03 · What You Need to Know
Scientific Research Has Shared Principles Without a Single Recipe
The familiar scientific method is a simplified model
The classroom version of the scientific method is useful because it introduces several important ideas. Researchers ask questions, develop possible explanations, gather evidence, analyze what they find, and revise conclusions in response to evidence.
The difficulty begins when this educational model is interpreted literally as the sequence followed by all working scientists.
The National Academies has explicitly rejected that interpretation. In Reproducibility and Replicability in Science, it states that scientists do not follow one fixed set of steps leading inevitably to scientific knowledge. A National Research Council framework for science education similarly cautions against the impression that there is one distinctive approach common to all science.
UC Berkeley's Understanding Science project was developed partly to correct the same misconception, portraying science as dynamic and creative rather than as the linear sequence frequently presented in textbooks.
Real research rarely moves neatly from Step 1 to Step 6
Consider what happens during an actual study. A researcher begins with a question, reads the literature, and realizes the question needs revision. Preliminary observations reveal an unexpected pattern. That pattern suggests another explanation. The available measurement turns out to be inadequate, requiring instrument development. Analysis raises a new question that sends the researcher back to theory.
None of this means the research has failed to follow the scientific method. It means research is iterative.
Scientific work often contains feedback loops. Evidence can modify questions. New theory can change what researchers decide to measure. Unexpected results can generate exploratory analyses. Replications can challenge assumptions that appeared settled. Technological developments can make previously inaccessible questions investigable.
The National Academies' public explanation of science captures this recurring movement through questioning, testing through observation or experimentation, confirmation, and revision rather than presenting scientific knowledge as the output of a one-way algorithm.
Different questions require different methods
The strongest reason there cannot be one universal research procedure is that research questions differ fundamentally.
An experimental researcher might manipulate an intervention and compare outcomes. An astronomer observes phenomena that cannot be experimentally rearranged. An epidemiologist may analyze naturally occurring exposures. A paleontologist reconstructs processes from surviving evidence. A computational scientist may develop simulations. A qualitative researcher may examine participants' experiences through interviews and observations. A historian may investigate archival records.
Forcing all of these inquiries through an identical procedural sequence would not make them more scientific. In some cases, it would make the method less appropriate to the question.
The National Research Council has noted that scientists use a wide range of methods to investigate phenomena and develop hypotheses, models, and theories rather than relying on one universally employed scientific method.
Experiments are important, but they are not universal
Experiments are especially valuable when researchers need to estimate causal effects and can manipulate relevant conditions ethically and feasibly. Experimental control and random assignment can sometimes provide powerful protection against alternative explanations.
But many scientific questions cannot be studied experimentally. Researchers cannot randomly assign people to many harmful exposures. They cannot experimentally recreate extinct ecosystems, planetary formation, historical events, or naturally occurring disasters simply to satisfy a methodological template.
Scientists instead use methods suited to the evidence available and the claims they seek to make. This is why research does not universally require an experiment or hypothesis.
Hypotheses are important without being the compulsory starting point of every study
The familiar scientific-method diagram often places "form a hypothesis" immediately after "ask a question." That arrangement fits confirmatory research in which investigators have a sufficiently developed theoretical or empirical basis for specifying predictions.
Other research is exploratory. Researchers may initially be trying to identify patterns, characterize a phenomenon, develop concepts, generate hypotheses, or determine which explanations deserve subsequent testing.
Observation can therefore precede a formal hypothesis. Theory can precede observation. An unexpected observation can modify theory. Exploratory analysis can generate a hypothesis that a later study tests confirmatorily.
Scientific reasoning involves these relationships without requiring every individual project to begin with a directional prediction.
Observation is not merely the stage before experimentation
Simple scientific-method diagrams can inadvertently make observation look preliminary: researchers observe something, formulate a hypothesis, and then move on to the supposedly more decisive activity of experimentation.
Observation is itself central to many sciences. Researchers systematically observe celestial bodies, geological formations, ecosystems, disease patterns, behavior, and numerous other phenomena. Scientific observation can involve sophisticated instruments, measurement protocols, classification systems, and theoretical assumptions about what should be recorded.
Research in biology likewise encompasses experimentation, observation, exploration, description, technology development, and hypothesis testing, with theory informing these different activities.
Observation should therefore not be confused with casual looking. Scientific observations are structured by questions, methods, instruments, concepts, and standards for evidence.
Models and simulations can play central roles in scientific reasoning
Some research investigates phenomena partly through models rather than through direct manipulation of the phenomenon itself. Climate science, epidemiology, physics, economics, ecology, engineering, and other fields use mathematical and computational models to represent systems, explore mechanisms, generate predictions, compare scenarios, and evaluate explanations.
Models do not remove the need for empirical evidence. Their assumptions, parameters, outputs, and predictions must be evaluated against relevant observations and knowledge. But their role demonstrates again why science cannot be adequately represented as "hypothesis, experiment, conclusion."
Theory, modeling, observation, experimentation, and analysis interact differently depending on the research problem.
Scientific methods are plural, but science is not methodologically arbitrary
If there is no universal scientific method, one might conclude that researchers can simply use whichever procedures they prefer. That does not follow.
Methodological pluralism means that several forms of investigation may be scientifically legitimate. It does not mean every method is appropriate for every question.
The National Academies notes that although scientists do not follow one fixed sequence, their work shares important principles, including the use of ideas and theories, reliance on evidence, logic and reasoning, and communication of results.
In educational research, the National Research Council similarly described six interrelated principles of scientific inquiry: posing significant empirically investigable questions, connecting research to theory, using methods that directly investigate the question, maintaining a coherent chain of reasoning, pursuing replication and generalization across studies, and disclosing research for professional scrutiny. Crucially, it described these as guiding principles rather than an algorithm.
This distinction is fundamental. Scientific research can have methodological diversity while retaining standards for what counts as defensible evidence and inference.
The method should fit the claim
A method becomes appropriate partly because of what the researcher wants to conclude.
If researchers want to estimate prevalence in a population, their sampling and measurement strategy must support that estimate. If they want to make a causal claim, the design must address plausible alternative explanations. If they want to understand lived experience, they need evidence capable of capturing that experience and an analytical approach suited to interpreting it.
A method that is rigorous for one question may be inadequate for another. A randomized experiment can estimate certain causal effects but may tell researchers relatively little about how participants interpret an intervention. An in-depth interview study can illuminate experience but generally cannot estimate population prevalence from a small purposively selected sample.
Scientific rigor therefore depends less on whether a study resembles a familiar diagram and more on whether the method is capable of supporting the intended inference.
Research methods differ within the same discipline
Methodological diversity is not merely a difference between physics and sociology or between quantitative and qualitative research. Researchers within a single field may use experiments, observational studies, simulations, longitudinal designs, surveys, case studies, secondary datasets, systematic reviews, or other methods.
Different methods can also investigate the same phenomenon from different angles. One study may estimate an effect, another identify a mechanism, another test whether the effect replicates, and another determine whether it persists under different conditions.
Scientific knowledge is often strongest when different methods produce converging evidence rather than when every researcher repeatedly applies the same procedure.
Individual studies are only part of the scientific process
The stepwise scientific-method model tends to end with "draw a conclusion." Real science does not.
A study becomes part of a larger research record. Other researchers may criticize its assumptions, attempt replication, collect evidence in another population, apply different methods, conduct a systematic review, develop a competing explanation, or discover evidence that requires the original conclusion to be qualified.
The National Academies emphasizes that confidence in scientific results develops through multiple studies and continued testing rather than through a single investigation. Its account of science includes confirmation and revision as integral to how knowledge changes.
This is why confirmatory and replication research is not an optional afterthought to discovery. It is part of the broader process through which claims become more or less credible.
Scientific knowledge remains open to revision
A rigid method can create another misconception: if researchers follow the correct steps, the result must be true.
No research procedure provides that guarantee. Studies contain uncertainty. Measurements can be imperfect. Samples may not represent every context. Assumptions can be mistaken. Previously unknown variables can matter. New evidence can expose limitations in an accepted explanation.
This does not mean scientific knowledge is merely opinion or that every established conclusion is perpetually in equal doubt. Some explanations are supported by extensive converging evidence. Rather, scientific claims remain in principle responsive to sufficiently strong new evidence. Recent National Academies guidance likewise emphasizes that science contains methodological diversity while relying on shared standards that allow knowledge to advance through iteration, disagreement, and uncertainty.
“The scientific method” and “scientific methods” are useful distinctions
There is nothing inherently wrong with using the phrase scientific method when it refers broadly to disciplined ways of developing and testing scientific explanations. Problems arise when the singular phrase is interpreted as one mandatory sequence.
The scientific method as a classroom model
A simplified sequence such as question, hypothesis, experiment, analysis, and conclusion that illustrates one recognizable pattern of scientific investigation.
Scientific methods in research practice
Diverse methods of observation, experimentation, measurement, modeling, comparison, analysis, replication, and other forms of systematic investigation selected according to the question and field.
Shared scientific principles
Commitments concerning evidence, logical reasoning, systematic investigation, methodological justification, transparency, uncertainty, and critical scrutiny that constrain how scientific claims are developed and evaluated.
The distinction allows us to preserve what is valuable in introductory models without mistaking the teaching diagram for a universal description of how research actually proceeds.
Not all scholarly research needs to be described as scientific
There is one final complication. The terms research and scientific research overlap, but scholarly research also occurs in fields whose methods are not always described as scientific in the conventional empirical-science sense.
Historical scholarship, philosophical inquiry, legal research, literary studies, and some forms of theoretical work may employ rigorous and systematic methods appropriate to their disciplines without presenting themselves as applications of a scientific method.
Consequently, asking whether all research follows one scientific method is doubly problematic. There is no single method followed by all sciences, and research itself extends beyond activities ordinarily classified as science.
