From Classroom to the Field: What Studying Statistics Never Prepared Me For in M&E
I studied Statistics; I knew regression models, hypothesis testing, probability distributions, and sampling theory. I could derive formulas and defend methodological choices before a panel. I thought I was ready for anything data-related. Then I walked into the field and realised that everything I knew was only the beginning.
When you sit across from a woman who has waited hours in the sun to participate in your data collection exercise, she is not a row. She is a person with a history, with fatigue, with suspicions about why you are asking these questions. Her willingness to give honest answers depends entirely on whether she trusts you, and you may have five minutes to earn that trust before she decides to tell you what she thinks you want to hear.
We covered social desirability bias in research methods, but I had little preparation for managing it in real field interactions. I had to learn how to introduce myself in a way that lowered a respondent’s guard, build trust within a limited amount of time, and recognise when someone was giving a socially acceptable response rather than expressing what they genuinely thought.
1. The Textbook World vs. The Real World
Classroom exercises often present data under controlled conditions. Variables are clearly defined, datasets are prepared, and the problem is usually well structured. Fieldwork rarely gives you that luxury. In Monitoring, Evaluation, Accountability and Learning (MEAL), the task is not simply to analyse data, but to understand whether development programmes are actually working, for whom, and why. It sounds straightforward until you try to do it in a community where roads are difficult, respondents are fatigued and suspicious, and pressure to show positive results quietly distorts how data gets collected and reported. I stepped into this space with strong technical skills and discovered quickly that they were necessary, but nowhere near sufficient.
2. People Are Not Data Points
Statistics taught me to think in populations and samples. A beneficiary, in that framework, can easily become a row in a dataset: clean, neutral, and abstract. The field destroys that abstraction immediately.
When you sit across from a woman who has waited hours in the sun to participate in your data collection exercise, she is not a row. She is a person with a history, with fatigue, and perhaps with suspicions about why you are asking these questions. Her willingness to give honest answers can depend on whether she trusts you, and you may have only a few minutes to begin building that trust.
We covered social desirability bias in research methods, but I had little preparation for managing it in real field interactions. Research on sensitive survey questions shows that respondents may modify what they report when they are concerned about embarrassment, judgement, or possible consequences, making the context of data collection important to response quality (Tourangeau and Yan, 2007).
I had to learn how to introduce myself in a way that lowered a respondent’s guard, build trust within a limited amount of time, and recognise when someone might be giving a socially acceptable response rather than expressing what they genuinely thought. I also had to develop active listening and cultural sensitivity on the job. These things can determine the quality of your data before a single number is ever recorded.
More participatory approaches to monitoring, evaluation, and learning similarly emphasise involving local actors in data collection and analysis and ensuring that tools reflect local contexts and voices (OECD, 2024). That lesson became much clearer to me in the field than it ever had in the classroom.
3. The Form Is Never Final
I understood survey design academically: question types, scale construction, and skip logic. What I was not prepared for was deploying an instrument in the real world.
The first time I worked with a KoboCollect form in the field, questions that seemed clear to us made no sense to respondents. A term we used routinely had a completely different local connotation. A skip-logic rule that worked in testing broke down under a combination of answers we had not anticipated.
These were not statistical errors. They were design and human errors, and catching them required structured field pre-testing.
A pilot is not a formality. Pre-testing helps identify problems with question comprehension, sequencing, skip patterns, administration, and other aspects of an instrument before full implementation (U.S. Census Bureau, 2021). It is the moment you discover the difference between the survey you designed and the survey that actually works.
Learning to communicate those flaws upward, diplomatically but clearly, is another skill I had little opportunity to practise during my statistical training.
4. Data Quality Is a Daily Battle
In the classroom, I often encountered data after much of the work required to produce and clean it had already been done. In MEAL, I became responsible for data quality from the point of collection, and that changed the way I understood the entire process.
Field data-quality threats are constant. Enumerators may skip questions. Respondents may misunderstand scales. Network failures can interrupt digital submissions. GPS coordinates can be recorded incorrectly. Sometimes, the threat is not accidental; there can also be subtle pressure to produce results that reflect well on a programme.
Addressing these issues requires regular data reviews, validation rules built into forms, and the ability to identify inconsistencies before they become bigger problems. Routine assessment of completeness, consistency, and other dimensions of data quality is also central to established data-quality guidance (World Health Organisation, 2023).
A quality issue caught on day one is manageable. The same issue discovered during analysis can be much harder to correct. My statistics course had not prepared me to think and operate at that level.
5. Indicators Reflect Choices, Not Just Measurement
In school, indicator design often appeared to be a technical exercise. A good indicator is SMART: Specific, Measurable, Achievable, Relevant and Time-bound. In practice, however, indicator selection is rarely purely technical.
Indicator selection often involves balancing donor reporting requirements, programme priorities, and the changes communities themselves consider important. These priorities do not always align neatly. The MEAL officer may find themselves at the intersection of these different interests, advocating for useful measurement while maintaining productive relationships.
Guidance on results frameworks similarly recognises that indicators should be selected carefully, reviewed iteratively and connected to the priorities and contexts in which programmes operate (OECD, 2024).
Sometimes, field experience shows that an agreed indicator is not capturing an important change. Raising that evidence and discussing possible adjustments with programme teams and funders becomes part of the MEAL role. Learning how to navigate those conversations is something experience has taught me to manage.
6. Communication Matters More Than Calculation
I spent four years becoming proficient at analysis, R, Python, Excel, and Power BI. What I was not trained to do was explain what any of it means to a non-technical audience.
In MEAL, your audience is rarely a statistician. It may be a programme manager who needs to make a decision before Friday, or a community stakeholder who simply wants to know whether the programme is making a difference in their lives.
The ability to translate technical findings into clear, actionable communication—through a well-structured report, an intuitive dashboard, or a concise verbal debrief—is arguably one of the most valuable skills in the MEAL toolkit. Research on evidence use has similarly found that clarity, relevance, accessibility, and relationships with decision-makers can influence whether evidence is actually used (Oliver et al., 2014).
I have seen brilliant analyses go unread because reports were too dense. I have seen simple findings drive major decisions because they were communicated at the right moment, to the right person, in the right format. Even strong analysis has limited value if the people making decisions cannot understand or use it.
Was My Statistics Degree Wasted?
Not for a moment.
My statistical foundation lets me interrogate a number before accepting it, ask whether a sample is representative, whether a correlation is being mistaken for causation, and whether an average is hiding a distribution that tells a more important story. Those instincts protect programmes from making decisions on faulty evidence.
But my degree was only the starting point. The classroom gave me the tools. The field is teaching me when to use them, how to adapt them, and sometimes, when to set them aside and simply listen.
If you are a statistics or social science graduate stepping into MEAL for the first time, come with your technical skills held firmly in one hand and your assumptions held loosely in the other. And keep learning from the people and contexts behind the data.
References
OECD. (2024). Effective Results Frameworks for Sustainable Development: Achieving Impact by Design. Paris: OECD Publishing. https://doi.org/10.1787/aaa4bb52-en
Oliver, K., Innvær, S., Lorenc, T., Woodman, J., & Thomas, J. (2014). A systematic review of barriers to and facilitators of the use of evidence by policymakers. BMC Health Services Research, 14, Article 2. https://link.springer.com/article/10.1186/1472-6963-14-2
Tourangeau, R., & Yan, T. (2007). Sensitive questions in surveys. Psychological Bulletin, 133(5), 859–883. https://psycnet.apa.org/record/2007-12463-007
U.S. Census Bureau. (2021). Statistical Quality Standard A2: Developing Data Collection Instruments and Supporting Materials. U.S. Department of Commerce. https://www.census.gov/about/policies/quality/standards/standarda2.html
World Health Organization. (2023). Toolkit for Analysis and Use of Routine Health Facility Data: General Principles. Geneva: World Health Organization. https://www.who.int/publications/i/item/9789240063938
ABOUT THE AUTHOR
Effiong Innocent is a MEAL Intern at Cloneshouse Foundation and a First-Class graduate of Statistics from the University of Jos. He writes about data, development practice, and the ongoing education of working in the field.
