the research team must obtain a human research subjects certification. Two commonly accepted courses that research team members can take to get certified are:
  • The National Institute for Health (NIH) offers a free course on protecting human research participants.
  • The Collaborative Institutional Training Initiative (CITI) offers a course on human subjects research.

• IRB approval. An institutional Review Board (IRB) is an organization which reviews and approves all aspects of a research study involving human subjects. After members of a research team obtain human research subjects certification, they must apply for an IRB approval.

• Informed consent. Before involving any subject in a research study, the research team must obtain informed consent from each individual. This means that the research team must clearly mention all possible risks and benefits from participating in a study, either for the survey pilot or as a respondent in the actual survey.

During Data Collection

In some cases, the research team does not need to obtain informed consent from each individual. For instance, in cases where the research team is using administrative data, they can just obtain consent from the concerned government agency. However, even in such cases, when the research involves analyzing data that contains sensitive or personally identifiable information (PII), the research team must follow certain guidelines during data collection:

• De-identification. Often data collection involves handling, cleaning and analyzing data that contains sensitive information like names, addresses, mobile phone numbers, bank or credit accounts, or location information. In such cases, it is important to ensure anonymity of individuals, by hiding the identity of individuals through de-identification. In general, the research team should always work with and analyze de-identified data, except in certain cases, such as monitoring data quality, or conducting a follow-up survey.

• Encryption. It is important to remember that simply de-identifying data is not enough. The members of the research team often share data using the internet, or store it in the cloud. In such cases, the data is exposed to several risks the data is exposed to some risk. Therefore, encryption of this data is also important to protect the identity of participants.

• Documentation and protocols. Another important aspect of ethical and reproducible research is to carefully document before public release of the results of a study. This also involves developing detailed protocols for handling and sharing data that contains personally-identifiable information (PII).

After Data Collection

Even after data collection is complete, there are certain guidelines for ethical research that the research team must follow before publishing results of a study.

• Bias. One of the most common concerns in research is the possibility of researchers manipulating the results of a study before publishing it in a journal. This is called publication bias, and can waste precious time and money. To ensure this does not happen, the research team can pre-register their study with a registry, in addition to developing a pre-analysis plan. Some of the common registries are:
  • The American Economic Association (AEA) hosts a trial registry specifically for randomized control trials.

• • The International Initiative for Impact Evaluation (3ie) provides a registry for experimental and quasi-experimental research in developing countries

• • The Open Science Framework offers a platform for registration, and houses [[registries].

• Reproducible research. Another way to ensure ethical research is to publically release data and codes for others to view and cross-check. This also ensures that research is not only transparent, but also reproducible.

Related Pages

Click here for pages that link to this topic.

Additional Resources

• Berk Özler (World Bank), Taking ethics seriously: Response #1
• Berk Özler (World Bank), Research with adolescents: issues surrounding consent
• David McKenzie (World Bank), A pre-analysis plan checklist
• DIME Analytics (World Bank), Research Ethics & Data Security
• Martin Ravallion (World Bank), Taking ethical validity seriously
• J-PAL, Ethics