A hypothesis is an idea that can be tested by using scientific methods, such as by performing experiments or statistical analysis or both. A well-framed research hypothesis helps identify the most appropriate experimental design to adopt and the exact nature of data to collect so that it can be tested effectively. It helps make the research objective as clear as possible and is an informed guess about how the experimental results may answer a research question.

In this post, I discuss how to frame a good research hypothesis.

Find a research theme or question

The big question students often ask me is how to find ideas that can be tested. The answer is simple —start with a why. Ask yourself why something piqued your curiosity and why you want to study it.

The next step is to figure out how you would answer the research question. Try to inculcate two important practices that may help you frame an apt research hypothesis:

1. Honing your observation and critical-thinking skills

The power of observation is the ability to spot detail in things that others might overlook. Scientists have discovered many fundamental truths by critically analyzing certain observations. For example, observing the fall of an apple and thinking critically about the same helped Newton develop a hypothesis about the force of gravity. He eventually explained the fundamental reason behind why things fall by performing experiments and mathematical calculations.

Thus, keenly observing events and reflecting deeply on what caught your attention is an important way to practice the scientific method, and framing a good hypothesis is the first step to mastering it.

2. Developing the habit of reading scientific literature

A hypothesis emerges out of existing theories and available knowledge. So, spend time learning more about topics of your interest. A simple way to do this is to develop a habit of reading popular science articles, science magazines, scientific review articles, and research papers.

Reading scientific literature may draw your attention toward new emerging areas in research and deepens your understanding of the subject matter. This will enable you to ask original questions that open a fresh line of investigation. Once you have a question, you can read more literature and convert your question into a specific, focused, and testable hypothesis.

Understand variables

A hypothesis is centered around variables. Consider the hypothesis ​“manuring helps plants grow tall faster.” You can test it by conducting an experiment wherein manure is added to one set of plants and not added to another. Next, you collect data by measuring the heights of plants in both the sets and comparing them to see if they differ. Thus, your hypothesis keeps you focused on the specific trait that you intend to study (plant height) and how a variable (manure) influences it.

Hypothesis framing and testing happens around collecting data for objects, features, events, and patterns referred to as ​“variables” and the relationship between them. Variables are of two types.

The first type is an independent variable: that which you can control while performing an experiment. In the above example, manuring is an independent variable. You can use different types of manure, add different amounts, or even use combinations of several kinds of manure. Thus, you can modify the independent variable in many ways. The second type is a dependent variable: that which you measure in your experiment to collect data. In the above example, plant height is the dependent variable and thus cannot be changed or altered.

If you change the dependent variable, your research question also changes. For example, if you replace plant height with flowering, your research hypothesis changes to ​“manuring helps plants to flower faster.” Now, you will measure the rate of flowering rather than plant height and thus answer a new research question.

If you change the independent variable manuring with watering, the hypothesis may be rewritten as ​“Regular watering helps plants grow tall faster.” To test this hypothesis, you will still measure the plant height – the dependent variable – which is fixed.

Thus, a clear understanding of variables and their relationships is important to coming up with a workable hypothesis and to staying focused on your original research query.

Learn to use the if/​then format

Commonly, hypothesis statements are framed using the if/​then format. This suggests an underlying cause-effect relationship, meaning that one variable influences the other, for example, ​“If you eat vegetables and fruits daily, then you will develop strong immunity.”

Fine-tune your hypothesis

Now consider this statement: ​“Exposure to pollution has detrimental effects on skin.” Such a hypothesis is ineffective because it does not indicate what specifically to consider and study as a detrimental effect. This lack of clarity may lead to ambiguity in data collection. For example, one may consider a gamut of features to describe the harmful effects of pollution on skin, such as dryness, pigmentation, allergy etc. Hence, the research hypothesis is too broad and needs to be narrowed down.

Now consider this: ​“Exposure to pollution leads to acne and related skin conditions.” This hypothesis clearly indicates that the experimental design should involve a comparative study of acne in people who are exposed to pollution and those who are not. This fine-tuning of a research hypothesis is key to developing a robust methodology.

Know different types of hypotheses

1. Simple hypothesis: This describes the relationship between two variables – one independent and the other dependent.

Example:

Drinking tea may reduce iron absorption in the body.

2. Complex hypothesis: This involves more than two variables. The combination may go from two independent variables and one dependent variable or vice versa.

Examples:

Tea consumption and vitamin C deficiency can both individually reduce iron absorption in the body.

Tea consumption and vitamin C deficiency can both individually reduce iron absorption in the body, but differently in men and women.

3. Empirical hypothesis: This is a hypothesis that is tested based on an assumption. Whether the assumption is true or not is decided based on the interpretation of the collected data.

Example:

Masks can protect against all coronavirus variants equally.

4. Null (H0) and alternate (H1) hypotheses: A null hypothesis describes an absence of relationship between variables. It is called a null hypothesis because researchers collect evidence to nullify it.

Example:

The use of hair oil or hair growth serum does not influence the rate of hair loss in men.

A null hypothesis cannot be proved; it can only be rejected. Hence, it is mostly supplemented by alternative hypotheses. An alternative hypothesis states the opposite of the null hypothesis. For the above example, an alternate hypothesis may be written as follows:

The rate of hair loss is lower in men using hair growth serum than in those using hair oil.

Considering null and alternate hypotheses while designing your experiments is a way to minimize flaws and get precise/​reliable results. Proving an alternate hypothesis without disproving the null hypothesis is acknowledged as an unethical research practice. This is because experimental results are never absolute but rather the closest approximation. Hence, researchers cannot prove an alternative hypothesis with 100% confidence. Thus, it is imperative to collect evidence to reject the null hypothesis before one proves an alternate one.

Let us understand this using the above example. You first need to provide evidence that hair oil/​growth serum affects the rate of hair loss in men. Such evidence would refute the null hypothesis. The next step would be to collect data to compare the efficacy of hair growth serum vs hair oil for promoting hair growth in men (collecting evidence to support your alternate hypothesis).

5. Statistical hypothesis: This is statistically tested on a fraction or subset of the population to generate statistical evidence and the findings are extrapolated to the remaining population. Such a hypothesis holds true if verified statistically even if it does not fall within the reigns of logic.

Example:

Seventy-five percent of the Indian population is deficient in vitamin D.

6. Logical Hypothesis: This hypothesis uses logic to explain an observation or suggest a relationship between variables, but for which, extensive evidence may be lacking. In most cases, it might not be possible to gather evidence, yet a logical hypothesis is often not rejected.

Example:

A fixed sleep-wake pattern improves focus and increases productivity in students.

Finally, use the following guidelines to frame a good research hypothesis.

• Always adhere to ethics. Consider the ethical demarcation between what you should test vs what you can test. Your hypothesis must respect scientific responsibility and laws that protect socio-cultural and scientific norms.
• Define variables clearly. Readers are able to visualize the experimental design if the relationship between variables is clearly described.
• Frame the hypothesis such that it is clear whether a cause-effect relationship is being explored.
• Account for testability. A hypothesis is an idea that can be tested, meaning it can be proved or disproved. If an idea, thought or observation cannot be tested within the confines of the scientific method, then it forms a weak or forced hypothesis. Thus, a hypothesis must allow the researcher to experimentally manipulate or control an independent variable.
• Use simple, clear, and concise language to write a hypothesis. It must be free of complex jargon.
• Make sure your hypothesis can answer a question in a way that adds value to the existing knowledge.