# Fractional Factorial Design for Cell Culture Screening ## Quick Answer - A fractional factorial design lets you screen many cell culture factors with fewer experimental runs by testing a carefully chosen subset of all possible combinations. - Select a resolution that separates main effects from two-factor interactions, then use the alias structure to identify which effects are confounded before you run the experiment. - The main limitation is that fractional designs cannot distinguish between confounded effects, so you must plan follow-up experiments to resolve any significant aliased terms. ## Understanding Fractional Factorial Designs in Cell Culture Cell culture experiments routinely require evaluation of multiple factors simultaneously. Media components, serum concentration, pH, temperature, oxygen tension, seeding density, and incubation time all influence cell growth and productivity. A full factorial design that tests every combination of these factors becomes impractical when the factor count grows. A design with seven factors at two levels requires 128 runs. A fractional factorial design reduces this workload by selecting a carefully chosen fraction of the full factorial, typically one half, one quarter, or one eighth of the total runs. The statistical foundation of fractional factorial designs rests on the concept of aliasing. When you run a fraction of the full factorial, some effects become mathematically indistinguishable from others. These effects are said to be aliased or confounded. The pattern of aliasing determines what you can learn from the experiment. A well-chosen fraction ensures that main effects are aliased only with higher-order interactions, which are often assumed to be negligible in screening experiments. ### Why Screening Matters in Cell Culture Cell culture systems are sensitive to many environmental and nutritional inputs. Researchers often begin with a large list of candidate factors and need to identify which ones meaningfully affect the outcome. Screening designs serve this purpose by estimating the direction and approximate magnitude of each factor's effect with a modest number of runs. The goal is not to build a precise predictive model but to rank factors by importance and decide which ones deserve further investigation. A fractional factorial design is particularly useful when the number of factors is large relative to the number of runs you can afford. Cell culture experiments are expensive in materials, time, and labor. Each run requires media, supplements, culture vessels, and incubator space. Reducing the number of runs from 128 to 16 or 32 can save substantial resources while still providing useful information about the factors under study. ### The Role of Design Resolution Resolution is a property of a fractional factorial design that describes the pattern of aliasing. Resolution III designs alias main effects with two-factor interactions. Resolution IV designs alias two-factor interactions with each other but keep main effects clear of two-factor interactions. Resolution V designs keep main effects and two-factor interactions clear of each other, but alias two-factor interactions with three-factor interactions. For cell culture screening, resolution IV designs are often a practical choice. They allow you to estimate all main effects without contamination from two-factor interactions, which is the primary goal of a screening experiment. Resolution III designs are more economical but risk having main effects confounded with interactions, which can lead to misleading conclusions. Resolution V designs provide cleaner estimates but require more runs. ### The Alias Structure The alias structure is the complete list of which effects are confounded with which other effects. You must examine this list before running the experiment to understand what your results can and cannot tell you. For example, in a 2^(7-4) design at resolution III, the main effect of factor A might be aliased with the interaction BC. If you observe a significant effect for A, you cannot know whether the true cause is A, the BC interaction, or both. In a 2^(7-3) design at resolution IV, main effects are not aliased with two-factor interactions, but two-factor interactions are aliased with each other. If you find a significant interaction term, you know that at least one of the aliased interactions is active, but you cannot identify which one without additional experiments. ### Choosing the Number of Factors and Runs The number of factors you can screen depends on the number of runs you are willing to perform. A 2^(k-p) design has 2^(k-p) runs, where k is the number of factors and p is the number of generators. For example, a 2^(7-3) design has 16 runs and can screen seven factors. A 2^(7-4) design has 8 runs and screens seven factors but at resolution III, which is generally too low for reliable screening. The choice of generators determines the resolution and the alias structure. Standard design tables list the generators for each design. You should use these tables instead of attempting to construct a design from scratch, because the tables have been verified to produce the best possible resolution for a given number of runs. ## At a Glance | Design | Number of Factors | Number of Runs | Resolution | Main Effects Clear of Two-Factor Interactions | Typical Use | |--------|-------------------|----------------|------------|-----------------------------------------------|-------------| | 2^(4-1) | 4 | 8 | IV | Yes | Screening four factors with moderate run budget | | 2^(7-3) | 7 | 16 | IV | Yes | Screening seven factors with a practical run budget | | 2^(7-4) | 7 | 8 | III | No | Rapid screening when run budget is very limited | ## Selecting Factors and Levels ### Factor Selection Criteria Choose factors based on biological plausibility and prior knowledge. Each factor must have a defensible reason for inclusion. In cell culture, common factors include serum concentration, growth factor supplementation, oxygen tension, pH buffer concentration, seeding density, and culture duration. The choice of factors should reflect the specific questions you are asking about your cell system. The number of factors you can include depends on the design you choose. If you have more candidate factors than the design can accommodate, you must prioritize. Factors with strong prior evidence of an effect should be included. Factors with weak or uncertain evidence can be left for a later experiment. The goal of screening is to identify the most important factors, not to test every possible variable. ### Choosing High and Low Levels Each factor must be assigned a high level and a low level. These levels should be chosen to represent the range of conditions that are biologically meaningful and practically feasible. The levels should be far enough apart to produce a detectable effect if one exists, but not so extreme that they cause cell death or other artifacts. For example, if you are screening serum concentration, the low level might be 1% and the high level might be 10%. If you are screening pH, the low level might be 7.0 and the high level might be 7.6. The levels should be based on the known operating range of your cell line. If you are uncertain about the appropriate range, you can consult published protocols for similar cell types. ### Center Points and Replication Fractional factorial designs at two levels assume a linear relationship between each factor and the response. This assumption may not hold in cell culture, where responses often show curvature. Adding center points, which are runs at the midpoint of all factor levels, allows you to test for curvature. Center points also provide an estimate of pure experimental error, which is useful for assessing the significance of effects. Replication of some runs is also valuable. Replication provides an estimate of run-to-run variability, which is essential for determining whether observed effects are larger than the noise in the system. In cell culture, variability can arise from differences in cell passage number, media preparation, and environmental conditions. Replication helps you distinguish real effects from this background noise. ## Building the Design Matrix ### Using Design Tables The design matrix is a table that specifies the level of each factor for each run. You can generate this matrix using statistical software or by following a standard design table. The design table lists the runs in a coded form, where -1 represents the low level and +1 represents the high level. You must convert these coded values to the actual experimental conditions before running the experiment. ### Randomization The order in which you run the experimental runs should be randomized. Randomization protects against systematic biases that can arise from changes over time, such as drift in incubator temperature, changes in media batches, or differences in cell passage number. Without randomization, an effect that is actually due to time could be incorrectly attributed to a factor. In cell culture, randomization can be challenging because some factors are difficult to change. For example, if you are testing different media formulations, you may need to prepare all media at the start of the experiment. In this case, you should randomize the order in which the media are used, even if the media are prepared in advance. ### Blocking If the experiment must be run in multiple batches, you should consider blocking. A block is a group of runs that are performed under similar conditions. Blocking allows you to account for systematic differences between batches, such as different incubators, different days, or different operators. The design can be arranged so that the block effect is not confounded with the factor effects. In cell culture, blocking is often necessary because you cannot run all experimental conditions simultaneously. For example, if you have 16 runs and can only handle 8 plates at a time, you would run the experiment in two blocks of 8. The design should be chosen so that the block effect is orthogonal to the main effects, meaning that the block effect does not bias the estimates of the factor effects. ## Running the Experiment ### Preparing the Cell Culture The quality of the cell culture is the foundation of the experiment. Cells should be in the logarithmic growth phase and at a consistent passage number. The passage number should be the same for all runs to avoid confounding the effects with passage-related changes. The cells should be counted and seeded at the same density for each run, unless seeding density is one of the factors being tested. ### Standardizing the Protocol The protocol should be standardized as much as possible. This includes the type of culture vessel, the volume of media, the incubation temperature, the CO2 concentration, and the duration of the culture. Any variation in these conditions can introduce noise into the experiment and reduce the ability to detect real effects. ### Recording the Data The outcome variable should be measured consistently across all runs. In cell culture, common outcomes include cell viability, cell count, metabolic activity, and protein production. The measurement method should be validated and consistent. For example, if you are using a colorimetric assay, the same plate reader and the same wavelength should be used for all runs. ## Analyzing the Results ### Estimating Effects The effect of each factor is estimated by comparing the average response at the high level with the average response at the low level. In a fractional factorial design, this estimate is a linear combination of the responses. The effect estimate is the difference between the average of the runs where the factor is at the high level and the average of the runs where the factor is at the low level. ### Interpreting the Alias Structure The interpretation of the results must be guided by the alias structure. If a main effect is significant, you can attribute it to that factor only if the main effect is not aliased with any other effect. In a resolution IV design, main effects are clear of two-factor interactions, so a significant main effect can be attributed to the factor itself. If a two-factor interaction is significant, you must consult the alias list to see which other interactions are confounded with it. You cannot conclude that the specific interaction is the cause of the effect. You can only conclude that at least one of the aliased interactions is active. ### Using a Pareto Chart A Pareto chart is a bar chart that displays the absolute values of the factor effects in descending order. This chart helps you identify which factors have the largest effects. The chart also shows a reference line that indicates the threshold for statistical significance. Factors with effects above the line are considered significant. ### Checking for Curvature If you included center points in the design, you can test for curvature by comparing the average response at the center points with the average response at the factorial points. A significant difference indicates that the relationship between the factors and the response is not linear. This finding suggests that the optimal conditions may be inside the range of the factor levels, and a response surface design may be needed for further optimization. ## Common Failure Patterns ### Confusing Confounded Effects A common error is to interpret a significant effect without consulting the alias list. For example, if a main effect is aliased with a two-factor interaction, a significant result could be due to the interaction instead of the main effect. This error leads to incorrect conclusions and wasted follow-up experiments. ### Choosing Levels Too Close Together If the high and low levels of a factor are too close, the effect may be too small to detect. The experiment will show no significant effect for that factor, even though the factor is important. The levels should be chosen to span a range that is likely to produce a detectable effect. ### Ignoring Blocking If the experiment is run over multiple days or with multiple operators, the block effect can be large. If blocking is not accounted for, the block effect can inflate the error and reduce the ability to detect real effects. The design should include blocks if the experiment cannot be run in a single batch. ### Failing to Randomize If the runs are not randomized, the results can be biased by time-related trends. For example, if all the high-level runs are performed first and the low-level runs are performed later, the effect of the factor will be confounded with the effect of time. Randomization is essential for valid inference. ### Overlooking the Need for Replication Without replication, you have no estimate of experimental error. You cannot determine whether an observed effect is larger than the noise in the experiment. Replication is essential for a valid analysis. ## Records and Measurements ### Data Recording The data should be recorded in a structured format that includes the run number, the factor levels, the response, and any notes about the run. The data should be recorded in a laboratory notebook or an electronic data system. The record should be complete enough that another researcher can reproduce the experiment. ### Documentation of the Design The design matrix, the alias list, and the randomization order should be documented before the experiment is run. This documentation is essential for the analysis and for the interpretation of the results. The documentation should be stored with the data. ### Measurement of the Response The response should be measured with a validated method. The method should be described in the records, including the instrument, the settings, and the calibration. The measurement should be performed in a consistent manner across all runs. ## Quality and Welfare Controls ### Cell Health Monitoring The health of the cells should be monitored throughout the experiment. The viability should be checked before and after the experiment. The morphology of the cells should be observed under a microscope. If the cells show signs of stress, such as detachment or vacuolization, the results may be affected. ### Contamination Control The experiment should be performed under sterile conditions. The media and the supplements should be checked for contamination. The incubator should be cleaned and maintained. If contamination is detected, the affected runs should be repeated. ### Biosafety The cell culture should be performed in a biosafety cabinet. The waste should be disposed of according to the institutional guidelines. The cells should be handled with appropriate personal protective equipment. The biosafety level of the cell line should be determined before the experiment. ## Limitations of Fractional Factorial Designs ### Confounding The main limitation of a fractional factorial design is the confounding of effects. The design cannot distinguish between effects that are aliased. This limitation is inherent in the design and cannot be overcome by the analysis. The researcher must accept this limitation and plan follow-up experiments to resolve the aliased effects. ### Assumption of Linearity The design assumes a linear relationship between the factors and the response. This assumption may not hold in cell culture. If the relationship is nonlinear, the design may miss important effects. The center points can detect curvature, but they cannot model the curvature. ### The Need for Follow-Up Experiments The results of a fractional factorial design are not the final answer. The design identifies the factors that are likely to be important, but the exact nature of the effects must be confirmed in follow-up experiments. The follow-up experiments can be designed to resolve the aliasing and to model the curvature. ## Professional Escalation Criteria If the results of the experiment are ambiguous or if the design has a low resolution, you should consider a follow-up experiment. The follow-up experiment should be designed to resolve the aliasing. If the results show a strong curvature, you should consider a response surface design. If the results are inconsistent with the prior knowledge, you should review the experimental protocol for errors. ## Building a Practical Decision Framework for Fractional Factorial Screening Selecting a fractional factorial design for cell culture experiments involves more than picking a resolution from a table. The practical challenge is matching the design to the biological question, the available resources, and the tolerance for ambiguity in the results. A structured decision framework helps you make these choices deliberately instead of defaulting to a familiar design. This section provides a stepwise approach to design selection, a record system for tracking decisions, and a troubleshooting method for common implementation problems. ### Step 1: Define the Screening Objective Before any design decisions, write a single sentence that states what the experiment must accomplish. A screening objective typically takes one of three forms. The first form is factor prioritization, where you need to rank a list of candidate factors by the size of their effects. The second form is go or no-go decisions, where you need to determine whether a specific factor warrants further investigation. The third form is interaction detection, where you need to know whether two factors jointly influence the response. The objective determines the minimum resolution you can accept. If the objective is factor prioritization, resolution IV is the minimum because main effects must be clear of two-factor interactions. If the objective is a go or no-go decision for a single factor, you may accept a resolution III design only if you can justify that the aliased interactions are biologically implausible. If the objective is interaction detection, you need a design that can at least indicate whether interactions exist, even if it cannot identify the specific interaction. Write the objective in one sentence and record it in the laboratory notebook before any other design work begins. This sentence becomes the reference point for all subsequent decisions. When you are tempted to add factors or change levels, check whether the change serves the stated objective. ### Step 2: Inventory the Candidate Factors List every factor that could plausibly affect the response. For cell culture, this list often includes media components, supplements, physical conditions, and handling variables. Do not filter the list at this stage. The purpose of the inventory is to capture the full set of candidates so that the design decision is based on complete information. For each candidate factor, record three pieces of information. The first is the biological rationale for inclusion. The second is the expected direction of the effect, if known. The third is the confidence in the prior evidence. This inventory becomes the basis for the factor selection decision. The inventory should also record the practical constraints for each factor. Some factors are difficult to change between runs, such as media formulation that requires bulk preparation. Other factors are easy to change, such as incubation time. The ease of changing a factor affects the randomization and blocking strategy. Factors that are difficult to change may need to be arranged in a split-plot structure, which changes the analysis and the alias structure. ### Step 3: Apply the Selection Criteria The inventory will almost always contain more factors than the design can accommodate. The selection criteria must be applied systematically to reduce the list to the number that fits the design. The criteria are applied in order of priority. The first criterion is the strength of the prior evidence. Factors with strong prior evidence of an effect should be included. The second criterion is the biological plausibility of the mechanism. A factor with a plausible mechanism is more likely to show a real effect than a factor included without a mechanistic rationale. The third criterion is the practical importance. A factor that is expensive or difficult to control in routine culture may be more important to screen than a factor that is easy to adjust. The fourth criterion is the interaction potential. Factors that are likely to interact with each other should be retained because the screening design may detect the interaction. Factors that fail all four criteria are excluded. The excluded factors should be recorded with the reason for exclusion. This record is important for the interpretation of the results. If a factor is excluded and the experiment shows a large unexplained variation, the excluded factor becomes a candidate for the follow-up experiment. ### Step 3: Determine the Run Budget The run budget is the maximum number of experimental runs that can be completed within the available resources. The budget is determined by the cost of each run, the time available, the incubator capacity, and the labor available. The budget should be recorded before the design is selected. The run budget must include the runs for the design, the center points, and the replication. A common error is to set the budget equal to the number of runs in the design table and then discover that the center points and replication cannot be accommodated. The budget should be set with the full experiment in mind. The run budget also determines the maximum number of factors that can be screened. A design with 16 runs can screen up to 7 factors at resolution IV. A design with 32 runs can screen up to 15 factors at resolution IV. The relationship between the number of runs and the number of factors is not linear. The design tables show the maximum number of factors for each number of runs at each resolution. ### Step 4: Select the Design Resolution The resolution is selected based on the objective and the run budget. The decision rule is to choose the highest resolution that fits within the budget. Resolution IV is the minimum for screening because it keeps main effects clear of two-factor interactions. Resolution V is preferred if the budget allows because it also keeps two-factor interactions clear of each other. The resolution decision should be recorded with the justification. The justification should state why the chosen resolution is sufficient for the objective and why a higher resolution is not feasible within the budget. This record is important for the interpretation of the results. If the results show a significant interaction, the resolution determines whether the interaction can be identified or only detected. The design choice also includes the selection of the generators. The generators determine the alias structure. Standard design tables provide the generators for each design. The generators should be selected to minimize the aliasing of the effects that are most likely to be important. For example, if two factors are known to interact, the generators should be chosen so that the interaction between those two factors is not aliased with a main effect. ### Step 5: Verify the Alias Structure The alias structure must be verified before the experiment is run. The verification is done by writing out the complete list of aliases for the design. This list is the reference for the interpretation of the results. The verification should focus on the main effects and the two-factor interactions. For each main effect, check whether it is aliased with any other main effect or any two-factor interaction. For each two-factor interaction, check which other two-factor interactions are aliased with it. The alias structure should be recorded in the research log. The record should include the complete list of aliases, beyond the main effects. The record is used during the analysis to interpret the significant effects. ### Step 6: Plan the Follow-Up Strategy The follow-up strategy is planned before the experiment is run. The strategy describes what will be done if the results show significant effects that are aliased. The strategy should include the specific follow-up design that will be used to resolve the aliasing. The follow-up design is often a fold-over of the original design. A fold-over reverses the signs of one or more factors, which changes the alias structure. The fold-over design can resolve the aliasing between the main effects and the two-factor interactions. The follow-up design can also be a full factorial for the significant factors only. The follow-up strategy should be recorded in the research log. The record should include the trigger for the follow-up, the design of the follow-up, and the resources required. The follow-up plan is not a commitment to run the experiment. It is a plan that is ready if the results require it. ## The Decision Record System The decision framework is only useful if the decisions are recorded. The record system is a structured set of documents that captures the design decisions, the rationale, and the data. The record system serves three purposes. The first purpose is to provide the information needed for the analysis. The second purpose is to enable the interpretation of the results. The third purpose is to support the reporting of the experiment. ### The Design Decision Record The design decision record is a single document that captures the following information. The screening objective is written in one sentence. The candidate factor inventory is listed with the rationale for each factor. The selected factors are listed with the high and low levels. The excluded factors are listed with the reason for exclusion. The run budget is stated. The design is identified by the number of factors, the number of runs, and the resolution. The generators are listed. The alias structure is recorded. The randomization order is recorded. The blocking plan is recorded. The center points and replication are recorded. The design decision record is completed before the experiment is run. The record is dated and signed by the researcher. The record is stored with the data. ### The Run Log The run log is a document that records the details of each experimental run. The run log includes the run number, the date, the time, the operator, the factor levels, the response, and any notes about the run. The notes include any deviations from the protocol, any signs of contamination, and any equipment issues. The run log is completed during the experiment. The log is the primary source of data for the analysis. The log is also the source for the troubleshooting if the results are unexpected. ### The Analysis Record The analysis record is a document that captures the analysis steps and the results. The analysis record includes the effect estimates, the Pareto chart, the significance tests, and the interpretation of the alias structure. The analysis record also includes the decisions about the follow-up. The analysis record is completed after the analysis is done. The record is stored with the design decision record and the run log. ## Troubleshooting the Design The troubleshooting method is a structured approach to diagnosing problems with the design or the experiment. The method is used when the results are unexpected or when the experiment fails to produce the expected information. ### Problem: The Results Show No Significant Effects If the results show no significant effects, the first check is the range of the factor levels. The levels may be too close together to produce a detectable effect. The second check is the measurement error. The measurement method may be too noisy to detect the effects. The third check is the cell culture quality. The cells may be stressed or the culture conditions may be inconsistent. The troubleshooting record should include the checks that were done and the findings. The record should also include the decision about the next step. The next step may be to repeat the experiment with wider levels, to improve the measurement method, or to improve the cell culture quality. ### Problem: The Results Show a Significant Effect That Is Aliased If the results show a significant effect that is aliased, the problem is the design resolution. The design cannot distinguish between the aliased effects. The troubleshooting record should include the alias list and the interpretation of the significant effect. The next step is to run the follow-up design to resolve the aliasing. ### Problem: The Results Show a Large Block Effect If the results show a large block effect, the problem is the blocking strategy. The block effect may be confounded with the factor effects. The troubleshooting record should include the block structure and the analysis. The next step is to review the block design and to consider a different blocking strategy. ### Problem: The Results Show a Strong Curvature If the results show a strong curvature, the problem is the linearity assumption. The two-level design cannot model the curvature. The troubleshooting record should include the center point analysis. The next step is to run a response surface design to model the curvature. ## Comparison of Design Selection Approaches The decision framework can be compared with the common approaches to design selection. The common approaches are the default approach, the budget approach, and the resolution approach. The default approach is to use the same design that was used in a previous experiment or in a published study. The default approach is simple but does not consider the specific objective of the experiment. The default design may have a lower resolution than needed or may not include the right factors. The budget approach is to choose the design that fits the available resources. The budget approach is practical but may choose a design with a lower resolution than needed. The budget approach may also choose a design that does not include the right factors. The resolution approach is to choose the highest resolution that fits the budget. The resolution approach is statistically sound but may not consider the specific objective. The resolution approach may choose a design that is more expensive than needed. The decision framework combines the strengths of the common approaches. The framework starts with the objective, then considers the budget, then chooses the resolution, then verifies the alias structure. The framework is more deliberate than the common approaches and produces a design that is tailored to the specific experiment. ## Practical Implementation Steps The implementation of the decision framework follows a sequence of steps. The steps are designed to be completed in order, with the record at each step. 1. Write the screening objective in one sentence. 2. List all candidate factors with the rationale for each. 3. Apply the selection criteria to reduce the candidate list. 4. Determine the run budget. 5. Choose the design and the generators. 6. Verify the alias structure. 7. Record the design decision. 8. Prepare the cell culture and the materials. 9. Run the experiment in the randomized order. 10. Record the data in the run log. 11. Analyze the results. 12. Interpret the results with the alias structure. 13. Record the analysis and the decisions. 14. Plan the follow-up if needed. The steps are not optional. Each step produces a record that is needed for the next step. The record system is the foundation of the experiment. ## Comparison of Design Selection Approaches | Approach | Basis for Selection | Strength | Limitation | When to Use | |----------|---------------------|----------|------------|-------------| | Table approach | Standard design tables | Simple and verified | Does not consider the objective | When the objective is standard | | Budget approach | Available resources | Practical | May choose a low resolution | When the budget is fixed | | Resolution approach | Highest resolution | Statistically sound | May be expensive | When the budget is flexible | | Decision framework | Objective and budget | The design matches the question | Requires more planning | When the experiment is important | ## Reporting the Design Decisions The reporting of the design decisions is part of the research record. The reporting should follow the guidelines for transparent research reporting. The [EQUATOR Network](https://www.equator-network.org/) provides a list of reporting guidelines for different study types. The reporting should include the design, the factors, the levels, the resolution, the alias structure, and the randomization. The reporting should also include the data management plan. The [NIH Data Management and Sharing Policy](https://sharing.nih.gov/data-management-and-sharing-policy) describes the expectations for data management and sharing. The data from the experiment should be stored in a format that can be shared and reused. The reporting should also include the authorship and the publication ethics. The [Committee on Publication Ethics](https://publicationethics.org/core-practices) describes the core practices for authorship, peer review, and data. The design decisions should be reported honestly and completely. ## The Role of the Decision Framework in the Research Process The decision framework is a tool for the planning of the experiment. The framework is not a substitute for the statistical analysis. The framework is a way to make the design decisions explicit and recorded. The framework is also a way to communicate the design decisions to other researchers. The record of the design decisions can be shared with the collaborators and the reviewers. The record shows the rationale for the design and the interpretation of the results. The framework is also a way to teach the design of experiments. The framework provides a structured approach that can be used by the researchers who are new to the fractional factorial designs. The framework is a practical tool for the design of the screening experiments in the cell culture. The framework is also a way to improve the quality of the research. The framework ensures that the design decisions are made deliberately and recorded. The framework reduces the risk of the design errors and the interpretation errors. The framework is a contribution to the quality of the research. ## Frequently Asked Questions ### What is the difference between a full factorial and a fractional factorial design? A full factorial design tests every combination of the factor levels. A fractional factorial design tests a subset of the combinations. The fractional design uses fewer runs but introduces confounding between effects. ### How do I choose the resolution of a fractional factorial design? The resolution should be chosen based on the number of factors and the number of runs you can afford. Resolution IV is a common choice for screening because it keeps main effects clear of two-factor interactions. ### What is an alias in a fractional factorial design? An alias is an effect that is confounded with another effect. The alias list shows which effects are indistinguishable from each other. The alias list must be consulted when interpreting the results. ### Can I use a fractional factorial design for optimization? A fractional factorial design is used for screening, not for optimization. The design identifies the important factors. A response surface design is used for optimization. ### How many runs do I need for a fractional factorial design? The number of runs depends on the number of factors and the resolution. A 2^(4-1) design has 8 runs. A 2^(7-3) design has 16 runs. The number of runs is a power of two. ### What is the role of center points in a fractional factorial design? Center points are runs at the midpoint of all factor levels. They are used to test for curvature and to estimate the experimental error. Center points do not affect the estimates of the main effects. ### How do I handle the confounding in a fractional factorial design? The confounding is handled by the design and the analysis. The design is chosen to minimize the confounding. The analysis must be interpreted with the alias list. Follow-up experiments can resolve the confounding. ### What is the difference between a resolution III and a resolution IV design? A resolution III design has main effects aliased with two-factor interactions. A resolution IV design has main effects clear of two-factor interactions, but two-factor interactions are aliased with each other. Resolution IV is preferred for screening. ## Related Diagnostic Guides - [Cell Culture Media: A Guide to Selection and Optimization](/knowledge/diagnostics/microbiology/cell-culture-media-a-guide-to-selection-and-optimization) - [Animal Biology](/blog/guides/animal-biology) - [Animal Research](/blog/guides/animal-research) ## References and Further Reading - [Research Methods Resources](https://www.ncbi.nlm.nih.gov/books). National Library of Medicine. - [EQUATOR Network](https://www.equator-network.org/). EQUATOR Network. - [Core Practices](https://publicationethics.org/core-practices). Committee on Publication Ethics. - [NIH Grants and Funding](https://grants.nih.gov/). National Institutes of Health. - [ORCID for Researchers](https://info.orcid.org/researchers). ORCID. - [Data Management and Sharing Policy](https://sharing.nih.gov/data-management-and-sharing-policy). National Institutes of Health. - [Laboratory Biosafety Manual](https://www.who.int/publications/i/item/9789240011311). World Health Organization. - [NCBI Literature Resources](https://www.ncbi.nlm.nih.gov/). National Center for Biotechnology Information. - [A statistical approach to improve compound screening in cell culture media.](https://pubmed.ncbi.nlm.nih.gov/32625011). Engineering in life sciences, 2019. - [Optimization of BY-2 cell suspension culture medium for the production of a human antibody using a combination of fractional factorial designs and the response surface method.](https://pubmed.ncbi.nlm.nih.gov/23721307). Plant biotechnology journal, 2013. - [Combinatorial optimization of pathway, process and media for the production of p-coumaric acid by Saccharomyces cerevisiae.](https://pubmed.ncbi.nlm.nih.gov/38528768). Microbial biotechnology, 2024. > This article is educational and does not replace validated laboratory procedures, institutional biosafety review, manufacturer instructions, or professional interpretation.