Identifying the Right Research Problem for Experimental Research

Every strong experiment begins with a strong question. But here is a fact that often surprises new researchers: not every research question can be answered through an experiment. Some questions are perfect for experimental investigation, while others demand surveys, case studies, or purely observational methods. Knowing the difference is the first and most important skill in experimental research. This article explains how to identify a research problem that genuinely suits an experiment, with examples drawn from library and information science and the wider social sciences.

Understanding research problems in experimental research
Where experimental problems appear in library and social science research
Suitable versus unsuitable problems
Key criteria for experimental problems
The variables must be manipulable
The variables must be measurable
Extraneous variables must be controllable
The problem must allow testable hypotheses
The study must be feasible and ethical
Bringing the criteria together

Understanding research problems in experimental research

An experiment is a method designed to establish cause and effect. The researcher changes one factor and watches what happens to another, while keeping everything else as constant as possible. This is why experimental research is best suited to explanatory questions rather than descriptive or exploratory ones. If your goal is simply to describe how often something occurs, or to explore a new area where little is known, an experiment is usually the wrong tool.

The central feature that separates experimental research from other approaches is control. In experimental work, the researcher directly controls and changes the independent variable, then measures the effect on the dependent variable. In non-experimental research, the researcher only observes and measures variables as they naturally occur, without manipulating them. This level of control is what allows an experiment to rule out alternative explanations and point to a genuine causal relationship.

So before anything else, ask yourself a simple question about your topic: Can I actually change something and observe the result? If the answer is yes, you may have an experimental problem on your hands. If the answer is no, you probably need a different method.

It is a common myth that experiments only belong in laboratories with test tubes and white coats. In reality, library science and the social sciences are full of testable cause-and-effect questions. Consider these examples:

Instructional methods: Does a gamified information literacy module help students retain database searching skills better than a traditional lecture? Here you can deliberately assign one group to each method and compare outcomes.

Library environment: Does lighting level affect reading comprehension in a study area? You can set up bright, moderate, and dim conditions and measure comprehension scores.

Service design: Does a mobile notification system reduce the number of overdue returns compared to email reminders? You can roll out each system to different user groups and track the results.

In each case, something is being deliberately changed by the researcher, and something measurable is expected to respond. That is the signature of an experimental problem. Surveys in library and information science still dominate, and quantitative LIS studies often rely on descriptive statistics rather than true experiments. But where a causal claim is needed, the experiment remains the strongest design available.

Suitable versus unsuitable problems

A quick way to test a research problem is to compare suitable and unsuitable examples side by side.

Suitable: “Does extending library opening hours during exam season increase student footfall?” Opening hours can be changed by the institution, and footfall can be counted. The relationship is causal and the variables are controllable.

Unsuitable: “What is the relationship between a student’s family income and their reading habits?” Income cannot ethically or practically be manipulated by a researcher. This is a correlational question best answered through a survey or observational study, not an experiment.

The difference is not about the topic being interesting or important. It is about whether the key variable can be deliberately and ethically changed by the researcher.

Key criteria for experimental problems

Once you suspect a problem might suit an experiment, you should test it against a clear set of criteria. A problem that satisfies all of these is far more likely to produce valid, meaningful results.

The variables must be manipulable

This is the cornerstone of all experimental research. The independent variable must be something you can change or control. The independent variable is the factor the experimenter manipulates to determine its influence, while the dependent variable is the response that is measured. It is called “independent” precisely because the researcher decides what values or conditions to create.

In the lighting study mentioned earlier, lighting level is the manipulated independent variable, and reading comprehension is the measured dependent variable. If a variable cannot be changed by the researcher, the study cannot be a true experiment, no matter how interesting the relationship may be.

The variables must be measurable

Both the independent and dependent variables need to be defined in ways that allow consistent, accurate measurement. Researchers call this operationalisation, turning an abstract idea into something countable. “Reading comprehension” is vague on its own, but it becomes measurable when defined as the score on a standardised comprehension test. “Student engagement” becomes measurable when defined as time spent on a task or number of resources accessed. Without reliable measurement, any conclusion you draw will be impossible to defend.

Extraneous variables must be controllable

An experiment only proves cause and effect if competing explanations can be ruled out. Extraneous variables are the outside factors that could secretly influence your results. In the lighting study, noise levels, the time of day, and participant fatigue could all affect comprehension scores. A well-chosen experimental problem is one where these factors can be held constant or accounted for.

The strongest tool for this is random assignment, where participants are placed into groups by chance rather than by choice. Random assignment reduces the potential for confounding variables by ensuring the groups are comparable at the start. When random assignment is possible alongside manipulation and a control group, you have the conditions for a true experimental design. When you can manipulate the variable but cannot randomly assign participants, you are working with a quasi-experimental design that resembles an experiment but cannot fully eliminate confounding variables. This distinction matters a great deal when you frame your problem, because it determines how confidently you can claim causation.

The problem must allow testable hypotheses

Experimental research revolves around hypothesis testing. A good experimental problem should lend itself to a clear, testable prediction about how the variables relate. Hypotheses often take an “if-then” form: if I manipulate X, then Y will change in a specific way. For example, “if the library extends its operating hours during exam periods, then student usage will increase significantly.”

A strong hypothesis has three qualities. It should state an expected relationship between the variables, such as predicting that gamified instruction produces higher retention than lecture-based instruction. It should be falsifiable, meaning it is capable of being proven false through observation, which is a fundamental principle of scientific inquiry. And it should be specific rather than vague, because the more precise your prediction, the more clearly you can design your study and interpret your findings.

Behind every hypothesis test sits a pair of competing statements. The null hypothesis states that no significant effect or relationship exists, while the alternative hypothesis states that an effect does exist. Researchers then use a significance level, usually set at 0.05, as the threshold for deciding whether the evidence is strong enough to reject the null hypothesis. If your research problem cannot be expressed as a contest between these two positions, it may not be suited to experimental testing at all.

The study must be feasible and ethical

A problem can satisfy every scientific criterion and still be a poor choice if it cannot be carried out responsibly. Feasibility covers the practical realities: the time available, the funding, and access to enough participants. Experimental studies with multiple phases can take months or even years, so a problem that demands resources you do not have is not a workable problem.

Ethics is equally decisive. Experiments that manipulate variables affecting people require extra scrutiny to protect participant welfare. Even a valuable research idea does not justify violating the rights or dignity of participants. Principles such as voluntary participation, informed consent, confidentiality, and the right to withdraw must be built into the design from the start. In an academic library setting these concerns are usually manageable, but they still shape what you can and cannot ethically manipulate. This is exactly why a question like manipulating a student’s family income is ruled out before the study even begins.

Bringing the criteria together

Identifying the right research problem is really a process of filtering. Start with a question you care about, then run it through each test in turn. Can the key variable be manipulated? Can both variables be measured reliably? Can extraneous influences be controlled? Can the problem be stated as a testable, falsifiable hypothesis? And can the whole study be done within your resources and ethical limits?

A problem that passes all five filters is a genuine candidate for experimental research. A problem that fails even one of them is not a bad problem, it is simply a problem for a different method. Recognising this early saves enormous effort and protects the credibility of your final results. The discipline of choosing well at the start is what gives an experiment its unique strength: the ability to link cause and effect with confidence that few other methods can match.

What do you think? Looking at a topic you find interesting in library or information science, could you reframe it as a question with a variable you are genuinely able to manipulate? And where might the line fall for you between a question that deserves a true experiment and one better served by a survey or observational study?

How useful was this post?

Click on a star to rate it!

Average rating 0 / 5. Vote count: 0

No votes so far! Be the first to rate this post.

References

PDF 📄

Table of Contents

Understanding research problems in experimental research 🔗

Where experimental problems appear in library and social science research 🔗

Suitable versus unsuitable problems 🔗

Key criteria for experimental problems 🔗

The variables must be manipulable 🔗

The variables must be measurable 🔗

Extraneous variables must be controllable 🔗

The problem must allow testable hypotheses 🔗

The study must be feasible and ethical 🔗

Bringing the criteria together 🔗