# How to Set Up Proper Controls in Biological Experiments ## Quick Answer - Select controls that match your experimental question and account for known confounders before you collect any data. - Document every control decision in your protocol and lab notebook so reviewers can trace how you ruled out alternative explanations. - No single control design fits all experiments, and even well-chosen controls cannot correct for poor measurement or biased sampling. ## Why Controls Determine Whether Your Biological Experiment Can Be Interpreted A biological experiment asks whether an observed effect comes from your treatment or from something else in the system. Controls are the measurements you take to distinguish those possibilities. Without them, you cannot know whether a change in your outcome variable is caused by the variable you manipulated or by an unintended factor such as temperature drift, reagent degradation, genetic background, or handling differences. The core principle is comparison. You compare results from a treated group against results from a group that did not receive the treatment but was otherwise handled identically. The control group establishes the baseline against which you judge the effect of your treatment. If the control group changes in the same way as the treated group, then your treatment did not produce a distinct effect. If the control group stays stable while the treated group changes, you have evidence that the treatment is responsible. This logic applies across biological fields. In cell culture, you compare treated wells against untreated wells from the same passage. In animal studies, you compare treated animals against littermates or matched animals housed under identical conditions. In molecular biology, you compare your experimental sample against a sample that went through every step except the one you are testing. In field ecology, you compare plots that received a manipulation against plots that did not. The cost of skipping controls is high. A result that looks significant may be an artifact of the assay, the handling procedure, or the environment. A result that looks null may be masked by a control that was not actually comparable. Reviewers and funding agencies expect to see controls described clearly enough that another laboratory could reproduce your conditions. The National Library of Medicine provides access to biomedical research methods references that describe how controls and comparison groups are used in rigorous study design. ## The Core Principle of Comparison in Experimental Design A biological experiment is a structured comparison. You define a question, you choose a system to study, you apply a treatment, and you measure an outcome. The control group is the part of the experiment that tells you what would have happened without the treatment. It is not an afterthought. It is the reference point against which every measurement is judged. The comparison must be fair. The treated and control groups must differ only in the variable you are testing. If they differ in any other way, you cannot attribute the difference in outcome to your treatment. This is the logic of the controlled comparison, and it is the same logic used in clinical trials, field studies, and laboratory experiments. In practice, a fair comparison requires that you hold constant every factor that could influence the outcome. These factors include the starting material, the environment, the timing, the equipment, and the person performing the measurements. If you change two variables at once, you cannot know which one caused the effect. This is why you design the experiment so that the only difference between groups is the treatment. ### The Role of Randomization in Making Comparisons Fair Randomization is the process of assigning experimental units to treatment or control groups by chance instead of by judgment. It reduces the risk that your groups differ in some systematic way before the experiment begins. If you assign all the larger animals to the treatment group and all the smaller animals to the control group, you cannot tell whether the outcome is due to the treatment or to the size difference. Randomization does not guarantee that the groups are identical. It makes it likely that differences are distributed by chance instead of by a systematic bias. The larger the number of units, the more effective randomization is at balancing unknown factors. In small experiments, randomization is still useful, but you should also record the characteristics of each group so you can check whether the groups were comparable at the start. ### Why Blinding Matters in Biological Experiments Blinding means that the person measuring the outcome does not know which units received the treatment and which received the control. This prevents the observer from unconsciously influencing the measurement. In animal studies, blinding is important because the person who handles the animals may treat them differently if they know which group they are in. In microscopy, blinding is important because the person counting cells may be more likely to find the expected result in the treated sample. Blinding is not always possible. In some experiments, the treatment is obvious, such as a surgical procedure or a visible color change. In those cases, you can still blind the outcome assessment by having a different person measure the result or by coding the samples so the assessor does not know the group identity. The goal is to reduce the risk of bias in the measurement step. ## Types of Controls and When to Use Each One Different experimental questions require different types of controls. The table below summarizes the main types and the situations where each one is appropriate. | Control Type | What It Does | When To Use It | | --- | --- | --- | | Negative control | Shows the outcome when no treatment is applied | Use when you need to establish the baseline level of the outcome | | Positive control | Shows the outcome when a known effective treatment is applied | Use when you need to confirm that your assay can detect an effect | | Vehicle control | Shows the outcome when the carrier or solvent is applied without the active agent | Use when your treatment is dissolved in a solvent or carrier | | Sham control | Shows the outcome when the procedure is performed without the active step | Use when your treatment involves a surgical or physical procedure | | Time zero control | Shows the outcome at the start of the experiment | Use when you need to measure change over time | | Reference control | Shows the outcome for a known standard or established sample | Use when you need to calibrate your measurement against a known value | ### Negative Controls A negative control is a sample or group that is not exposed to the treatment. It establishes the background level of the outcome. If your negative control shows a high level of the outcome, your assay may be detecting something other than the treatment effect. If your negative control shows a low level, you have a baseline against which to compare your treated group. In a cell culture experiment, the negative control is a well that receives the same culture medium and handling but no treatment. In an animal study, the negative control is an animal that receives the same housing and handling but no treatment. In a molecular assay, the negative control is a sample that goes through every step of the protocol but contains no target material. ### Positive Controls A positive control is a group that is known to produce the outcome. It confirms that your assay is working. If your positive control does not show the expected result, then your assay may be failing, and any negative result from your treated group cannot be interpreted. In a gene expression experiment, the positive control is a sample that is known to express the gene you are measuring. In a drug response experiment, the positive control is a compound that is known to produce the effect you are studying. In a microbial growth experiment, the positive control is a strain that is known to grow under your conditions. ### Vehicle Controls A vehicle control is a group that receives the solvent or carrier used to deliver the treatment, but not the active agent. This is necessary when your treatment is dissolved in a solvent such as water, ethanol, or dimethyl sulfoxide. The solvent itself can affect the outcome, so you need to compare your treated group against a group that received the same solvent without the active agent. The vehicle control must match the treated group in every way except the active agent. The volume, the timing, and the route of administration should be the same. If you inject a drug dissolved in saline, the vehicle control receives the same volume of saline by the same route. ### Sham Controls A sham control is used in experiments that involve a procedure. The sham group undergoes the same procedure as the treated group, but the active step is omitted. In a surgical experiment, the sham group receives the same incision and closure but not the actual intervention. This controls for the effect of the procedure itself, such as the stress of surgery, the anesthesia, and the wound healing. The sham control is important because the procedure can affect the outcome independently of the treatment. If you compare a surgically treated group against an untreated group, you cannot tell whether the effect is due to the treatment or to the surgery. The sham group gives you the baseline for the procedure. ### Time Zero Controls A time zero control is a measurement taken at the start of the experiment. It establishes the starting point for each group. This is important in experiments where the outcome changes over time. If you measure the outcome only at the end, you cannot tell whether the change was due to the treatment or to the passage of time. The time zero control is often the same as the baseline measurement. You record the outcome for each unit before the treatment begins, then you apply the treatment, and then you measure the outcome again at the end. The change from time zero to the end is the effect you interpret. ### Reference Controls A reference control is a sample or group with a known value. It is used to calibrate your measurement. In a quantitative assay, the reference control is a standard with a known concentration. In a behavioral experiment, the reference control is a group with a known response. The reference control confirms that your measurement system is producing values that are comparable to known standards. ## How to Choose the Right Controls for Your Experiment The choice of controls depends on your experimental question and the potential confounders in your system. The decision tree below shows the main steps in selecting controls. ```mermaid flowchart TD A[Define the experimental question] --> B[Identify the outcome you will measure] B --> C[Identify the factors that could affect the outcome] C --> D[Is there a known effective treatment?] D -->|Yes| E[Include a positive control] D -->|No| F[Use a negative control only] E --> G[Is the treatment delivered in a solvent?] F --> G G -->|Yes| H[Include a vehicle control] G -->|No| I[Is the treatment a procedure?] H --> I I -->|Yes| J[Include a sham control] I -->|No| K[Include a time zero control] J --> K K --> L[Document all control decisions] ``` ### Step 1: Define the Experimental Question The first step is to write down the exact question you are asking. The question should specify the treatment, the outcome, and the population. For example, "Does compound X reduce the growth of bacterial strain Y in culture?" This question tells you that the treatment is compound X, the outcome is growth, and the population is bacterial strain Y. The question also tells you what you are not asking. If you are not asking about the effect of the solvent, you still need a vehicle control to rule out the solvent as a confounder. If you are not asking about the effect of the procedure, you still need a sham control to rule out the procedure as a confounder. ### Step 2: Identify the Confounders A confounder is a factor that is associated with both the treatment and the outcome. It can make it look like the treatment has an effect when it does not. Common confounders in biological experiments include the solvent, the procedure, the handling, the time of day, the temperature, and the genetic background of the sample. To identify the confounders, ask yourself what else changes when you apply the treatment. If you dissolve the treatment in a solvent, the solvent is a confounder. If you inject the treatment, the injection is a confounder. If you culture the cells in a new medium, the medium is a confounder. Each confounder needs a control. ### Step 3: Select the Minimum Set of Controls The minimum set of controls is the smallest number of groups that allows you to interpret your result. You need a negative control to establish the baseline. You need a positive control if you want to confirm that your assay can detect an effect. You need a vehicle control if the treatment is delivered in a solvent. You need a sham control if the treatment is a procedure. The minimum set is not always the best set. If you have the resources, you can include additional controls to strengthen your interpretation. But you should not include controls that are not relevant to your question. A control that does not address a confounder in your design is a waste of resources. ### Step 4: Document the Decision The decision to include or exclude a control must be documented. The documentation should state which controls you used, why you used them, and what you expect them to show. This documentation is part of your experimental record. It allows you to explain your design to reviewers and to reproduce the experiment later. The documentation should also state the limitations of your controls. If you did not include a positive control because you did not have a known effective treatment, you should state that. If you did not include a vehicle control because the treatment was not delivered in a solvent, you should state that. The documentation makes your reasoning transparent. ## Practical Implementation of Controls in the Laboratory The implementation of controls requires attention to the details of your protocol. The controls must be treated exactly the same as the treated group, except for the treatment itself. Any difference in handling between the control and the treated group is a source of bias. ### Preparing the Control Groups The control groups must be prepared from the same stock, the same culture, or the same batch of animals as the treated group. If the control group comes from a different source, you cannot tell whether the difference in outcome is due to the treatment or to the source. The control group must be handled at the same time, in the same room, and with the same equipment. The number of units in the control group should be comparable to the number in the treated group. If the control group is much smaller, the comparison is less reliable. The sample size in each group should be large enough to detect the effect you expect to see. ### Running the Controls in the Same Assay The controls must be run in the same assay as the treated group. If you run the controls in a separate assay, you cannot compare the results because the conditions may differ. The controls and the treated group must be processed at the same time, with the same reagents, and on the same plate or in the same batch. In a plate-based assay, the controls should be on the same plate as the treated group. In an animal study, the controls should be housed in the same room and handled by the same person. In a molecular assay, the controls should be processed in the same run. ### Recording the Control Results The control results must be recorded in the same way as the treated group. The measurements should be taken with the same instrument, by the same person, and at the same time. The data should be recorded in the same format, so that the comparison is straightforward. The control results should be reported in your results section. You should state the mean and the variability of the control group, and you should compare the treated group against the control group. The control results are part of the evidence, and they must be available to the reader. ## Records and Measurements for Control Verification The records of your controls are the evidence that your experiment was valid. The records should include the protocol, the data, and the analysis. The protocol should describe the controls and the reason for each one. The data should show the measurements for each group. The analysis should show the comparison between the groups. ### What to Record Record the following for each control group: - The type of control (negative, positive, vehicle, sham, time zero, or reference) - The number of units in the group - The source of the units - The conditions under which the group was handled - The measurements for each unit - The mean and variability of the group Record the same information for the treated group. The records should be complete enough that another researcher can reproduce your experiment from the records alone. ### How to Check the Controls Check the controls before you interpret the treated group. The negative control should show the expected baseline. The positive control should show the expected effect. The vehicle control should show the same baseline as the negative control. The sham control should show the same baseline as the negative control. If any control does not show the expected result, you cannot interpret the treated group. The assay may be failing, or the conditions may be wrong. You need to identify the problem and repeat the experiment. ### Common Failure Patterns The following are common failure patterns in control design: - The negative control shows a high level of the outcome, which means the assay is detecting something other than the treatment effect. - The positive control does not show the expected effect, which means the assay cannot detect the effect. - The vehicle control differs from the negative control, which means the solvent has an effect. - The sham control differs from the negative control, which means the procedure has an effect. - The control group is not comparable to the treated group because of a difference in handling or source. ## Welfare and Safety Context for Controls The design of controls must respect the welfare of the animals and the safety of the laboratory personnel. The controls should not be subjected to unnecessary procedures. The sham control should receive the same anesthesia and pain management as the treated group. The negative control should not be deprived of food, water, or housing. The safety of the laboratory personnel is also part of the control design. The controls should be handled with the same safety precautions as the treated group. If the treatment is hazardous, the controls should be handled in the same containment. The controls should not be used to test a hazardous substance without the same safety measures. The use of animals in research is regulated by institutional policies and national laws. The design of the control group must be approved by the institutional animal care and use committee. The committee reviews the protocol to ensure that the number of animals is the minimum needed to answer the question and that the controls are designed to minimize suffering. ## Limitations of Controls Controls are necessary, but they have limitations. A control cannot eliminate all sources of bias. The control group is not identical to the treated group. It is a group that is handled the same way, but it is still a different group. The difference between the groups may be due to a factor that you did not control. The control group cannot account for the effect of the treatment on the measurement itself. If the treatment changes the way the outcome is measured, the control group cannot correct for that. For example, if the treatment changes the color of the sample and the color affects the measurement, the control group cannot show that. The control group cannot account for the effect of the treatment on the sample over time. If the treatment changes the sample in a way that affects the measurement, the control group cannot show that. The control group is a snapshot of the baseline, but the treated group is a snapshot of the treated state. ## When to Escalate to a Professional If you are designing an experiment and you are unsure which controls are necessary, you should consult a professional. The professional can be a senior researcher, a statistician, or a laboratory manager. The professional can review your experimental question and your design and recommend the appropriate controls. You should also escalate if your controls are not showing the expected results. If the negative control is high or the positive control is low, you should stop and consult a professional before you interpret the treated group. The problem may be in the assay, the reagents, or the protocol, and the professional can help you identify the cause. You should escalate if you are working with a hazardous substance or an animal model and you are unsure about the safety or welfare requirements. The institutional safety officer or the animal care committee can provide the guidance you need. ## A Practical Decision Framework for Matching Controls to Your Experimental Question The decision tree in the previous section tells you which controls to consider, but it does not tell you how to weigh the costs and benefits of each control in your specific situation. A practical framework helps you move from the abstract question of "which controls do I need" to a concrete list of groups you will actually run. This section provides a structured method for making those decisions, recording them, and troubleshooting when your controls fail. ### The Control Justification Matrix Before you prepare a single sample, build a control justification matrix. This is a simple table with one row for each potential control and four columns: the confounder it addresses, the expected result if the control works, the cost of including it, and the consequence of excluding it. The matrix forces you to state explicitly why each control exists and what it protects against. | Control | Confounder Addressed | Expected Result | Cost of Inclusion | Consequence of Exclusion | | --- | --- | --- | --- | --- | | Negative control | Baseline outcome level | Low or zero signal | One extra sample group | Cannot interpret treated group | | Positive control | Assay sensitivity | Known positive signal | One extra sample group | Cannot confirm assay detects effect | | Vehicle control | Solvent or carrier effect | Same as negative control | One extra sample group | Cannot separate solvent from treatment | | Sham control | Procedure effect | Same as negative control | One extra surgical group | Cannot separate procedure from treatment | | Time zero control | Baseline drift over time | Starting value for each group | One measurement per unit | Cannot measure change from baseline | | Reference control | Measurement calibration | Known standard value | One standard sample | Cannot confirm measurement accuracy | The matrix is not a checklist. It is a reasoning tool. For each row, you must be able to state the confounder in one sentence and the expected result in one sentence. If you cannot state the expected result, you do not understand what the control is supposed to show, and you should not run it. ### Scoring Control Necessity To decide whether a control is necessary, score it on three criteria. The first criterion is the likelihood that the confounder is present in your system. The second is the magnitude of the effect the confounder could have on your outcome. The third is the feasibility of running the control with your current resources. For the likelihood score, ask whether the confounder is a known feature of your system. If you are dissolving a drug in dimethyl sulfoxide, the solvent effect is likely present. If you are using a surgical procedure, the procedure effect is likely present. If you are measuring a gene that is not expressed in your cell line, the positive control may be difficult to find. For the magnitude score, ask whether the confounder could produce an effect large enough to change your conclusion. A solvent that changes cell viability by 2 percent may not matter if your treatment effect is 50 percent. A solvent that changes viability by 30 percent will obscure a treatment effect of 20 percent. For the feasibility score, ask whether you can run the control with the same cells, animals, or samples as the treated group. If you cannot, the control may introduce a new confounder. A control that requires a different batch of cells is less useful than one that uses the same batch. Add the three scores. A control with a high total score is necessary. A control with a low total score can be excluded, but you must document the exclusion and the reason. The scoring is not a substitute for judgment. It is a way to make your judgment explicit and reviewable. ### A Worked Example of the Matrix in Practice Consider a study of a new compound that you dissolve in ethanol and apply to cultured neurons. You plan to measure cell survival after 24 hours. The confounders are the ethanol solvent, the handling of the culture plates, and the assay itself. The negative control is a well with culture medium and no ethanol. The vehicle control is a well with the same concentration of ethanol but no compound. The positive control is a well with a known toxic compound that reduces survival. The time zero control is a measurement of survival at the start of the experiment. The matrix for this experiment has four rows. The negative control addresses the baseline survival of the cells. The vehicle control addresses the ethanol effect. The positive control addresses the sensitivity of the survival assay. The time zero control addresses the starting survival of the cells. The vehicle control is necessary because ethanol is a known solvent that can affect cell survival. The positive control is necessary because you need to confirm that the survival assay can detect a reduction. The time zero control is necessary because cell survival changes over 24 hours even without treatment. The sham control is not relevant because there is no procedure. The reference control is not relevant because you are not calibrating against a known standard. The matrix shows that four controls are necessary and two are not. ### The Control Verification Record Once you have decided which controls to run, create a control verification record. This is a separate section of your lab notebook or electronic record that tracks the performance of each control. The record should have one entry per control per experiment. Each entry should state the control type, the expected result, the observed result, and whether the control passed or failed. The verification record is not the same as your raw data. It is a summary that allows you to check the controls before you interpret the treated group. The record should be updated on the same day you collect the data. If you wait until the end of the experiment, you may not remember the conditions under which the control was run. The record should also include the batch number of the reagents, the passage number of the cells, and the date and time of the experiment. These details allow you to trace a control failure to a specific reagent or condition. The National Library of Medicine provides access to research methods references that describe how detailed experimental records support the interpretation of study results. ### The Control Acceptance Criteria Before you run the experiment, write the acceptance criteria for each control. The acceptance criteria are the specific values or ranges that the control must meet for the experiment to be interpretable. These criteria are not the same as the expected result. The expected result is what you predict. The acceptance criteria are the boundaries within which the control is considered valid. For a negative control, the acceptance criterion might be that the survival is above 90 percent. For a positive control, the acceptance criterion might be that the survival is below 50 percent. For a vehicle control, the acceptance criterion might be that the survival is within 5 percent of the negative control. For a time zero control, the acceptance criterion might be that the starting survival is between 85 and 95 percent. The acceptance criteria must be written before you collect the data. If you write them after you see the results, you can adjust them to fit the data. This is a form of bias. The criteria should be based on your knowledge of the system and the assay, not on the data you collect. ### The Control Failure Response Protocol When a control fails, you need a response protocol. The protocol is a set of steps you follow to identify the cause of the failure and decide whether to repeat the experiment. The protocol should be written before the experiment starts, so you do not have to make decisions under pressure. The first step is to confirm the failure. Check the raw data and the verification record. Make sure the failure is not a recording error. If the control value is outside the acceptance criteria, confirm that the measurement was taken correctly and that the sample was labeled correctly. The second step is to identify the likely cause. The cause is usually one of the following: the reagent is degraded, the cells are contaminated, the instrument is miscalibrated, or the protocol was not followed. Check the reagent batch, the cell morphology, the instrument calibration, and the protocol log. The third step is to decide whether to repeat the experiment. If the cause is a degraded reagent, replace the reagent and repeat. If the cause is contaminated cells, start a new culture and repeat. If the cause is a miscalibrated instrument, calibrate the instrument and repeat. If the cause is a protocol error, correct the protocol and repeat. The fourth step is to document the failure and the response. The documentation should include the control type, the expected result, the observed result, the cause, and the action taken. This documentation is part of your experimental record and should be available to reviewers. ### The Control Failure Log The control failure log is a running list of every control failure you encounter. The log should include the date, the experiment, the control type, the expected result, the observed result, the cause, and the response. The log is useful for identifying patterns. If you see the same control failing repeatedly, you have a systematic problem in your protocol or your reagents. The failure log is also useful for planning future experiments. If you know that a particular reagent batch is causing control failures, you can order a new batch before you start the next experiment. If you know that a particular step in the protocol is prone to error, you can add a check at that step. The failure log is not a record of your mistakes. It is a record of the problems you identified and the solutions you applied. The log shows that you are monitoring the quality of your experiment and that you are responding to problems when they appear. ### The Control Audit A control audit is a review of your control design and your control records. The audit is performed after the experiment is complete and before you interpret the results. The audit checks that the controls were run correctly, that the acceptance criteria were met, and that the control results are consistent with the interpretation. The audit should be performed by a person who was not involved in the experiment. This person can be a colleague, a supervisor, or a statistician. The auditor reviews the control matrix, the control verification record, the failure log, and the raw data. The auditor checks that the controls are appropriate for the question and that the records are complete. The audit is not a formality. It is a check that the experiment can be interpreted. If the audit finds that a control was not run or that a control failed, the experiment cannot be interpreted. The audit is the final step before you draw conclusions from your data. ### When to Escalate a Control Problem You should escalate a control problem when you cannot identify the cause of a failure or when the failure persists after you have corrected the obvious causes. If you have repeated the experiment twice and the control still fails, you should consult a senior researcher or a statistician. The problem may be in the design of the control itself, not in the execution. You should also escalate when you are unsure whether a control is necessary. If you cannot state the confounder that the control addresses, or if you cannot state the expected result, you should consult a professional before you run the experiment. The professional can review your control matrix and help you decide which controls are necessary. You should escalate when the control failure suggests a safety or welfare issue. If the control failure is caused by a hazardous reagent or an animal welfare problem, you should stop the experiment and consult the institutional safety officer or the animal care committee. The control failure is a signal that the system is not working, and the system must be corrected before you continue. ### The Control Review Meeting A control review meeting is a scheduled discussion of the control design and the control results. The meeting is held after the experiment is complete and before the results are interpreted. The meeting includes the researcher, the senior researcher, and the statistician. The meeting reviews the control matrix, the verification record, the failure log, and the audit report. The purpose of the meeting is to confirm that the controls are adequate and that the results can be interpreted. The meeting is not a presentation of the results. It is a review of the control system. The meeting should be documented in the minutes, and the minutes should be stored with the experimental records. The control review meeting is a practical step that many laboratories skip. The result is that control problems are discovered after the results are interpreted, and the experiment has to be repeated. A review meeting before interpretation is a low-cost way to catch control problems early. ### The Control System as a Whole The control matrix, the verification record, the acceptance criteria, the failure response protocol, the failure log, the audit, and the review meeting form a single system. The system is designed to answer one question: can the results of this experiment be interpreted? The system is not a set of separate steps. It is a continuous process that starts when you design the experiment and ends when you interpret the results. The system is also a record. It shows that you made your control decisions deliberately, that you monitored the controls during the experiment, and that you responded to failures. This record is part of the evidence that your experiment is valid. The record is also part of the evidence that you can present to reviewers and funding agencies. The system is not a guarantee. A control system cannot eliminate all sources of bias. It can make the bias visible and it can provide a record of the decisions you made. The system is a practical way to ensure that the controls you choose are the controls you need and that the controls you run are the controls you can interpret. ## Frequently Asked Questions ### What is the difference between a negative control and a positive control? A negative control is a group that receives no treatment and establishes the baseline level of the outcome. A positive control is a group that receives a known effective treatment and confirms that the assay can detect an effect. You need both to interpret your results. ### Do I always need a positive control? You need a positive control when you want to confirm that your assay can detect an effect. If you do not have a known effective treatment, you cannot include a positive control. In that case, you should state that limitation in your records. ### What is a vehicle control and when do I need one? A vehicle control is a group that receives the solvent or carrier used to deliver the treatment, but not the active agent. You need a vehicle control when your treatment is dissolved in a solvent, because the solvent itself can affect the outcome. ### What is a sham control and when do I need one? A sham control is a group that undergoes the same procedure as the treated group but without the active treatment. You need a sham control when your treatment involves a surgical or physical procedure, because the procedure itself can affect the outcome. ### How many controls do I need? The number of controls depends on your experimental question and the confounders in your design. You need at least a negative control to establish the baseline. You need additional controls for each confounder, such as a vehicle control for the solvent or a sham control for the procedure. ### Can I use the same group as both a negative control and a vehicle control? You can use the same group if the vehicle is the same as the baseline. If the vehicle is the same as the baseline, the vehicle control is the same as the negative control. If the vehicle is different from the baseline, you need separate controls. ### What should I do if my control does not show the expected result? If your control does not show the expected result, you should stop and identify the problem. The assay may be failing, the reagents may be contaminated, or the conditions may be wrong. You should not interpret the treated group until you have resolved the problem. ### How do I document my controls? You should document the type of control, the number of units, the source, the handling, the measurements, and the analysis. The documentation should be complete enough that another researcher can reproduce your experiment from the records alone. ## Related Diagnostic Guides - [Animal Biology](/blog/guides/animal-biology) - [Animal Research](/blog/guides/animal-research) - [Biology Facts](/blog/guides/biology-facts) ## References and Further Reading - [Research Methods Resources](https://www.ncbi.nlm.nih.gov/books). National Library of Medicine. > This article is educational and does not replace validated laboratory procedures, institutional biosafety review, manufacturer instructions, or professional interpretation.