Research Resources

Institutional Review Board (IRB) proposals

Alyssa Staats

Institutional Review Boards (IRBs) review research involving human subjects to ensure that participants are protected from potentially harmful research. This resource provides an overview of the roles of IRBs and ethics guidelines. It also includes practical tips for researchers preparing IRB proposals, including an annotated informed consent checklist, guidance on timeline planning and responding to common concerns of IRBs, and handling unanticipated events. Researchers who are already familiar with the roles of IRBs and obtaining human subjects training certificates may wish to begin at the section "Preparing the proposal."

Important terms

  • Institutional Review Board (IRB): Ethics review committee. US requirements often govern the contents and elements of an IRB application, even if you also submit the application to an ethics board somewhere else. All J-PAL funded projects need to pass IRB review at MIT or cede authority to another host institution.
  • Common Rule: Shorthand term for the Federal (US) Policy for the Protection of Human Subjects, which outlines the criteria and mechanisms for IRB review of human subjects research.
  • Exempt research, expedited review, and full review: Different levels of scrutiny that are required for research protocols. IRBs sometimes use slightly different procedures to determine which category a project falls under and what the review process is for each. You can find guidance on the homepage of your IRB, but in the end only your IRB can decide which category your study is.
    • Exempt research involves minimal risk and fits under one of the exempt review categories described in the Common Rule. Exempt status means the study is exempted from some (but not all) regulatory review, though not from ethics review, and IRBs may still place conditions on exempt research. It is up to the IRB, not the researcher, to determine whether the study qualifies for exempt status1. The revisions to the Common Rule include new categories and clarification of existing categories of exemption. See Human Subjects Decision Chart 2: “Is the Research Involving Human Subjects Eligible for Exemption Under 45 CFR 46.104(d)?” 
    • Expedited review involves research with minimal risk and fulfills a set of other criteria listed by the US Department of Health and Human Services, Office of Human Research Protections (DHHS OHRP, 2017). Expedited review is carried out by one member of the IRB Committee.
    • Full review is required if the protocol is greater than minimal risk and/or does not fall into one of the expedited or exempt review categories, according to both pre-2018 and 2018 regulations. This may happen, for example, when a study targets a vulnerable population. Full review requires that the protocol is discussed by the full Board, which requires a Board meeting (DHHS OHRP 2009 & 2018).
  • Amendment: You should submit an amendment request to the IRB to make any change to the original protocol. This includes updates to personnel, consent scripts or processes, surveys questions, funding, and participating institutions.
  • IRB Authorization Agreements (IAA): Agreements between two IRBs that one of them will take on all reviewing responsibilities for a project for both of them. For example, if a research team includes a principal investigator (PI) at Tufts University and one at Cornell, Cornell can cede reviewing responsibilities to Tufts. DHHS OHRP provides a sample IAA , though many IRBs now process IAAs online via Smart IRB. Please check with your IRB to see which is preferred.
  • Adverse event: A violation of IRB protocols or an event during the research that carries the risk of potential harm to the subjects. This includes data breaches. Adverse events must be reported to the IRB.

The basics

Institutional Review Boards (IRBs) review research involving human subjects to ensure protection of research participants from potentially harmful research.

Research involving human subjects obtains information through interventions or generates identifiable private information (including bio-specimens). Identifiable information means that the identity of the subject may readily be ascertained or associated with an individual. Private information refers to behavior taking place in a context in which an individual can reasonably expect no observation, recording, or information that the individual provided for a specific purpose and can reasonably expect will not be made public (e.g., a medical record).

The "Common Rule" is the popular term for the Federal (US) Policy for the Protection of Human Subjects45 CFR part 46, which outlines the criteria and mechanisms for IRB review of human subjects research. The revised Common Rule went into effect January 21, 2019. All new protocols submitted after January 21, 2019, are required to comply with the revised requirements. It is up to individual IRBs, however, whether to apply the previous or revised version of the regulations to ongoing studies that were submitted before that transition date (DHHS OHRP 2019a).

J-PAL IRB requirements

All research projects that are either funded or implemented by J-PAL must be reviewed by the IRB at the host institutions of all PIs on the project and adhere to all policies and protocols approved by the IRB. Effective all rounds of J-PAL funding starting October 1, 2022, and for all projects where data collection is supposed to start after January 1, 2023, at least one of the IRBs reviewing the project must have status as an IRB Organization (IORG). An IRB’s status can be found by consulting the database of IORGs.

Under the revised Common Rule, only one institution will be named the “IRB of Record,” and the other institutions will cede authority to it. If MIT is the IRB of Record, then all key personnel must have completed CITI training (see below). J-PAL also requires this of  all J-PAL research staff. Field surveyors generally do not need CITI training but should be given training by the study staff on ethical research procedures and sign confidentiality agreements before contact with research participants2. Some IRBs may require CITI training for all staff; it is important to check specifics with your IRB.

CITI/Human Subjects training certification

  • IRBs have different requirements as to which study personnel must receive CITI human subjects training and certificates, though requiring it of key personnel is common. As stated above, CITI training is required if MIT is the main IRB of record and of all J-PAL research staff. All US research institutions require human subjects training of their staff.
  • You can do one of these two:
    • The CITI training course on “Social and Behavioral Research Investigators” is preferred by MIT (and most US institutions) — you can find the course either via your host institution’s IRB or by going to the MIT COUHES website and following the links for MIT affiliates or unaffiliated personnel. See also this guide for instructions on navigating the CITI site.3
    • NIH training course recognized by MIT and the NBER: Protecting Human Research Participants (PHRP). This course now charges a fee for learners, but CITI can be taken for free via the MIT link above.
  • Make sure to also keep your certificate updated; it requires refreshing every three years (if you give CITI a permanent email address they will remind you). One way to complete this renewal requirement more efficiently is to take the refresher course in social and behavioral research; these courses are designed to be shorter and test learners on the most important content. Note that the reviewing agency may need to have the original certification on file for these courses to be acceptable (CITI Program). See also this guide, particularly highlights on page 3, for instructions on enrolling in CITI courses, including refresher courses.

Ethics guidelines and principles for research

The principles of ethical research are detailed in IRB training. These principles, described below, are formulated in the Belmont report (DHHS OHRP 2017). For a more detailed overview of the role of ethics in social science research, see our resource on ethical conduct of randomized evaluations. 

Respect for persons

Research subjects need to be given a chance to independently decide and volunteer to participate in the study given knowledge of all aspects of the study that concern them: procedures, costs, risks, and benefits. Derived from this principle is the obligation for researchers to:

  1. Carry out an informed consent procedure or an assent procedure (for a minor), and
  2. Protect vulnerable populations, such as prisoners and children.


Researchers should consider the burdens and risks of the research in all aspects, including the length and effort of the survey, possible embarrassment from answering questions, psychological burden on subjects and surveyors, and risk of personal data being exposed. In medical research, beneficence is interpreted very strictly; for example, researchers must end a trial and begin treating the whole study population if it becomes clear that the treatment significantly increases welfare.

Note: The rules of beneficence (not doing harm and minimizing possible harm) should be applied not just to the subjects of your research but all human actors involved, for example your surveyors (who need to be protected from, e.g., psychological harm from conducting very difficult interviews, say, with torture survivors). They apply to both the study procedures (e.g., how burdensome the interview process is) and any interventions you conduct (e.g., how burdensome the policy is).


Researchers should ensure that a just distribution of burdens and benefits governs the selection of the sample, subjects, and treatment recipients. The justice principle requires that the benefits of research don't go only to one group and the burdens to another

Random assignment and the justice principle

It may appear that simply randomly assigning a program to some people but not others falls under the definition of "injustice" above. Do the phrases "when some benefit to which a person is entitled is denied without good reason or when some burden is imposed unduly" and "equals ought to be treated equally" rule out randomized designs? An article by Glennerster and Powers (2016) discusses this question in some detail. The relevant part starts on p. 12 of the ungated version.

Here are some additional thoughts:

  • Suppose the intervention is part of your research and provides a benefit to some subjects but not all of them. These benefits do not represent an entitlement: if you do not conduct the research then no one will receive the benefit. The hypothetical alternative of giving the benefit to all study subjects (which makes the study worthless) is not relevant if the intervention is not implemented by the researchers.
  • If the intervention is implemented by someone else, and the research is only measuring the effects of an existing program, the benefit or harm from the intervention should not be a factor in the decision if the research is ethical. We need to study programs that are conducted by governments or NGOs, especially if they may be unethical.
  • The real ethical questions arise in the gray area in between. What if the researchers implement a program that poses a risk? What if the researchers get involved in the randomization or implementation of a program carried out by another organization? The latter is particularly problematic if a government program is tested that is in principle available to all study subjects: in this case, a benefit might be withheld as part of the research design to which the study subjects are actually entitled. In such cases, real ethical trade-offs need to be made between the inherent, general benefits of studying a program, and the potential for harm or injustice done to specific subjects. The benefits from learning may sometimes justify the disadvantages to the study subjects from receiving or not receiving a study treatment, but if they do not, the researchers may have to reconsider the randomized design. A possible alternative might be a so-called encouragement design, in which all study subjects have access to the program, but only a random subset is given additional information or "nudges" to encourage use.

Roles of researchers and the IRB

Weighing the general benefits of research against individual costs is the key task of an IRB. However, you should, of course, also ask yourself this question. You should only go ahead with study procedures after you have made every effort to minimize the burden on your subjects and any other affected parties, and made sure that you don't find your own research ethically questionable. A first check as to whether a given research protocol is problematic or not might be introspection: Would you feel upset, treated unfairly, or subjected to considerable potential for harm in your own study? Knowing all the details of the study, would you choose to participate? 

More on ethics considerations are covered in the corresponding resource, and Glennerster & Powers (2016) discuss gaps in IRB review and ethics frameworks.

Only if you are sure that you can justify the research to yourself should you turn to the matter of describing your research protocol, and its acceptability from an ethical standpoint, to the IRB. Having thought through these questions will help you design procedures that truly minimize the burden on subjects and anticipate questions from the IRB. We will discuss some common concerns and issues that you should consider and be able to address further below in the section "Responding to common concerns of IRBs." 

Choosing an IRB and planning a timeline

Which IRB?

  • Biomedical testing or any health procedures more invasive than measuring weight or height typically have to be reviewed by a medical review board. Social science research can go to a general IRB or a social and behavioral science IRB.
  • An IRB at the institution of each PI must review the project, either as a full IRB application or as a cede to another institution. 
  • A local IRB (for J-PAL projects, typically at the host university of the J-PAL regional office) may need to review as well if the study is in a different country. Keep in mind that regulatory procedures, standards, and timelines can all vary substantially by country. For example, Zambia requires both IRB approval and research approval from the government; a research review and an ethics review would therefore be distinct processes there. If you are not sure which institution should be reviewing your project, especially in countries where J-PAL has not established a regional office, it may take some additional time to research and identify the appropriate institution to conduct a review. You might start this research process by consulting the OHRP's (incomplete) lists of general international standards and regulations or social and behavioral research standards, or searching the OHRP's database of registered IRBs & IORGs, but you should still confirm your specific project country's regulatory procedures with local staff or experts (Department of Health and Human Services, Office of Human Research Protections 2018).
  • J-PAL-funded projects must be reviewed by MIT COUHES, either as a full IRB application or as a cede (ceding is preferable if the project has no MIT PIs involved). 
  • Many US institutions can sign IAAs between them, so only one US institution needs to review the project. Under the new Common Rule, IRB oversight for most federally-funded collaborative research projects located in the U.S. is required to use a single IRB (commercial, academic, or hospital-based).

Timeline planning (based on experience with US IRBs)

Plan when to submit your IRB application by calculating backwards from the piloting and study start. Typically, the procedure is that an IRB officer reviews the application first for completeness and then a Board member receives it. For full review, the proposal is then discussed at the next Board meeting; otherwise the decision is made by the Board reviewer. If you are not entirely sure if you will need a full review, plan for the full review and also budget for needing to resubmit.

  1. Full review: Deadline is usually 3 weeks before the Board meeting. IRB staff and Board members will pre-review and bring up any questions; the approval is made at the Board meeting. Work on the study cannot start until IRB approval is received.
    • Find out when the relevant IRBs meet: some do not meet in the summer or winter break. Some US IRBs don't have more than one or two meetings between the Spring and Fall terms.
  2. Expedited or exempt: Review by IRB staff and Board members and has no fixed timeline (usually short but up to several weeks is possible).
    • Inquire about the turnaround times of the IRBs you can submit to - sometimes expedited reviews can take longer than submitting to a well-timed Board meeting (particularly if the IRB is busy around the Board meeting time).
  3. Any additional procedures can add significant time to the review process.
    • Some ethics boards outside the US may have different procedures, e.g. require a presentation of your project to the Board (e.g., the Comité National d'Ethique pour la Santé et les Sciences de la Vie, CNESS, in Mali).
    • Some IRBs (e.g., at Brown University) use a "cultural consultant" to make sure the study protocols are culturally appropriate. 
  4. Resubmission: The Board may give only conditional approval and/or request changes that require a second submission.

Some additional notes on the timing:

  • Two IRBs: if one of two IRBs requests protocol changes you may need to resubmit to the other one.  Make sure you plan for this.
  • Initial submission plus amendment: Often, it is faster to get an amendment to an existing study approved rather than submitting a completely new protocol. Therefore, it may be better to submit an initial proposal that outlines the study and describes pilot procedures, for example, and then to submit protocol amendments as you get a better idea of the exact study procedures. In some cases, you need initial IRB approval for a study to receive funding so your study will typically later require an amendment (e.g., J-PAL funded projects require IRB approval before funds can be released; this is an MIT requirement). 

Administrative data

Research involving administrative data is very likely to be considered human subjects research and thus subject to review by the IRB of the researcher’s home institution, particularly if the data contains identifiers. Beyond the purview of the IRB, an overlapping web of U.S. federal laws and regulations, state laws and regulations, and institutional restrictions and procedures governs access, storage, and use of administrative data.

Because compliance requirements and definitions of “identifiable” data differ by field, source, and geography, the guidance of an IRB and legal counsel at the researcher’s home institution can be critical in sorting through compliance and reporting requirements. Local regulations may apply as well (e.g., research involving the use of European data conducted by a US-based PI may involve IRB review by the PI’s home university, and would also be subject to the EU’s General Data Protection Regulation (GDPR)). 

Data Use Agreement (DUA) documents the terms under which a data provider shares data with a researcher’s home institution for use by the researcher. This agreement, which typically must be approved by legal counsel at the researcher’s home institution, contains a number of provisions that can significantly impact the underlying research. Many universities have a standard template that includes terms and conditions that are acceptable to the university, and were created with researchers’ needs in mind. More on DUAs is covered under the resource Using administrative data for randomized evaluations.

IRBs often require researchers to furnish signed DUAs before approving a study protocol, and data providers often require IRB approval before signing a DUA. Research teams may request provisional approval from one party, making clear to all parties the process and constraints, to find a path forward. Research teams can facilitate the process by proactively and frequently communicating with the IRB and data partners. 

More information on working with IRBs and DUAs with administrative data can be found in the IDEA Initiative’s Handbook on Using Administrative Data for Research and Evidence-based Policy and in the resource Using administrative data for randomized evaluations

Preparing the proposal

This section is based on experience with US IRBs. Check if your IRB has a submission form that pre-specifies the information they want to see. These are typically found on each IRB’s website along with templates for consent forms and other helpful documents. With the revised Common Rule, some universities such as MIT have changed the application and forms (e.g., closure, adverse event, etc.). Please download these forms directly from the your IRB's site rather than using old versions you have on hand.

Contents of a proposal

The information required is very similar across most proposals.

  1. Research objectives and purpose: Be brief on the purpose or scientific background of the study. A comprehensive literature review is not required. Make this accessible to a well-educated lay person.
  2. Procedures of the study: Be detailed. The IRB tends to care most about the "how." This is partly because the first review is not done by a researcher but by an IRB officer. They may request a lot of detail upfront in order to avoid additional back and forth for the Board member who will review the protocol. Walk through what the subject experiences in the study step-by-step, including duration of participation and location. Let someone else read what you wrote to see if they understand your study procedures.
    • Submit all the interview guidessurvey instruments, and recruitment materials you want to use, and provide translations. Translations may need to be certified; ask the IRB about that.
    • Describe the consent procedure in full and submit the consent form exactly as you plan on using it. Waivers or alterations of consent/assent may require additional materials. Once approved, the consent form cannot be changed without an amendment so it's worth putting some work into this (see next section and our resource on intake and consent processes). You may need to use the IRB-stamped form.
    • Data handling: An important risk of study participation for subjects is that their personal data is exposed in some way. Make sure you have good procedures for both physical and electronic data protection, including storing data in locked files and clearly stating the timeline and process for storage/destruction of personal information. Moreover, make sure you describe in the consent form to whom and how the subjects' data will be accessible. State clearly which specific information will not be shared with others. This is particularly important if you intend to publish the study data (which most researchers now want to do). Make sure you describe how you will de-identify the data and use language in the consent form that does not prevent you from publishing the de-identified data later on. See the “Permission to publish” section of our data publication resource for sample consent language that permits subsequent data publication.
      • Deception and debriefing: The IRB will need a good explanation if you need to withhold information from subjects or deceive them. Some journal referees have rejected papers that they perceive to contain unnecessary deception. If you do use deception, the IRB will likely require that you debrief subjects afterwards and explain the true study purpose or procedures to them. A good rule of thumb is that the subjects will need enough information about the risks, costs, and benefits so that they can make an informed decision if they want to participate; everything else will require special reasons and/or a debriefing.4
      • Study participants: Describe exclusion/inclusion criteria and justifications.
        • Are you working with vulnerable populations (children, prisoners, pregnant, mentally disabled, or developmentally challenged individuals), and if so, why?
        • Subject numbers: you cannot use more subjects than you originally planned to enroll unless you submit an amendment. If you are not sure of the final subject numbers, provide an upper bound estimate. There is no penalty for surveying fewer subjects than stated, but IRBs do not like to see approved subjects exceeded.
        • For an RCT, describe the selection method for treatment and control groups.
      • Risk assessment: Explain all the potential risks for the subjects and how you are guarding against them.
        • IRBs will read this very closely and they may come up with risks that you feel are remote. You may be required to safeguard in some way against such identified risks.
        • Whatever risks and benefits you identify you need to mention in your consent form (see below and in our Define consent and intake process resource). 

      Some additional hints, based on experience:

      • There may be elements of your study that change later. If these changes are not substantial, you should write your protocol to allow for them. For example, you may want to adjust the amounts paid to subjects, change your estimate of how long it takes to conduct a survey, or rephrase some survey questions. If you get a reasonable range of options for these fields approved in your initial application then making the needed adjustments will be easy. For changes not written into the protocol always submit an amendment (and wait for approval) before implementing them in the field, even changes as small as the contact information on a consent form.
      • Similarly, if you plan to choose between two different procedures, say, based on piloting results, you may request approval for both of them, and explain that you will choose one (and how). This can save you time later on. 
      • Note that even if you do not identify any risks, IRBs may still require you to mention risks and benefits in the consent form, e.g., by saying "there are no risks or benefits to you from this study." There is no need to try and bend over backwards to convince yourself, the IRB, or the subjects that there is absolutely zero risk. Depending on your study population and context, this can make subjects actually more suspicious or at least confused: if there are no risks, and so far the study sounded completely harmless, why is this explicitly mentioned? Thus, on balance it is actually not necessarily going to reduce participation or harm your relationship with the subjects if you mention small risks. It can help the subjects see that you thought carefully about this but that these are the only risks you came up with. 

      Procedure for informed consent

      Normally, written, signed consent is required from all study subjects.

      Consent procedures for research involving minors are more complicated, so you should check with your IRB on the specifics. Depending on the age of the child (typically seven or older) children have to give assent to study procedures that involve them, though the method of obtaining assent will vary depending on age. In addition, parent(s) have to consent to research involving their children. If the study involves more risk, both parents' consent may be required. MIT requires that research involving minors have written informed consent of a parent as well as the suitably documented assent of the minor, if the child is over seven years old. 

      The revised Common Rule includes a new "Key Elements" section and a rearrangement of content that is designed to facilitate a potential subject's decision of wheter to participate. Under the revised 2018 Common Rule, the requirements for informed consent have changed, with the addition of: 

      • "Key information" to be presented at the beginning of the consent form
      • New consent elements
      • Changes to waiver criteria and documentation (plus other process changes)
      • A "broad consent" option for unspecified future use of identifiable data/bio-specimens

      There are exceptions to the need for written consent. These may be important if the written consent procedure deters or prevents subjects from participating in the research. Examples could be subjects who cannot read, who do not want to be linked with their study answers, or who are in a rush and do not want to sign unfamiliar documents before studying them carefully. 

      Some of the alternatives available for such cases are:

      • Oral consent with a witness (witness signature replaces subject signature)
      • Waivers of written consent (replaced by an oral procedure), especially in the case of minimal risk
      • Consent from only one or none of the child's parents (e.g., unavailable parents or risks from involving the parents, e.g., in cases of abuse)
      • Waivers of assent (e.g., if the child is too young or incapable of assenting)

      See Human Subjects Decision Chart 13: “When Can Informed Consent Be Waived or Altered Under 45 CFR 46.116(f)?” for more details.

      In our experience, for studies involving young students, where getting every parent to consent would be impossible, typically the head of school consents for all students. Examples of such scenarios include boarding schools or when it is likely parents will not all read or return the forms and when the study consists of non-invasive research such as math games conducted on school time. For high school students involved in educational interventions, both students and the head of school (or parent) provide legal binding consent. 

      An important provision of US law is a so-called Certificate of Confidentiality, which prevents law enforcement from accessing research data by subpoena for the purpose of criminal persecution.5 The National Institutes of Health offers helpful resources on what a certificate of confidentiality is and answers some frequently asked questions about how it works. The law on certificates of confidentiality was originally conceived to allow research on illegal drug use by Vietnam veterans without compromising subjects or deterring them from participating. These certificates can help a researcher protect their subjects and assure subjects of the confidentiality of their responses. For some background, see for example Wolf et al. (2015, 76 pages).

      More details on informed consent are covered in J-PAL's Define intake and consent process resource.

      Consent Forms

      MIT has useful guidelines on providing and communicating adequate information and ensuring comprehension when obtaining informed consent. In particular, the language researchers use in their consent forms should be simple and clear. They include the following:

      • Wording should be understandable to those with an 8th grade reading level.
      • Avoid technical language or jargon.
      • Use commonly recognizable terms.
      • Explain terms that may not be easily understood.
      • When minors are involved, use age-appropriate language.
      • Note that one needs explicit permission from the IRB to obtain verbal consent, in place of the usual written consent, even if the survey is conducted over the phone. See Transitioning to CATI: Checklist and Resources for additional adaptations that should be made for phone surveys

      Researchers should at a minimum include the following in their consent forms, even if an oral consent: 

      1. A brief explanation of the purpose of the study 
      2. That subjects can stop at any time or choose not to answer any questions they want 
      3. A local contact and phone number/email for questions. 

      Subjects should get a copy of the consent form with this information. They should also receive a link to the local university's IRB office in case they have concerns they do not want to voice to the study researchers, e.g., “If you feel you have been treated unfairly, or you have questions regarding your rights as a research subject, you may contact the Chairman of the Committee on the Use of Humans as Experimental Subjects, MIT, Room E25-143b, 77 Massachusetts Ave, Cambridge, MA 02139, phone +1-617-253-6787."

      MIT has consent templates on their website that researchers can use. The consent templates have also undergone changes with the Common Rule revisions. Check with your IRB with questions about modifying a consent template provided by MIT or another institution.

      If you plan to collect personal data from participants residing in the European Union you should include additional language in your consent forms, in accordance with the European Union's General Data Protection Regulation (GDPR). This should include information on how long you will retain subjects' information, whether their personal information will be transferred to the US (which may not offer the same level of protection as the EU), etc.

      You should review the Annotated informed consent checklist, based on the OHRP checklist, to make sure you are including all elements of informed consent in your form. The checklist document contains information on:

      • The various exceptions to written consent above, 
      • Elements of a written consent statement, and 
      • Sample language that makes sure J-PAL data management and data publication requirements are met and the study data can be published in de-identified form.

      Broad Consent

      The Revised Common Rule allows "Broad Consent" in lieu of informed consent. Broad consent is obtained only with respect to storage, maintenance, and secondary research use of individuals' identifiable data without obtaining additional consent. You are not required to obtain informed consent through broad consent; this is simply an option that is now available to researchers.

      Broad consent under Common Rule has several mandated elements, including:

      • A general description of the types of research that may be conducted with the identifiable private information.
      • Whether sharing of identifiable private information might occur.
      • Types of institutions or researchers that might conduct research with the identifiable private information.
      • Description of the period of time that the identifiable private information may be stored and maintained and description of period of time that they may be used for research.

      You can find the complete list of mandatory elements to broad consent on the HHS website.

      If an individual refuses to provide broad consent, the researcher can choose to de-identify the subjects' data to conduct further research. Your IRB can provide guidance on institutional policies regarding refusals, including on re-requesting broad consent or obtaining informed consent.

      Not all IRBs have implemented use of Broad Consent. Be sure to check with your IRB to understand whether this option is available for your research. Implementing Broad Consent requires careful thought about the language of the consent form. When the IRB evaluates secondary research proposals, it must determine that the proposed secondary research is within the scope of the identified types of research in the original consent form.

      Responding to common concerns of IRBs

      Your IRB may ask you for clarification or flag things in your research protocols as problematic that may seem surprising to you. It is useful to know ahead of time what type of designs and issues can raise concerns or are particularly important to IRBs. Conversely, there are research design considerations you should keep in mind that may help you overcome some common concerns an IRB might have.

      • Who is conducting the intervention? Recall the principle above: it is an injustice if "some benefit to which a person is entitled is denied without good reason or when some burden is imposed unduly." Researchers are not allowed to commit injustices in their research. You are still allowed to study programs, policies, or events that constitute an injustice by this definition. If an NGO carries out a questionable program (say, re-education camps for transgender youth) you can study the effects of these programs (and it may be a great service to humanity if you do). However you cannot run such a re-education camp, or help the NGO run it, or even select subjects from a potential pool for them. That would be unethical because there is good reason to think that these re-education efforts harm the children who are subjected to them. The more involved you are in the design, administration, targeting, or any other aspect of the intervention you are studying, the more the intervention itself falls under the scrutiny of the IRB and under your obligation as a researcher. 
      • The IRB will be particularly concerned if you get involved in a study of a benefit that people are legally entitled to receive. Note the phrasing "some benefit to which a person is entitled is denied." If part of your research plan is to suggest to the government to withhold an official program for research purposes, you will probably (rightfully) not be allowed to do that by your IRB. As another example, say there is a policy that could be harmful or unpleasant to some people. For example, suppose the government is choosing some households for tax audits. It is possible that the IRB will be concerned if you as the researcher get involved in the selection process (for example, by proposing a randomization strategy to pick households for auditing). It can help to make the argument very explicit that the government would have audited someone either way, and that without the randomization it may just have been a different, arbitrarily chosen person.
      • Related to the argument above, IRBs are typically much more concerned if there is potential for undue burden or harm to the study participants or subjects, as opposed to potential benefits. Helping the government to randomly assign a benefit is often an easier case to make than helping to randomly assign a burden. This means, for example, if you are studying a demand elasticity and you want to vary prices, it will be looked upon more kindly by the IRB if you reduce prices (i.e., provide subsidies) to a random subset of people rather than increase prices. 
      • You may not always agree with the concerns that are raised by the IRB — take the example of the audit study. You may think, for example, that it is not unethical to audit people based on random selection rather than let the government choose by some other criteria that may actually be more questionable. If you truly think so, you should make your case to the IRB. In the author's experience, a request by the IRB for clarification is sometimes just that: there is not necessarily an issue, they just want clarity. Therefore, do not back away from protocol details that are important to your study unless you find that the IRB's concerns about the research design are justified. You may have to be prepared to explain your procedures at length. However, if you truly think that what you are doing is (a) ethical and (b) important research, do not self-censor in order to take the path of least resistance.
      • IRBs can be particularly sensitive if the cultural context is unfamiliar to them or they feel that research interferes with traditions or deeply-held beliefs of subjects. Your IRB may, for example, worry about a special burden on women from conducting research on things like sexually transmitted diseases, gender empowerment, or fertility in a very patriarchal or religiously conservative society. In those cases, the IRB may consult a cultural advisor or require special protections for subjects. It often helps to demonstrate a deep understanding of the cultural context of your study and to have local researchers on the research team.
      • Physical contact as part of your research is almost always subject to special scrutiny. For example, if you interview mothers about the health of their child, the child's assent is typically not required, but if you measure the kid's weight or even take their blood for a malaria test, not only do you need assent, you need the consent of one or both parents (and typically biomedical clearance). Relatedly, as soon as any body fluids are involved in your research procedures, biomedical IRBs will get involved. You need to think about things like how to safely dispose of any contaminated waste products. Note that some IRBs may go a bit overboard, and require for example that your medical testing is done by trained nurses, even if the procedures are designed to be safely carried out by lay persons (like a rapid detection test for malaria). In those cases, it is again sometimes worth pushing back and making your argument that it is not unethical to instead train your surveyors to carry out these procedures.
      • Inclusiveness and screening: even after asking you a lot of questions about the potential harm that your study might do to subjects, some IRBs will then turn around and ask if your procedures are not unfairly excluding some people from the research. This is the flip side of fairness; if you are not sure that the research is harmful or beneficial, preventing someone from participating, say, based on their physical condition, age, gender, or race, or based on literacy or education, can be considered unjust. If you screen out some subjects from your study procedures make sure to carefully explain why.
      • Payments and compensation: many IRBs will require you to compensate your subjects for the time and effort they spend on your research. Indeed, it is the ethical thing to do if you are burdening your subjects by using up a lot of their time. Paradoxically, however, paying too much may also be a problem; namely if the amount is so high that it has coercive effects. We provide some tips below, with a more detailed discussion in the “Compensation” section of J-PAL’s Define intake and consent process resource.
        • The general consensus within the ethical community is that payments are not “coercive” if the study passes the bar of risk to potential benefit (to an individual or society) assessment (with the payments not in the “benefit” side of the calculation). 
        • The concern over payments is particularly strong when it involves vulnerable populations. For example, IRBs typically do not allow you to pay mothers for the enrollment of their children in a study. Rather, you can reward study participation by giving a gift to the child. This is in order to prevent mothers from silencing their concerns over the study and agreeing to their child's participation in order to collect the payment. 
        • The IRB may also ask that the payment is not emphasized in the recruitment materials, and you cannot call payments a "benefit" of the study in the consent form.
        • Both sides of this argument have some truth to them, and you should carefully weigh the costs and benefits of each approach. One way of dealing with the problem may be to not mention the full amount you are paying when you ask subjects if they agree to participate (to avoid coercion) but still pay an amount that fully compensates them for their time later (to avoid exploitation). This could be done by paying a baseline participation reward which everyone receives who "shows up" (essentially a token of appreciation for hearing you out, even if the subject later withdraws from the study or does not complete the survey). In addition, you can then pay an hourly fee at the end of a completed survey. 
        • A good approach may also be to first get approval from a local IRB and use their agreement when submitting the protocol to the US IRB, because they will often defer to what is considered appropriate by local ethics committees.

      Handling IRB violations, unanticipated problems, and adverse events

      After your study starts, you will need to remain aware of potential changes or issues with the study that require reporting and/or other processes. Researchers should ensure that there is a good communication system between field staff and PIs so that any changes or events are communicated to the PIs to then report to IRB. 

      It is the researchers' obligation to report the following to the IRB that approved the study:

      • Violations of the IRB-approved study protocols.
      • "Unanticipated problems" with the research.

      Protocol violations or incidents

      Protocol violations are any unapproved changes, deviations or departures from the study design or procedures of a research project that are under the investigator’s control and that have not been reviewed and approved by the IRB. Protocol violations are divided into two categories: major (reportable) or minor (non-reportable). 

      Changes to the study, such as to study team members, sample size, survey instrument, consent forms, etc. must be reviewed and approved by the IRB before implementation. Any changes in the research study design and/or procedures that are both within the investigator’s control and implemented prior to IRB approval, must be reported as protocol violations. Such changes concern the IRB because they may unintentionally affect the participant’s rights, safety, or well-being, or the completeness, accuracy, and reliability of the study data. 

      If the changes in the research study design and/or procedures do not have a major impact on either the participants' rights, safety or well-being, or the completeness, accuracy and reliability of the study data you do not need to report them to the IRB but should document them in the study files.

      Any problematic or unanticipated events involving the conduct of the study or participant participation also need to be reported to the IRB. This could include, for example, breaches of study participants' privacy or confidentiality (see below), non-compliance, and changes to measures to mitigate risks to participants.

      Data security and personally identifiable information

      Potential breaches of privacy or confidentiality of study participants’ data are “major (reportable) incidents” that must be submitted to the IRB. The PI is responsible for ensuring that research data is secure when it is collected, stored, transmitted, or shared. All members of the research team should receive appropriate training about securing and safeguarding research data.

      Note that IRB protocol violations often involve the exposure of confidential data, accidental or not (such as losing tablets, computers or cameras which have subjects’ data – whether or not it is encrypted). When personal or identifiable data is exposed, the IRB needs to make a determination on what risk of harm this exposure inflicts on the subjects, and what mitigating measures the research team needs to undertake in order to address these risks. They might decide, for example, that the researchers need to contact the study subjects to inform them of the breached data, or that additional psychological or legal assistance must be offered to subjects to deal with any consequences of the data breach. The IRB may also decide that data collected in violation of the study protocols or without full informed consent from the subjects cannot be used in research, though this is an extreme outcome that only rarely occurs. In most cases, the risk of harm is low, and the IRB will only request to halt data collection and amend the study protocols to ensure compliance with the original IRB proposal. 

      Unanticipated problems and adverse events

      You must also report any unanticipated problem involving potential risk to participants or others. An unexpected, research-related event is one where the risk exceeds the nature, severity, or frequency described in the protocol, study consent form or other study information previously reviewed and approved by the IRB.

      OHRP gives further guidance on "unanticipated problems," including how and when researchers should report the problems to the IRB. Unanticipated problems that are serious adverse events should be reported to the IRB within one week of the investigator becoming aware of the event, while other unanticipated problems should be reported within two weeks.

      Adverse events encompass both physical and psychological harms. They occur most commonly in the context of biomedical research but can occasionally occur in the context of social and behavioral research. If an adverse event is an unanticipated problem, then it must be reported to IRB. Some examples of adverse events are listed below.

      Emergencies/when do you have to report or stop?

      Any time that there is a significant adverse event, the IRB of Record should be contacted immediately to see what needs to be done (e.g., does the study need to stop or do you just need to file an Adverse Event Form?). This is at the IRB’s discretion. Significant adverse events include:

      1. Non-consented surveying, sampling, or data use
      2. Participant deaths
      3. Work in dangerous situations
      4. Breaches by staff

      For example, an unexpected death is a serious adverse event and must be reported to the IRB. Typically, unexpected deaths must be reported within 48 hours, although it is best to report as soon as possible. An unexpected death is defined as the death of a participant that was not related to a risk of participation listed in the protocol or consent document, and was more likely than not caused by the research procedures and/or study interventions. 

      Depending on the circumstances, the IRB may need to take immediate action to minimize further harm to participants, such as halting the enrollment of additional participants or suspending approval of the research. 

      Continuing review

      The IRB is required to review and approve all non-exempt research projects at intervals appropriate to the degree of risk, but not less than once per year. This is called "continuing review." 

      Under the revised Common Rule, continuing review is not required for:

      • Research that is eligible for expedited review,
      • Exempt research conditioned on limited IRB review,
      • Research that has completed all interventions and now only includes analyzing data, even if the information or biospecimens are identifiable,
      • Research that has completed all interventions and now only includes accessing follow-up clinical data from clinical care procedures.

      Under the new Common Rule, for the MIT IRB new protocols approved under expedited review (not going to the Board meeting) no longer require yearly renewals unless otherwise determined by COUHES. Instead, investigators must submit an annual progress report (very minor) and a Continuing Review Questionnaire every three years. Previously approved studies still need to be reviewed annually. MIT also requires annual review until the project is completely done with data analysis.

      For continuing review, researchers must give a report to their IRB before the approval expires. MIT requires researchers to submit a Continuing Review Questionnaire, updating the IRB on the progress of their research, 60 days before the IRB approval expiration date. The form includes information on activities conducted in the past year, subject recruitment and attrition, adverse events, subject complaints and publications. If researchers do not submit the form by the required date, the study will be automatically terminated, research grants related to the study will be suspended, and no further research with human subjects can be conducted under the protocol.

      The IRB makes its determination on whether to re-approve research based on whether there is any new information, especially with respect to potential benefits or risks to the subjects, that could alter its decision. Such information may also require the IRB to update its risk assessment for the protocol and its corresponding review category, so be aware that this is a possibility when budgeting the review time for your study's continuing review. The IRB also assesses whether there is any new information that would necessitate revision of the protocol and/or the informed consent document.

      For more information, see Human Subjects Decision Chart 11: “Is Continuing Review Required Under 45 CFR 46.109(f)?”

      Reporting and publishing

      Publishing: Researchers should report once a study has been published as a working paper or other publication in their yearly review form. Often IRB protocols are kept open for years after field work is complete, as multiple papers may come out of one study. MIT’s IRB allows you to close a study once “the study is closed to enrollment; all subjects have completed all study procedures; data collection is complete; AND data analysis is complete.” Researchers should report when de-identified data is made public in their IRB Closure.

      Trial registry: Researchers should include their IRB information when registering their trial with the AEA RCT Registry. Specifically, they should state whether they received IRB approval (mandatory), the IRB name, approval date, and approval number.

      For J-PAL projects: All researchers conducting research projects funded or implemented by J-PAL should submit IRB certificates and consent forms to J-PAL.

      Last updated March 2023.

      These resources are a collaborative effort. If you notice a bug or have a suggestion for additional content, please fill out this form.


      We are grateful to Jack Cavanagh, Sarah Kopper, Jim Shen, Keesler Welch, and Evan Williams for review and copy-editing of this resource. Any errors are our own.

      For example, per MIT COUHES, studies that are begun without an evaluation of exempt status by the IRB (even if the researcher believes the study is exempt) are in violation of MIT policy and will be terminated immediately.
      More information can be found in our ethics, consent, and surveyor training resources.
      CITI has over 70 courses; make sure you take the right one. For example NOT this one (even though it could be very useful in certain situations!)
      Note that experimental (lab) research in economics also frowns on deception for other reasons. One is that many equilibrium concepts require that subjects hold beliefs that are correct on average. Another is that subjects in experimental labs typically participate in more than one experiment and may start to distrust the researchers' explanations of the study procedures if they are deceived repeatedly and find that out later. It is, however, acceptable to withhold information if you tell the subjects that the information is withheld. For example, suppose subjects interact in a one-shot game and a subject may be randomly assigned to one of two types. It is ok if they do not know which type they are as long as they are told that this is the case. By contrast, it would not be acceptable to say that there is only one type, or not mention at all that there are different types that affect the game, or assign types with different probabilities from what the subjects are told.  Observe that this is a stronger information requirement than the typical RCT, where people may not know that they are in a control group or that a control group exists beside the treatment group; but most RCTs also do not explicitly test equilibrium interactions between members of the two groups. 
      A famous case where research data was subpoenaed was a study on the Irish Republican Army (IRA) conducted at Boston College (see this New York Times article by Jim Dwyer from May 2011). 
        Additional Resources
        1. CITI Program, "Social-Behavioral-Educational (SBE) Refresher 1" (last accessed September 2019). 

        2. Department of Health and Human Services, Office of Human Research Protections, 2017, "Institutional Review Board (IRB) Authorization Agreement" (last accessed Dec 2017). A sample IAA.

        3. Murphy, Kathleen. 2020. “Collaborating with the Institutional Review Board (IRB).” In: Cole, Dhaliwal, Sautmann, and Vilhuber (eds), Handbook on Using Administrative Data for Research and Evidence-based Policy. Accessed at on 2021-03-18.

        4. O’Hara, Amy. 2020. “Model Data Use Agreements: A Practical Guide.” In: Cole, Dhaliwal, Sautmann, and Vilhuber (eds), Handbook on Using Administrative Data for Research and Evidence-based Policy. Accessed at on 2021-03-18.

        5. Staats, Alyssa, Kristianna Post, Lindsey Shaughnessy, and Anita Desai, 2019. "Human Subjects and IPA IRB." Delivered at 2019 J-PAL/IPA Global Staff Training. An overview of the IRB process and how it applies to IPA/J-PAL projects.

        Department of Health and Human Services, Office of Human Research Protections, 2019, "The Revised Common Rule Compliance Dates and Transition Provision (45 CFR 46.101(l))" (last accessed September 2019). 

        Department of Health and Human Services, Office of Human Research Protections, 2018a, "Electronic Code of Federal Regulations" (last accessed September 2019). 

        Department of Health and Human Services, Office of Human Research Protections, 2018b, "Table 1: Social-Behavioral Research Standards: Description" (last accessed September 2019).

        Department of Health and Human Services, Office of Human Research Protections, 2017, "OHRP Expedited Review Categories" (last accessed Dec 2017). 

        Department of Health and Human Services, Office of Human Research Protections, 2009, "Pre-2018 Requirements" (last accessed September 2019). 

        Department of Health and Human Services, Office of Human Research Protections, 2007, "Unanticipated Problems involving Risks & Adverse Events Guidance" (last accessed September 2019).  

        Glennerster, Rachel and Shawn Powers, 2016. "Balancing Risk and Benefit: Ethical Tradeoffs in Running Randomized Evaluations". In: George DeMartino and Deirdre McCloskey, Oxford Handbook of Professional Economic Ethics, Oxford University Press. 

        MIT Committee on the Use of Human Subjects website, "Forms and Templates," (last accessed January 18, 2019).

        National Institutes of Health (NIH). "Certificates of Confidentiality: FAQs" (last accessed September 2019). 

        National Institutes of Health (NIH). "What is a Certificate of Confidentiality?" (last accessed September 2019). 

        Wolf, Leslie E., Patel, Mayank J., Williams Tarver, Brett A., Austin, Jeffrey L., Dame, Lauren A., and Laura M. Beskow, 2015, "Certificates of confidentiality: protecting human subjects research data in law and in practice", Minn J Law Sci Technol. 14(1): 11–87.

        In this resource