The contention that our attempts to explain and understand politics will always remain ethically or politically suspect in some way is correct, but it requires clarification of what “suspect” truly means. For the purposes of this essay, “suspect” will be understood as questionable or deserving suspicion. Importantly, this does not mean research is invalid, but rather that it raises legitimate questions about power and values. This essay will argue that research is always suspect in both elements. It begins by examining the ethical problems in research, specifically the inherent power imbalances between researchers and those researched, drawing on Wood’s (2006) analysis of power asymmetries and Fujii’s (2012) work on compromised consent to highlight the structural problems that persist. Then, drawing on various scholars (Kuhn 1996; Wacquant 2002; Biruk 2018), it shows that political commitments are embedded at every stage of research through paradigm choice, category construction, and positioned interpretation. The essay then considers the counterargument that randomized controlled trials (RCTs) constitute a methodological “gold standard” that avoids these problems (Dunning 2008), before demonstrating that even RCTs remain ethically and politically suspect (Noble et al. 2025). Together, these arguments will prove that our attempts to explain and understand politics will always remain ethically and politically suspect. Acknowledging suspicion strengthens rather than weakens political research.
Power Asymmetries, Unpredictable Harms, and Compromised Consent
Research relationships inherently involve power asymmetries, create unpredictable harms, and compromise consent. These exist as structural features of the research process, rather than individual researcher failures. Fujii (2012, p. 718) identifies “the power imbalance between researcher and researched” as “one of the major sources of ethical dilemmas”. While not immediately apparent, researchers are supported by institutional resources and mobility and are entrusted with representing participants, whereas those researched do not possess such advantages. Fujii (2012, p. 719) notes that “people may consent not only because of social pressure, but also because they believe that establishing a relationship with the researcher will be beneficial in and of itself.”
Researchers are granted unique mobility and power during the research process that is not replicable for those they observe. Plainly, consent is shaped by the very power imbalance it supposedly mitigates; participants may agree not from genuine voluntary choice but in part because they need or want something from the researcher. This dynamic persists even when researchers aim to limit it. Pachirat’s (2009) work in an industrial slaughterhouse demonstrates this perfectly; despite technically working undercover as a “low-level” employee, he remained in “networks of power” (p. 144) possessing an education, mobility, and most importantly, the choice to leave, one his coworkers (and subjects) lacked. As he concludes in his work, “neutrality in fieldwork is an illusion” (Pachirat 2009, p. 143). Pachirat’s work reveals that power imbalances cannot be mitigated by individual researchers but are built into the structure of research itself.
Beyond power imbalances, research also creates unpredictable harms that researchers cannot predict. Despite their intentions, researchers can inflict harm in any number of ways, with their mere presence playing a vital role in power dynamics. Published research can leave subjects exposed, interviews can bring back traumatic memories, and basic contact can lead to physical consequences. As Fujii (2012) explains, social scientists can cause psychological, physical, and social harm to study participants.
Elisabeth Wood’s (2006, p.373) work interviewing civilians in conflict zones demonstrates these risks starkly, where “the ethical imperative of research (‘do no harm’) is intensified… by political polarization, the presence of armed actors, and the precarious security of most residents”. By simply talking to Wood, civilians could risk their lives by being marked as informants, exposing them to retaliation from armed groups. Unlike Wood, her subjects are unable to appeal to international bodies or institutions for protection the way she is. Even if researchers are in positions to protect their subjects, they often have minimal understanding of the risks in comparison to the local residents they interview. As Wood (2006, p. 380) notes: “Rural residents… had a more highly developed sense of evolving risks… than I did.” Regardless of experience or resources, researchers lack the local knowledge necessary to fully understand the dangers their presence creates. If researchers cannot prevent harm that they also cannot predict, the question becomes whether participants can at least give meaningful consent to these unknown risks.
Informed consent is structurally unattainable in research because participants can never be fully informed about uncertain outcomes, future uses of data, or the power relations shaping their agreement. Consent then functions not as a safeguard, but as an ethically suspect foundation for research itself. As Fujii (2012, p. 718) argues, formal consent procedures often amount to performative compliance rather than meaningful understanding: “compliance can be meaningless,” since signed consent forms “have little value if participants do not understand what they were consenting.” This limitation does not stem from inadequate explanation by researchers, but from the impossibility of anticipating how research will be spread, understood, or create consequences beyond the field. Consent also cannot be treated as a one-time act, because ethical dilemmas show up unpredictably during research. Fujii therefore emphasizes that “no set of rules can cover every ‘ethically important moment’ that arises in the field” (2012, p. 718). In such moments, researchers must act without renegotiating consent, effectively making decisions on behalf of participants. This leads Fujii to conclude that “the responsibility to act ethically rests ultimately on the individual researcher” (2012, p. 718). These dynamics highlight that consent cannot resolve the ethical tensions of research but instead displaces them onto the researcher’s judgment. As with power asymmetries and unpredictable harms, compromised consent is not a procedural failure but a structural feature of political research itself.
Political Commitments Are Embedded Throughout the Research Process
Beyond ethical concerns, political research is also politically suspect more broadly. That is not to say that research is partisan, but that it is shaped by unavoidable value judgments which remain unavoidable throughout the research process. Decisions about what questions matter, how concepts are defined, and what counts as a valid explanation are all shaped by assumptions. As Kuhn argues, “the choice between competing paradigms proves to be a choice between incompatible modes of community life” (1996, p. 94), and there exists “no standard higher than the assent of the relevant community” by which paradigms can be judged (Kuhn 1996, p. 94). Paradigms, therefore, shape not only methods but also what counts as a legitimate question, idea, or topic in the first place. When paradigms change, “there are usually significant shifts in the criteria determining the legitimacy both of problems and of proposed solutions.” (Kuhn 1996, p. 109).
This means that paradigm choice cannot be justified through neutral scientific standards alone, since those standards are themselves defined internally. Dahl makes this explicit, arguing that “identifying a question that is important is a moral and normative issue, not a scientific issue” (Dahl 2007, p. 134). For example, choosing to study poverty through randomized controlled trials focused on individual interventions reflects different political assumptions than studying poverty as a structural issue tied to inequality and power. Both approaches may claim rigor, but they rest on fundamentally different views of what the problem is.
Political commitments also shape research through the categories used to produce and interpret data. As Wacquant (2002, p. 1524) warns, failing to utilise proper theoretical processes allows “ordinary notions issued out of common sense” to fill the gap. This allows ideological assumptions to be slotted into seemingly neutral categories. Biruk’s (2018) work illustrates how data are actively produced rather than simply collected. In her analysis of development research in Malawi, researchers produce “clean data” through negotiating with interviewers, deciding when to probe, and judging which responses count as valid. While the name suggests otherwise, clean data remains far from clean, but is the outcome of countless differing assumptions about truth, knowledge, and fact, disguising political power as a natural process. Her discussion of “ghost numbers” and “failed numbers” in AIDS policy further shows how indicators that appear objective are shaped by institutional priorities and policy demands. Interpretation is similarly situated. Pachirat insists on the “embodied nature of all vision” (2009, p. 143), showing how his position within the slaughterhouse shaped what he could observe and understand, forcing attention to “the inevitable partiality of sight” (Pachirat 2009, p. 159). Davenport’s research in Rwanda demonstrates the same dynamic, noting that “being African American was my way in” (2014, p. 1), as assumptions about his identity shaped access and interpretation. Research findings cannot be separated from who produces them. Political commitments are therefore embedded throughout the research process, not just as external pressures, but as fundamental features of political inquiry itself.
The Case for Randomized Controlled Trials as a Neutral Methodological Standard
If research is always suspect, it must also remain suspect in the most neutral methodologies, regardless of their careful design. RCTs are often treated as the methodological ideal because they provide “the strongest causal inference” (Dunning 2008, p. 243), removing researcher discretion through randomisation and ensuring ethical oversight through institutional review boards. Ethically, RCTs appear to resolve many of the problems discussed earlier. Random assignment eliminates researcher discretion over who receives benefits, directly addressing power asymmetries by ensuring allocation is determined by chance rather than any biases. Noble et al.’s (2025) study of randomised cash transfers exemplifies this logic. In the study, mothers were randomly assigned to receive either $333 per month or $20 per month, with the “randomization process designed so that interviewers could not influence the assigned cash gift level” (Noble et al. 2025, p. 7). In addition, Institutional Review Board (IRB) oversight is used to ensure that potential harms are assessed and minimised before research begins. As Noble et al. (2025, p. 8) note: “all study procedures were approved by the Institutional Review Boards of Teachers College, Columbia University, and the New York State Psychiatric Institute”. Dunning (2008) and other proponents may argue that standardised protocols also mean that all participants receive the same information, ensuring meaningful consent.
On political grounds, RCTs are similarly presented as eliminating value judgments through methodological neutrality. Randomisation limits researcher bias at the point of assignment, while outcomes are standardised and measured using pre-registered instruments. Noble et al. (2025, pp. 8-9) highlight this approach through four primary child outcomes: language, executive function, social-emotional development, and resting high-frequency brain activity. These are measured using standardised tests such as the Receptive One Word Picture Vocabulary Test. Within this framework, the intervention is treated as a neutral policy tool rather than a politically motivated choice. Statistical analysis then yields quantifiable assessments of effectiveness, seemingly bypassing the interpretive judgments that complicate other forms of political research. If this claim holds, RCTs seem to offer a methodological approach capable of escaping the ethical and political suspicion that characterises political research.
Randomized Controlled Trials Remain Ethically and Politically Suspect
However, even RCTs, designed to be neutral, remain ethically and politically suspect. Although Noble et al.’s (2025) study appears neutral at first glance, a closer look reveals ethical compromises and political choices at every stage. Ethically, the study is compromised by design. At its core, the study relies on vulnerable, desperate mothers as its subjects, providing them with varying sums of money (some get $333, others just $20) to analyse differences in their children’s development. Noble et al. (2025, pp. 6-7) describe how “mothers reporting household incomes below the U.S. federal poverty threshold… were randomly assigned to receive either $333/month or $20/month.”
While participants were told “they could spend the money however they wished, and that receipt of the funds did not require continued participation” (Noble et al. 2025, p. 7), this framing hides a fundamental ethical problem: mothers living in poverty and desperate for resources cannot give truly informed consent, given their participation is shaped by need, not truly free choice. The power asymmetry between researchers and impoverished participants shapes “voluntary” participation in ways that approval systems cannot eliminate. As Wood (2006, p. 385) notes: “even with research practices and protocols tailored to specific field conditions, inevitably field researchers rely on their judgment.” Effectively, IRB approval formalizes acceptable harm rather than preventing it. The control group (receiving only $20/month) experiences deprivation by design, with their children being monitored while denied meaningful support. This means that those in the study cannot meaningfully consent or withdraw consent without a limiting external pressure, making Noble’s study inherently unethical.
Beyond ethical concerns, Noble et al.’s (2025) study is also politically suspect. Its design embeds value judgments about what counts as a legitimate political problem. Although the authors acknowledge that poverty is entangled with broader forces such as structural racism, their decision to focus on income reflects a political choice. Noble et al. (2025, p. 4) note that it is “difficult to isolate income from its many correlates, including parental education, community socioeconomic factors, and structural racism”, yet only income is tested experimentally. This reflects a paradigm choice: a liberal paradigm treats poverty as a lack of resources, while a structural paradigm treats it as a matter of power relations. As Kuhn (1996, p. 100) argues: “without commitment to a paradigm there could be no normal science.” Testing income interventions requires assumptions about what poverty fundamentally is. Both are valid questions, but choosing between them is political, not methodological. Their choices of what to measure reinforce this, as the study evaluates child development through standardised indicators benchmarked against normed averages (Noble et al. 2025, pp. 8-9). Children “performing approximately 0.2-0.3 SD below the normed average for receptive vocabulary” (Noble et al. 2025, p. 13) are framed as deficient, but the norms are not specified. The decision to measure language skills rather than political efficacy or logical development is a directly political decision. As Wacquant (2002, p. 1470) observes, research should “dissect the social mechanisms and meanings that govern practices.” By focusing on individual developmental deficits, the study sidesteps structural causes of poverty, thereby limiting its scope through its choice of paradigm.
Interpretation also depends on the political paradigm. When Noble et al. (2025, p. 22) conclude that “income alone may not affect children’s early development,” the null result lends support to contradictory interpretations. A liberal interpretation is reached: income doesn’t matter. The structural interpretation, which would highlight the intertwining of poverty with power relations, is not discussed. The same result yields opposite conclusions, confirming Kuhn’s (1996, p. 94) point that paradigm choice involves “incompatible modes of community life” with “no standard higher than the assent of the relevant community.” The RCT doesn’t eliminate political judgment but reorganises it, prioritizing certain questions while presenting choices as methodologically neutral. It adopts a liberal analysis, disregarding other views, thereby compromising the study’s political neutrality.
Conclusion
The proposition that our attempts to explain and understand politics will always be ethically or politically suspect is valid. Ethical problems, such as power asymmetries, unpredictable harms, and compromised consent, remain structural features of research relationships. Political suspicion is equally unavoidable, with paradigm choice, category construction, and interpretation requiring value judgments that cannot be neutral. Even RCTs, designed to minimize bias, cannot escape these problems. Noble et al.’s study demonstrates how ethical compromises (lack of true consent, allowing harm) and political choices (liberal paradigm choice, scales of deficiency) are embedded throughout research. Recognizing this does not invalidate political science but transforms practice. Rather than claiming false neutrality, researchers must acknowledge positioned vision and paradigm commitments. This means practicing Wood’s “situated ethics,” being explicit about normative assumptions as Dahl advocates, and embracing Pachirat’s reflexivity. Maintaining impossible claims to neutrality is negative, as it obscures power relations and makes research less rigorous by hiding limitations. However, acknowledging this openly produces more honest and more trustworthy research than false claims to objectivity. Suspicion, when recognized, strengthens rather than weakens political research.
References
Biruk, C. (2018) Cooking Data: Culture and Politics in an African Research World. Durham, NC: Duke University Press: Chapters 4 and 5.
Dahl, R.A. (2007) ‘Normative Theory, Empirical Research, and Democracy’, in Munck, G.L. and Snyder, R. (eds.) Passion, Craft, and Method in Comparative Politics. Baltimore: Johns Hopkins University Press, pp. 114-139.
Davenport, C. (2014). Researching While Black: Why Conflict Research Needs More African Americans (Maybe). Available at: https://politicalviolenceataglance.org/2013/04/10/researching-while-black-why-conflict-research-needs-more-african-americans-maybe/.
Dunning, T. (2008). Improving Causal Inference: Strengths and Limitations of Natural
Experiments. Political Research Quarterly, 61(2), pp.282-293. https://doi.org/10.1177/1065912907306470.
Fujii, L.A. (2012). Research Ethics 101: Dilemmas and Responsibilities. PS: Political Science & Politics, 45(04), pp.717-723. Available at: https://www.cambridge.org/core/journals/ps-political-science-and-politics/article/research-ethics-101-dilemmas-and-responsibilities/90D5442392CD0716F04E764939054FF7.
Kuhn, T.S. (1996) The Structure of Scientific Revolutions. 3rd edn. Chicago: University of Chicago Press.
Noble, K., Magnuson, K., Duncan, G., Gennetian, L., Yoshikawa, H., Fox, N., Halpern-Meekin, S., Troller-Renfree, S., Han, S., Egan-Dailey, S., Nelson, T., Mize, J., Black, N., Georgieff, M. and Karhson, D. (2025) The effect of a monthly unconditional cash transfer on Children’s development at four years of age: A randomized controlled trial in the U.S. Cambridge: National Bureau of Economic Research.
Pachirat, T. (2009) ‘The Political in Political Ethnography: Dispatches from the Kill Floor’, in Schatz, E. (ed.) Political Ethnography: What Immersion Contributes to the Study of Power. Chicago: University of Chicago Press, pp. 143-161.
Wacquant, L. (2002). Scrutinizing the Street: Poverty, Morality, and the Pitfalls of Urban Ethnography. American Journal of Sociology, 107(6), pp.1468-1532. https://doi.org/10.1086/340461.
Wood, E.J. (2006) ‘The Ethical Challenges of Field Research in Conflict Zones’, Qualitative Sociology, 29(3), pp. 373-386.
Further Reading on E-International Relations
