Choosing a question, a supervisor and a project

The most important decisions in an honors project are made before any data are collected. Who will supervise you, what broad topic you will work on, and how you narrow that topic into a question that can be answered in eight months together determine most of what follows. A good question asked well, with a supportive supervisor and a realistic plan, will carry a student through bad weather and failed experiments. A poor question, or one far too large, can make even a hard-working student's year miserable.

This lesson follows Priya from her first email to Dr. Okafor in January 2025 to the one-sentence research question she wrote at the end of February. Along the way it covers how to approach potential supervisors, how to judge whether a project is feasible, and how to sharpen a topic into a question that data can answer.

Finding a supervisor

Most departments publish a list of faculty who are willing to supervise honors students, often with short project descriptions. Some assign projects; others expect students to find their own supervisor. Either way, the relationship with your supervisor matters as much as the topic. A brilliant project with a supervisor who is rarely available can be harder than a modest project with one who meets weekly and reads drafts quickly.

Priya already knew Dr. Okafor from her summer job, which made the first approach easy. Students who do not know any faculty members can start by reading department research pages, attending research seminars and talking to instructors from courses they enjoyed. When writing to a potential supervisor, a short, specific email works best.

• Say who you are, your program and year, and when you hope to do the project.
• Show that you know something about the person's research: mention a paper, a project or a seminar, and say what interested you.
• Mention relevant experience or courses, briefly.
• Ask whether they are taking honors students and whether you could meet to talk.

Avoid sending the same long message to every faculty member in the department. Supervisors can tell, and a generic message suggests a student who has not thought about what they want to do. If a faculty member says no, it is usually because they already have as many students as they can supervise well, not because of anything about you.

Questions to ask at the first meeting

The first meeting is an interview in both directions. The supervisor wants to know whether you are reliable and interested. You want to know whether the project and the working relationship will suit you. Priya went to her meeting with a list.

Dr. Okafor answered Priya's questions frankly. She met her students every two weeks, more often during fieldwork; she could pay for the eDNA analysis from a grant if the costs stayed under a set budget; and the station's long-term fenced and grazed wetlands, along with their water chemistry records, were available. Her one condition was that Priya's work fit within the existing study's animal care approval and permits, which would need to be amended to cover her methods.

From topic to question

Students usually arrive with a topic, such as frogs and cattle, rather than a question. A topic is an area of interest; a question asks something specific that data can answer. The process of narrowing is called scoping, and it often takes several weeks of reading and discussion.

Priya's first draft question was: How does cattle grazing affect amphibians in prairie wetlands? Dr. Okafor pointed out that this question has dozens of possible answers depending on which amphibians, which wetlands, which aspect of the animals' lives and which kind of grazing. Answering it properly would take a team many years. Over several weeks, Priya narrowed it.

• Which amphibians? The boreal chorus frog, because it is common enough at the station to provide data at every wetland.
• Which aspect of grazing? Cattle access to the wetland margin, because the ranch had fenced some margins and not others, providing a natural comparison.
• Which response? Breeding activity and early tadpole numbers, because these can be measured in one spring and are directly linked to population success.
• What mechanism? Water clarity and emergent vegetation, because cattle trample margins and stir up sediment.

The final question, written at the end of February, was: Do wetlands where cattle are excluded from the margins support more chorus frog breeding activity and higher tadpole densities than paired grazed wetlands, and is any difference associated with lower turbidity? It is long, but every part of it can be measured.

What makes a good research question

Researchers use various checklists for judging questions. One useful set of criteria is that a good honors question is focused, answerable, feasible, interesting and ethical.

Feasibility is where most honors projects need adjustment. It has several parts: time (can data be collected within the season or schedule?), money (are supplies, travel and analyses affordable?), skills (can you learn the methods in time?), access (do you have the sites, organisms or equipment?) and approvals (can permits and ethics reviews be obtained before you need them?). A question that fails any one of these must be changed or narrowed.

Questions, hypotheses and predictions

Biologists distinguish three related statements. A research question asks what is true. A hypothesis proposes an answer, usually with a mechanism. A prediction states what you will observe if the hypothesis is correct. Priya's statements, in the form she used in her proposal, were these.

• Question: Does excluding cattle from wetland margins increase chorus frog breeding success?
• Hypothesis: Cattle access reduces breeding success by increasing turbidity and removing emergent vegetation used for egg attachment.
• Predictions: Compared with paired grazed wetlands, fenced wetlands will have lower turbidity, more nights with chorus calling and higher tadpole densities; across wetlands, tadpole density will decrease as turbidity increases.

The predictions are directional and measurable. They also allow for partial support. If turbidity is lower in fenced wetlands but tadpoles are not more abundant, the mechanism might be wrong, or the effect might be too small to detect with 12 pairs. Thinking through the possible outcomes in advance, and what each would mean, is one of the best habits a researcher can develop. Some researchers make a simple table of outcomes and interpretations before collecting any data.

A related distinction is between a null hypothesis and an alternative hypothesis in statistics. The statistical null hypothesis for Priya's tadpole prediction is that the mean difference in tadpole density between fenced and grazed wetlands in a pair is zero. That is a statement used in a test, not a biological explanation. Lesson 9 returns to how statistical tests relate to biological hypotheses, and why estimating the size of an effect, with its uncertainty, is usually more informative than a yes-or-no test.

Building on what exists

Many of the strongest honors projects add a well-defined piece to an existing study. Priya's project used wetlands that had already been fenced, matched into pairs and monitored for water chemistry for four years. That gave her a design she could never have set up alone and allowed her to compare her turbidity measurements with earlier records. In exchange, her data joined the station's long-term database, with her name attached.

Building on existing work brings obligations. You must use the existing protocols where they apply, so that data are comparable; respect agreements with landowners and partners; and credit the people who created the data. Before starting, Priya read the station's data-use agreement, which stated that her thesis could use station data with acknowledgment and that any publication would include Dr. Okafor and the station's data manager as co-authors.

Aisha's revised question changed from cataloging resistance genes in soil bacteria to asking which resistance genes are found in multidrug-resistant isolates from manured and unmanured fields. It is narrower, cheaper and better, because it compares two conditions and so can test a hypothesis.

When the project changes

Questions often change after a project begins. A pilot study may show that a method does not work, an organism may be scarcer than expected, or early data may reveal something more interesting than the original question. Changing course is not failure, but it should be deliberate and documented. Priya kept a project log in which she recorded each major decision, the date and the reason. When her examiners later asked why she had added a dilution step to her eDNA analysis, the log showed exactly when and why the change was made, which made her answer convincing.