There is no single “scientific method,” instead there are common principles. These include observing, forming questions, proposing hypotheses, testing through experiments, and analysing results. The cycle may repeat many times.
Different fields adapt these steps to suit their subject. Astronomers cannot experiment with stars, so they rely on observation and modelling. Biologists often use controlled experiments. Social scientists may use surveys or interviews. What they have in common is an underlying rigour, and a commitment to careful testing, verification and reporting.
Scientific rigour means being thorough, careful, and transparent. Methods are scrupulously recorded, including details of what materials, implements, software and other parameters were used. The process of statistical analysis and tabulation of findings will also be described. This enables others to repeat the study or experiment and check whether the results hold true. Without good records, findings cannot be verified.
Credible scientific papers will describe methods in detail. The management of research materials including lab notebooks, research protocols, and data archives is also important, as they capture and present the pathway from question to conclusion.
Rigour also builds trust. When we know how a study was done, we can judge its strengths and weaknesses and use its findings with greater confidence.
Statistics are the tools scientists use to make sense of data. They help distinguish patterns from chance, and signal how confident we can be in results.
For example, if two groups eat different diets and one group shows better health, statistics test whether the difference is likely real or just random variation. They also estimate how strong the effect is.
Ultimately, statistics give an indication rather than absolute proof. They help weigh evidence, showing whether results are meaningful and reliable.
A meta-analysis combines the results of many individual studies to see the bigger picture. By pooling data, it increases statistical power and reveals whether findings are consistent.
For example, dozens of small studies on a specific nutrient may give mixed results. A meta-analysis can bring them together, showing overall trends and clarifying what the evidence really suggests.
Meta-analyses are powerful tools for developing guidelines, because they summarise broad evidence rather than relying on one study alone.
About science
How science works
Research standards and credible evidence
Identifying credible science
Money and science
Science begins with curiosity. This step involves identifying a specific, testable question based on observations or gaps in knowledge. For example, "Why do plants grow faster in some soils?" The question should be clear and focused to guide the investigation. A good question asks for a measurable answer and may lead to a hypothesis.
Before experimenting, scientists gather existing information on the topic. This involves reading studies, articles, or data to understand what is already known and inform new work. For instance, researching background information on the nitrogen cycle might reveal factors potentially affecting plant health. This helps refine the research question and informs the hypothesis.
A hypothesis is a speculative idea or a testable prediction about a possible answer to a research question. It’s often phrased as an "if-then" statement, like, "If plants receive more nitrogen, then they will grow taller." The hypothesis is based on research and sets the stage for testing.
An experiment tests the hypothesis under controlled conditions. Scientists design experiments to isolate variables (for example, changing only the amount of added nitrogen while keeping water and sunlight constant). They collect data through measurements or observations. For example, measuring the height and heath of plants grown with varying amounts of added nitrogen.
During and after the experiment, scientists gather data such as plant heights, root system health or quantity of useable biomass produced. They analyse this data using tools like graphs, statistics, or comparisons to identify patterns or trends. For instance, calculating the average growth rate of plants in each test group to see if the hypothesis holds.
Based on the data analysis, scientists determine whether the hypothesis was supported or not. For example, if plants with more nitrogen grew taller, the hypothesis is supported. If not, scientists consider why and what it means. This step summarises what was learned and whether the question was answered.
Scientists share their findings through papers, presentations, or reports. This allows others to review, replicate, or build on the work. For example, publishing a study on how nitrogen affects plant growth. Clear communication ensures the scientific community and public can benefit from the findings.
Science values repeatability. Other researchers may repeat the experiment to confirm the results or test them under different conditions. Replication strengthens confidence in findings. For instance, another scientist might test the nitrogen experiment in a different climate or with different plant varieties to verify the results.
In science, experiments are the hands-on ways we test hypotheses and gather evidence. While the scientific method provides a framework, the type of experiment chosen depends on the research question, ethical considerations, and practical constraints.
The best experiment depends on the goal. Controlled experiments excel in precision for basic science, while observational or natural ones are better for exploratory research or where there are ethical constraints. Many studies combine elements, like using observational data to inform a controlled follow-up. Or, for example in medicine, controlled experiments (like clinical trials) test new drugs, while observational studies track side effects in the general population.
No experiment is perfect, because factors such as bias, sample size, and measurement errors can affect outcomes. That's why replication and peer review are crucial for robust research.
These are typically done in a lab or controlled setting where the researcher actively manipulates one or more variables (the independent variable) while keeping everything else constant. For example, testing how different light levels affect plant growth by growing identical plants in separate rooms with varied lighting, while controlling soil, water, and temperature.
Pros:
High level of control reduces outside influences, making it easier to establish cause-and-effect relationships.
Results are often repeatable and reliable, as conditions can be precisely replicated.
Allows for randomisation (e.g., assigning subjects randomly to groups) to minimise bias.
Cons:
May not reflect real-world complexity, leading to results that don't apply outside the lab.
Can be expensive and time-consuming to set up controlled environments.
Ethical issues may arise, especially with human or animal subjects, if manipulation causes harm.
Here, researchers observe and record phenomena as they occur naturally, without interfering or manipulating variables. For instance, watching bird migration patterns in their natural habitat to study environmental impacts, or surveying people's diets and health outcomes over time. These can be cross-sectional (a snapshot at one time) or longitudinal (over a period).
Pros:
Ideal for studying things that can't be ethically or practically manipulated, like the effects of smoking on people’s long-term health.
Often more realistic, as they capture natural behaviours and real-world conditions.
Can be less invasive and more cost-effective, especially for large-scale or long-term data collection.
Cons:
Harder to prove cause-and-effect because other uncontrolled factors (confounders) might influence results, and correlation doesn't equal causation.
Potential for observer bias, where the researcher's expectations affect what they record.
Results may not be as repeatable, as natural conditions can vary unpredictably.
Similar to controlled experiments but conducted in real-world settings rather than a lab. For example, testing a new teaching method by implementing it in some classrooms (treatment group) while others use the standard approach (control group), then comparing student performance.
Pros:
Balances control with realism, making findings more applicable to everyday life.
Can reveal unexpected real-world interactions that lab settings miss.
Useful for social sciences, ecology, or applied research where lab isolation isn't feasible or appropriate.
Cons:
Less control over external variables such as weather or participant behaviour, which can introduce unreliable information or “noise” into the data.
Logistically challenging and potentially more expensive due to real-world variables.
Ethical concerns if the experiment affects people or environments without full consent or awareness.
These occur when external events or policies create "experimental" conditions without the researcher's involvement. For example, studying the impact of a new law on public health by comparing regions where it's implemented versus those where it's not. No direct manipulation happens; nature or society provides the variation.
Pros:
Highly ethical, as no one is deliberately exposed to treatments.
Can provide insights into large-scale phenomena, like economic policies or natural disasters, that couldn't be ethically replicated.
Often cost-effective, relying on existing data or events.
Cons:
Limited control means it's hard to isolate causes, and results might be influenced by unrelated factors.
Opportunities are rare and unpredictable, so most research happens after an event or incident.
Data quality can vary, andit's challenging to ensure groups being compared are truly similar.
