This methodological weakness of Nepal’s development research has been documented for decades. In the 1980s, researchers found that estimates of Nepali firewood use varied so wildly between studies that the numbers were simply too unreliable to say whether Himalayan forests were disappearing at all. Yet, these same numbers were treated as facts, informing a World Bank prediction that no accessible forest would remain in Nepal by 2000.

Likewise, anthropologist Judith Justice found that Nepal’s health data was produced to satisfy bureaucratic reporting requirements rather than to understand conditions on the ground. Planning documents described a Nepal that existed only on paper. Even forty years ago, research was being produced for reporting, not for understanding.

These shortcomings persist today. A large amount of research claims a ‘causal relationship’ between microfinance and women’s empowerment. These methods typically rely on before-and-after surveys. Researchers asked if the women felt empowered, and they usually responded positively because demonstrating progress provided them with continued programme access. This is a social desirability bias. The survey finds empowerment because the method is looking for empowerment. Anthropologist Katharine Rankin's ethnography shows exactly what those methods missed. Rather than asking women if they felt empowered, she asked who held authority over labour and loans within the household. She found women were targeted less for empowerment than for their reliability in repaying. As a result, repayment pressure fell on women, while investment decisions stayed with men. This shows that the method determines the finding.

In recent years, a newer method has entered Nepal’s development research landscape. Designed by Silicon Valley pioneers, human-centred design (HCD) asks researchers to build ‘empathy with users’ through short interviews, then turn the learnings into a redesigned product or intervention. HCD suffers from what is known as ‘design tourism’, where outsiders help communities build a prototype, without engaging with their deep structural power dynamics. According to communication scholar Sasha Costanza-Chock, sympathy does not redistribute authority; feeling for a community is not the same as sharing power over what gets built. Anthropologist Anke Schwittay’s research on microfinance codesign confirms this pattern. Even with genuine sympathy and good intentions, outside expertise tends to favour outsider, technological and commercial solutions over political action or indigenous practice. Nepal’s development sector needs to adopt this method with scrutiny.

Research in Nepal is overwhelmingly commissioned by donors, designed around their accountability needs, conducted by researchers selected through processes favouring prior donor relationships and interpreted for donor reporting requirements. Communities being researched have almost no role in defining the research questions, little access to findings and almost no mechanisms for challenging conclusions that might misrepresent them. According to anthropologist David Mosse, research commissioned to check whether a programme works tends to find that it does because the methodology is designed within the programme’s own framework.

Having traced these methodologies’ weakness, it is worth highlighting what rigour looks like. The Nepal Demographic and Health Survey (NDHS) and the Nepal Living Standards Survey (NLSS) show this in practice. The NDHS uses stratified probability sampling across thousands of households, standardised modules for comparison across survey rounds, and caste and ethnicity disaggregation, making inequality visible. It openly admits it relies on verbal reports, not clinical measurement, and cannot explain the causal mechanisms behind the trends it finds. The NLSS adds two further dimensions. It follows the same households across time, allowing longitudinal tracking of welfare change and spreads data collection across 12 months to capture seasonal variation. In both these cases, rigour does not mean complexity for its own sake. It means choosing methods appropriate to the questions being asked, being transparent about limitations, triangulating across multiple data sources, and designing deliberately against the social dynamics that distort findings.

Every research design should be justified against the research questions, and not chosen for convenience. A method that cannot detect a negative or an unexpected result is built to confirm whatever it was designed to find. This means being explicit about what a method can and cannot show, reflecting on likely biases, triangulating across sources before trusting a result, using mixed methods in qualitative inquiry and adopting longitudinal designs whenever observing change.

Research depends on the social context, and rigorous research helps to design against the hierarchies that shape it—through stratified data collection, working with local researchers rather than one-off enumerators, validating findings before finalising and reporting results in language communities can access and contest. Research participation is not about being a respondent, but about being part of the process that decides what the research concludes.

In Nepal’s development research, methods are largely chosen not for their ability to find evidence, but to suit the assumptions made by programmes and generate reports. Therefore, rigour is necessary because it requires designing a method that is open to new findings. That commitment requires donors willing to fund research with depth, producing inconvenient findings; institutions willing to build evidence systems communities can actually contribute to and contest; and researchers willing to design and conduct research with the possibility of learning and improvising.