Controlled Experiment – Definition, Process, Importance, Examples

A controlled experiment tests whether a deliberate change affects an outcome while other relevant influences are kept as comparable as possible. By comparing experimental conditions with an appropriate control or baseline, researchers can judge more confidently whether the manipulated factor produced the observed difference.

What Is a Controlled Experiment?

A controlled experiment is a scientific experiment in which a defined factor is deliberately changed and the effect of this change on a measurable response is observed, while other relevant factors are kept similar or managed by the experimental design.

It is basically a comparison performed under controlled conditions. Here, the term “controlled” means that the factor being tested is separated as much as possible from other changes that may also affect the result. It does not simply mean an experiment “done in a laboratory”.

In science and biology, a controlled experiment has a defined manipulation, measurable outcome, a control or another meaningful comparison, and management of the possible confounding factors.

Basic Logic of Experimental Control

The basic logic of experimental control is based on manipulation and comparison. A defined independent variable is changed and its dependent variable or response is measured.

The treatment condition is compared with a control or reference condition. In this condition, the factor being tested may be absent, kept unchanged, or present at another defined level.

Other variables are kept constant wherever possible. When this is not possible, methods such as randomization, blocking, or other related experimental designs can be used to keep unwanted variation from getting mixed with the treatment effect. The intended manipulation must differ between the conditions.

Controlled Experiments in Biology and Laboratory Science

In biology and laboratory science, controlled experiments are used to separate the effects of a treatment, genotype, nutrient, temperature, time, or another experimental factor from unwanted variation.

Positive and negative controls are also used when required for the interpretation of the experiment.

A laboratory setting makes it easier to control conditions such as medium, temperature, handling, timing and other factors. But a laboratory experiment and a controlled experiment are not the same thing. A laboratory experiment can still have confounding when important conditions change together with the treatment.

Controlled experiments can also be performed outside the laboratory. These are field experiments, where an intervention is introduced under a natural setting and compared with an appropriate control or comparison condition.

How to Tell Whether an Experiment Is Well Controlled

A well-controlled experiment has a defined manipulation and a measurable response. The experimental condition and comparison condition should remain comparable except for the factor or factors that are deliberately changed.

Appropriate controls are present. Possible confounders are managed by keeping conditions constant, standardization, random assignment, blocking, blinding, or another suitable method depending on the experiment.

If several uncontrolled differences occur along with the treatment, the observed response cannot be clearly separated from the effects of those other factors.

Components of a Controlled Experiment

The following are some of the important components of a controlled experiment:

1. Research question or hypothesis- A controlled experiment starts with a defined research question about the effect of one factor on another. In hypothesis-based research, the observations required to support or reject the hypothesis are decided before carrying out the experiment.
2. Independent variable- It is the factor which is deliberately changed or manipulated by the researcher. The independent variable may be concentration of a nutrient, temperature, drug treatment, light exposure, or a particular growth condition. In a simple controlled experiment, it should be the major planned difference between the conditions being compared.
3. Dependent variable- The dependent variable is the measurable response observed after changing the independent variable. It can be growth rate, enzyme activity, cell number, survival, gene expression, or another measurable outcome depending on the experiment. The method of measurement has to be properly defined so that the same response can be recorded from both experimental and control conditions.
4. Experimental or treatment group- This is the group that receives the factor or treatment being investigated. Several treatment groups can also be prepared when different concentrations, doses, genotypes, or other experimental conditions are compared.
5. Control group or control condition- It provides a reference for comparison with the experimental treatment. A negative control normally lacks the factor being tested and can be used to detect background effects or artifacts. Positive controls are used when it is necessary to confirm that the experimental procedure and measurement system are working as expected. The same type of control is not required for every experiment.
6. Controlled variables or standardized conditions- The factors other than the variable under study are kept as similar as possible between the groups. Temperature, incubation time, sample handling, medium, measurement method and other relevant conditions may need to be controlled. If another factor changes along with the treatment, it may become a confounding variable and its effect becomes difficult to separate from the treatment effect.
7. Experimental units and replication- Experimental units are independent samples, organisms, cultures, plots, or other units on which the treatments are applied. Several independent replicates are generally used for estimating biological variation. Repeated measurements taken from the same experimental unit are not automatically considered as separate biological replicates. Treating them as independent replicates can result in “pseudoreplication”.
8. Randomization- Experimental units or treatments are assigned randomly where the experimental design allows it. Randomization is used to prevent unwanted differences in space, time, batches, sample characteristics, or other factors from becoming systematically associated with a particular treatment group. Such differences may occur even in apparently uniform laboratory experiments.
9. Blinding- In some controlled experiments, the person receiving the treatment, performing the measurement, or analysing the result is kept unaware of the treatment identity. It is especially useful when the expectations of participants or investigators can influence the observations. Blinding is not possible or required in every type of laboratory experiment.

How to Design and Conduct a Controlled Experiment

The following are the major steps used to design and conduct a controlled experiment:

1. Define the research question- The experiment begins with a clear question or hypothesis that can be tested by measurement. Decide what factor will be changed and what response will be recorded. The main outcome to be measured should preferably be decided before starting the experiment.
2. Identify the experimental variables- Select the independent variable, which will be deliberately manipulated, and the dependent variable whose response will be measured. Other factors which can affect the dependent variable should also be identified. These may later become controlled variables or possible confounding factors.
3. Set up treatment and control conditions- An experimental group is given the defined treatment, while an appropriate control condition provides the reference for comparison. Depending on the experiment, negative control, positive control, untreated control, vehicle control, or another suitable comparison can be selected. The treatment and control conditions should differ mainly in the experimental factor being tested.
4. Decide what should be controlled- Factors other than the intended treatment that can reasonably change the measured response are kept similar between the groups. In biological experiments these may include temperature, pH, light, nutrient composition, sample age, incubation period, volume, humidity, handling method, time of measurement and equipment used. The variables requiring control depend on the biological system and the question being tested.
5. Select experimental units and replication- The organisms, cultures, cells, samples, plots, or other units receiving a treatment are defined before the experiment. More than one independent experimental unit is generally used for each condition. Replication gives an estimate of the variation between experimental units and should not be confused with repeated measurements of the same sample.
6. Assign the experimental conditions- Where possible, experimental units are randomly assigned to the treatment groups. Randomization prevents a particular location, batch, time, cage, plate position or another uncontrolled factor from repeatedly occurring with one treatment. If an important source of variation is already known, samples can first be arranged into suitable blocks and randomized within them.
7. Maintain the experimental conditions- In this step, the same procedure is followed for the control and experimental groups except for the planned manipulation. Temperature, incubation time and other controlled factors are maintained at the selected values. Equipment settings, reagent preparation, sample handling and measurement procedures are also kept comparable. Samples from different treatments should not be separated unnecessarily into different days or processing batches, because a “batch effect” can then become mixed with the treatment effect.
8. Apply the defined treatment- The independent variable is changed according to the experimental design. Its amount, concentration, duration, timing, or other treatment level should be applied in the same defined manner to the required experimental units. Any change from the planned procedure is recorded.
9. Measure the response- The dependent variable is measured using the same method and measurement criteria for all comparable groups. Where observer expectation can influence measurement, treatment identities can be blinded during data collection or analysis. This is especially useful for measurements containing a subjective component.
10. Record observations and experimental conditions- Data are recorded as they are obtained. Treatment identity, time, environmental conditions, unexpected events and excluded samples are also documented where relevant. Criteria for removal or exclusion of data are preferably defined before examining the experimental result.
11. Analyze the control and treatment data- Measurements from the treatment condition are compared with those of the appropriate control. The analysis is selected according to the experimental design, type of data and number of groups being compared. Effects of blocking or other planned design factors are included where required.
12. Repeat when required- Independent repetition can be used to examine whether the observed experimental result is obtained again under comparable conditions. Biological replication, randomization and proper control of experimental conditions help to separate treatment effects from ordinary experimental variation.

Controlled Experiment Examples

The following are some examples of controlled experiments in biology and laboratory science:

• Effect of nitrogen addition on plant growth- A controlled experiment can be set up to test whether additional nitrogen affects the growth of seedlings. Plants of the same species are grown under comparable conditions. The control plants receive the basal nutrient treatment without added nitrogen, while the experimental plants are supplied with a defined amount of additional nitrogen. Here, nitrogen addition is the independent variable and plant growth is the dependent variable.The experiment can be arranged as follows:
  • Question- Does additional nitrogen affect seedling growth?
  • Control group- Seedlings receiving the same basal treatment, but without the added nitrogen.
  • Experimental group- Seedlings receiving the defined nitrogen treatment.
  • Measured response- Shoot dry mass, plant height, or another growth measurement selected before the experiment.
  • Controlled conditions- Plant species, growth period, pot size, growth medium, water supply, light conditions and temperature are maintained similarly between the groups.
  • Several independent plants are used in each condition. Pot positions can also be randomized so that one treatment is not always kept at a particular position where light or temperature may be different. Replication and randomization are important parts of comparative biological experiments because uncontrolled variation can otherwise become associated with the treatment.
• Effect of an antibiotic on bacterial growth- A bacterial culture can be divided into control and treatment groups. The treatment culture receives a defined concentration of antibiotic, while an appropriate control is prepared without the antibiotic (or with the same solvent when a solvent is used). Same bacterial strain, inoculum, growth medium, incubation time and temperature are maintained. Bacterial growth is then measured, such as by optical density or viable cell count.
• Effect of light intensity on photosynthesis- An aquatic plant can be exposed to different defined light intensities and its photosynthetic response measured from oxygen production. Water temperature, plant material, exposure period and carbon dioxide availability are kept comparable. Light intensity is changed. The measured oxygen production becomes the response variable.
• Effect of temperature on enzyme activity- Equal enzyme-substrate mixtures are incubated at selected temperatures, while pH, enzyme concentration, substrate concentration and reaction time are maintained under defined conditions. The rate of product formation or substrate disappearance is measured. In this experiment, temperature is the manipulated factor.
• Effect of a nutrient on microbial growth- Microbial cultures can be grown in the same basal medium, with one group receiving the nutrient being tested and another suitable group without its addition. Equal inoculum and incubation conditions are used. Growth after the defined period can then be compared between the conditions. A controlled biological experiment requires defined independent, dependent and standardized variables rather than simply changing a treatment and observing what happens.

Importance of Controlled Experiments

Controlled experiments are important because a measured change can be compared against a suitable reference while other possible influences are managed. Some of the important advantages are:

• Tests the effect of a defined factor- In a controlled experiment, the factor of interest is deliberately changed and its effect is measured. The treatment and control conditions are made comparable except for the experimental factor being studied. This gives a stronger basis for relating the observed response with that manipulation.
• Reduces the effect of confounding variables- Other factors can produce differences that appear to be caused by the treatment. These are referred to as confounding variables. Control of relevant conditions, randomization, blocking and a balanced experimental design can prevent such unwanted factors from becoming associated with one particular treatment. Even laboratory experiments can give misleading results when this is not considered.
• Provides a reference for comparison- A control group or control condition gives a baseline against which the experimental response can be compared. Negative controls can also reveal background effects, contamination, nonspecific responses or other unexpected sources of error. Positive controls are useful for checking whether the experimental system is capable of producing the expected response.
• Improves internal validity- Experimental results can be affected by selection, handling, measurement and investigator-related biases. Randomization and blinding are used where suitable to reduce these effects. When such sources of bias are managed, differences between the experimental groups are less likely to arise from the way the experiment was conducted.
• Separates biological effects from experimental noise- Variation may come from batches, position, time, environmental conditions or technical procedures rather than the biological factor under study. A controlled design distributes or limits these sources of variation. Replication is then used to measure the variation occurring among independent experimental units.
• Makes experimental comparisons more reliable- If one treatment is processed on one day and its control on another, for example, a day or batch difference can become mixed with the treatment effect. Processing comparable groups under balanced conditions and randomizing their order reduces this problem.
• Supports reproducibility- Proper controls, randomization, replication and clear experimental procedures make it possible to examine whether a result occurs again under comparable conditions. These design practices are important for reproducible biological research and for detecting results produced mainly by methodological bias or uncontrolled variation.

Limitations of Controlled Experiments

Some of the important limitations of controlled experiments are:

• Complete control is difficult- All experimental variables cannot always be kept under control. Biological systems contain variation between organisms, samples, days, batches and environmental conditions. Some of these factors may not even be known when the experiment is designed. Randomization and blocking can reduce their influence, but unwanted variation can still remain.
• Highly controlled conditions may be less representative- Laboratory conditions are commonly made uniform to reduce experimental variation. But the natural environment is not uniform. A treatment giving a particular response under one standardized condition may behave differently when temperature, diet, housing, microbial environment or other conditions change. Treatment-environment interactions have been shown to produce results having poor external validity in highly standardized animal experiments.
• Results cannot always be generalized to a larger population- Random assignment of experimental units helps in comparing the treatment groups, but it does not itself make the experimental sample representative of every population. A study performed on one strain, cell line, age group, species or restricted group supports inference mainly for the experimental population represented by those samples.
• Ethical and practical restrictions- Some factors cannot be experimentally manipulated. Harmful exposures, certain diseases and many human characteristics cannot be assigned simply to create treatment and control groups. In clinical experiments, informed consent and protection of participants also restrict what treatments and procedures can be used.
• Controlled experiments may require considerable resources- Adequate controls, independent replicates, equipment, experimental material and repeated measurements increase the requirement of time, money and labor. Sample number is commonly limited by these practical and economic factors. Small sample sizes can reduce the ability of an experiment to detect a real but small effect.
• Measurement error and bias can remain- Instruments do not measure with unlimited precision. Background noise, sample handling and other technical effects can enter into the result even after careful control. Replication reduces random noise, but repeated measurements will not remove a systematic bias in the measurement itself.
• Hidden confounding can still occur- An unnoticed difference in reagent batch, incubation day, plate position, operator or equipment may become associated with one treatment. Then the effect of that factor and the experimental treatment become difficult to separate. Negative controls, randomization and balanced experimental arrangements are used to pick up or reduce such problems, but they do not identify every possible source of error.
• Complex biological interactions can be missed- A controlled experiment commonly isolates a limited number of variables so that their effects can be studied clearly. Biological responses, however, can depend on several interacting factors. When most of this variation is removed from an experiment, the observed effect may apply only to the particular combination of conditions that was tested.

Controlled Experiments vs Other Study Designs

An experiment is a broader term in which the researcher deliberately changes an exposure, treatment, or experimental condition and then measures the response. Experiments may be controlled or uncontrolled.

A controlled experiment includes an appropriate control or comparison strategy. It is used so that the effect produced by the experimental treatment can be separated from other possible changes.

Controlled vs Uncontrolled Experiments

Both controlled and uncontrolled experiments involve deliberate manipulation by the researcher. The major difference is in the comparison.

In a controlled experiment, the response obtained after treatment is compared with an adequate control condition. In an uncontrolled experiment, a treatment may be applied and the outcome is measured without such an internal comparison. Experimental studies can therefore be controlled (with comparison) or uncontrolled (without comparison).

For example, a bacterial culture is treated with an antimicrobial compound and the bacterial number decreases. If no untreated culture is present, this decrease cannot be properly separated from changes caused by incubation conditions, normal loss of viability, or another factor. The control culture provides a reference for comparison.

Controlled Experiment vs Observational Study

The main difference is based on who determines the exposure. In a controlled experiment, the treatment is assigned or manipulated by the investigator. In an observational study, naturally occurring differences are observed without assigning the exposure by the researcher.

For example, researchers may give different nutrient treatments experimentally to separate groups of plants. It is an experiment.

If the researchers instead measure naturally occurring nutrient levels in different soils and compare the plants already growing in those soils, it is an observational study. Observational studies can contain comparison groups and statistical adjustment for confounding, but the exposure groups may already differ in other characteristics which are related to the measured outcome.

Randomized Controlled Experiments

Randomization is another method used for experimental control. Experimental units such as organisms, cultures, plots, or human participants are assigned into treatment conditions by a chance mechanism instead of placing them into groups according to investigator choice or their already existing characteristics.

Random assignment tends to distribute known as well as unknown characteristics between different groups. It reduces systematic confounding between the treatment groups.

A controlled experiment does not always have to be randomized. Nonrandomized controlled experiments are also present.

A randomized controlled trial (RCT) is a particular type of controlled experiment, commonly used for testing health interventions. In this type, participants are randomly allocated into intervention and control groups. It is not another name used for every controlled experiment carried out in biology or laboratory science.

Controlled Experiment at a Glance

References

1. Carey, T. S., Sanders, G. D., Viswanathan, M., Trikalinos, T. A., Kato, E., & Chang, S. (2012). Framework for considering study designs for future research needs (Methods Future Research Needs Reports No. 8). Agency for Healthcare Research and Quality. https://www.ncbi.nlm.nih.gov/books/NBK95273/
2. Chidambaram, A. G., & Josephson, M. (2019). Clinical research study designs: The essentials. Pediatric Investigation, 3(4), 245–252. https://doi.org/10.1002/ped4.12166
3. Curtis, M. J., Bond, R. A., Spina, D., Ahluwalia, A., Alexander, S. P. H., Giembycz, M. A., Gilchrist, A., Hoyer, D., Insel, P. A., Izzo, A. A., Lawrence, A. J., MacEwan, D. J., Moon, L. D. F., Wonnacott, S., Weston, A. H., & McGrath, J. C. (2015). Experimental design and analysis and their reporting: New guidance for publication in BJP. British Journal of Pharmacology, 172(14), 3461–3471. https://doi.org/10.1111/bph.12856
4. Haas, J., Manro, J., Shannon, H., Anderson, W., Brozinick, J., Chakravartty, A., Chambers, M., Du, J., Eastwood, B., Heuer, J., Iturria, S., Iversen, P., Johnson, D., Johnson, K., O’Neill, M., Qian, H.-R., Sindelar, D., & Svensson, K. (2012). In vivo assay guidelines. In S. Markossian, A. Grossman, & H. Baskir (Eds.), Assay guidance manual. Eli Lilly & Company and National Center for Advancing Translational Sciences. https://www.ncbi.nlm.nih.gov/books/NBK92013/
5. Holmes, S., & Huber, W. (2019). Modern statistics for modern biology. Cambridge University Press. https://doi.org/10.1017/9781108551441
6. Hyndman, T. H., Bowden, R. S., Woodward, A. P., Pang, D. S. J., & Hampton, J. O. (2024). Uncontrolled pain: A call for better study design. Frontiers in Veterinary Science, 11, 1328098. https://doi.org/10.3389/fvets.2024.1328098
7. Landis, S. C., Amara, S. G., Asadullah, K., Austin, C. P., Blumenstein, R., Bradley, E. W., Crystal, R. G., Darnell, R. B., Ferrante, R. J., Fillit, H., Finkelstein, R., Fisher, M., Gendelman, H. E., Golub, R. M., Goudreau, J. L., Gross, R. A., Gubitz, A. K., Hesterlee, S. E., Howells, D. W., . . . Silberberg, S. D. (2012). A call for transparent reporting to optimize the predictive value of preclinical research. Nature, 490, 187–191. https://doi.org/10.1038/nature11556
8. Lipsitch, M., Tchetgen Tchetgen, E., & Cohen, T. (2010). Negative controls: A tool for detecting confounding and bias in observational studies. Epidemiology, 21(3), 383–388. https://doi.org/10.1097/EDE.0b013e3181d61eeb
9. National Academies of Sciences, Engineering, and Medicine, & Federal Judicial Center. (2025). Reference manual on scientific evidence (4th ed.). National Academies Press. https://doi.org/10.17226/26919
10. Olimpo, J. T. (2015). The Biology Experimental Design Challenge: An interactive approach to enhance students’ understanding of scientific inquiry in the context of an introductory biology course. Journal of Microbiology & Biology Education, 16(1), 75–76. https://doi.org/10.1128/jmbe.v16i1.817
11. Pen before pipette. (2017). Nature Methods, 14, 929. https://doi.org/10.1038/nmeth.4460
12. Richter, S. H., Garner, J. P., & Würbel, H. (2009). Environmental standardization: Cure or cause of poor reproducibility in animal experiments? Nature Methods, 6, 257–261. https://doi.org/10.1038/nmeth.1312
13. Singh, K., Gupta, P., & Shivashankar, R. (2021). Primer of epidemiology IV: Study designs II: Interventional or experimental designs. National Medical Journal of India, 34(4), 228–231. https://doi.org/10.25259/NMJI_373_19
14. Spielman, R. M., Jenkins, W. J., & Lovett, M. D. (2020). Psychology 2e. OpenStax. https://openstax.org/books/psychology-2e/pages/1-introduction
15. Sullivan, L. M., Weinberg, J., & Keaney, J. F., Jr. (2016). Common statistical pitfalls in basic science research. Journal of the American Heart Association, 5(10), e004142. https://doi.org/10.1161/JAHA.116.004142
16. Tanious, R., Manolov, R., Onghena, P., & Vlaeyen, J. W. S. (2024). Single-case experimental designs: The importance of randomization and replication. Nature Reviews Methods Primers, 4(1), Article 27. https://doi.org/10.1038/s43586-024-00312-8
17. Vollert, J., Macleod, M., Dirnagl, U., Kas, M. J., Michel, M. C., Potschka, H., Riedel, G., Wever, K. E., Würbel, H., Steckler, T., EQIPD Consortium, & Rice, A. S. C. (2022). The EQIPD framework for rigor in the design, conduct, analysis and documentation of animal experiments. Nature Methods, 19, 1334–1337. https://doi.org/10.1038/s41592-022-01615-y
18. Wagner, M. R., & Kleiner, M. (2025). How thoughtful experimental design can empower biologists in the omics era. Nature Communications, 16, 7263. https://doi.org/10.1038/s41467-025-62616-x
19. Wilson, E., Ramage, F. J., Wever, K. E., Sena, E. S., Macleod, M. R., & Currie, G. L. (2023). Designing, conducting, and reporting reproducible animal experiments. Journal of Endocrinology, 258(1), e220330. https://doi.org/10.1530/JOE-22-0330
20. Zhao, Y., Sampson, M. G., & Wen, X. (2020). Quantify and control reproducibility in high-throughput experiments. Nature Methods, 17, 1207–1213. https://doi.org/10.1038/s41592-020-00978-4
21. Zwahlen, M., & Salanti, G. (2018). Causal inference from experiment and observation. Evidence-Based Mental Health, 21(1), 34–38. https://doi.org/10.1136/eb-2017-102859