Factorial Designs: How to Study Multiple Variables Efficiently

Factorial designs let you study two or more factors in a single experiment, measuring both the individual effect of each factor and whether factors interact with one another. For farmers and animal scientists, this means you can test combinations of treatments, genetics, diets, or management practices in one trial instead of running separate experiments for each variable. A 2x2 factorial design, the simplest form, uses two factors at two levels each, creating four treatment combinations. This approach gives you more information per animal or plot than testing one variable at a time, and it reveals interaction effects that single-factor experiments cannot detect.

What Factorial Designs Do That Single-Factor Experiments Cannot

A factorial design assigns every combination of factor levels to experimental units. In a 2x2 design with factors A and B, you have four groups: A1B1, A1B2, A2B1, and A2B2. Each observation contributes information about the effect of factor A, the effect of factor B, and the interaction between A and B. This efficiency is the central advantage of factorial designs.

The one-variable-at-a-time approach, where you hold all factors constant except one, cannot detect interactions. If the effect of diet depends on breed, or the effect of a vaccine depends on age at administration, a one-factor-at-a-time experiment will miss that relationship. Factorial designs reveal these dependencies because every combination is tested.

Evidence from animal research supports this efficiency. Factorial experimental designs have been used to study controllable variables such as treatment, sex, strain, age, diet, and prior treatment of animals on a defined response. These designs generally provide more information than the one-variable-at-a-time approach because each animal contributes information on the effect of every factor, and because such designs can highlight interactions among variables. In drug discovery settings, factorial designs have effectively reduced the numbers of animals used in routine screens while maintaining or improving the information gained (Reduction in laboratory animal use by factorial design).

The practical implication for farm research is direct. If you want to know whether a new feed additive works across both breeds you raise, or whether a vaccination protocol performs differently in young versus older animals, a factorial design answers those questions in one trial. Running separate experiments for each factor would require more animals, more time, and more resources, and would still leave the interaction question unanswered.

Core Principles of Factorial Design

Factors, Levels, and Treatment Combinations

A factor is a categorical or continuous variable you manipulate, such as diet type, stocking density, breed, or vaccination status. A level is a specific value or category of that factor. In a 2x2 design, each factor has two levels. The total number of treatment combinations equals the product of the number of levels for each factor. A 2x2 design has four combinations, a 2x3 design has six, and a 3x3 design has nine.

The term "full factorial" means every possible combination of factor levels is tested. A "fractional factorial" tests only a subset of combinations, which reduces the number of experimental units needed but sacrifices the ability to estimate some higher-order interactions. Fractional designs are useful when many factors are being explored and the number of treatment combinations would otherwise become unmanageable (Reduction in laboratory animal use by factorial design).

Main Effects and Interactions

The main effect of a factor is the average difference in response between its levels, averaged across all levels of the other factors. The interaction is the extent to which the effect of one factor depends on the level of the other factor.

Consider a hypothetical example with weaning weight as the response. Factor A is creep feed (present or absent) and factor B is breed (Breed X or Breed Y). If creep feed increases weaning weight by 5 kg in Breed X but by only 1 kg in Breed Y, there is an interaction between creep feed and breed. The main effect of creep feed, averaged across breeds, would be 3 kg, but that average hides the breed-specific responses. Reporting only the main effect would mislead a farmer who raises Breed Y, where the response to creep feed is small.

Interaction effects can be synergistic, where the combined effect is larger than the sum of individual effects, or antagonistic, where the combined effect is smaller. In a study of pesticide and heavy metal contamination in sediment, second-order interaction effects of pollutant concentrations dominated the adsorption of the pesticide metalaxyl, contributing 79 percent of the explained variation compared with 21 percent for main effects. The interaction effects included both synergistic and antagonistic components, showing that the combined pollutants did not behave additively (Adsorption Mechanism of Metalaxyl Pesticide in Pesticide/Heavy Metal Sediment Using Fractional Factorial Design/Fixed Effects Models).

Replication and Randomization

Replication means assigning each treatment combination to multiple experimental units. Without replication, you cannot estimate experimental error, and you cannot test whether observed differences are statistically significant. The number of replicates needed depends on the variability of your response variable and the size of the effect you want to detect.

Randomization means assigning treatments to experimental units by chance. This protects against confounding, where an uncontrolled variable systematically varies with your treatment. If all animals receiving treatment A are in one pen and all animals receiving treatment B are in another pen, you cannot separate the treatment effect from the pen effect. Randomization spreads uncontrolled variation across treatment groups.

The Experimental Design Assistant from the NC3Rs provides a free tool for planning experiments with proper randomization and blinding. This resource is designed for animal researchers and helps identify design weaknesses before the experiment begins.

At a Glance: Factorial Design Decision Table

Setting Up a 2x2 Factorial Experiment

Step 1: Define the Research Question and Response Variables

Start with a clear question that involves two factors. For a farm setting, examples include:

• Does the response to a probiotic depend on whether calves receive colostrum within 6 hours of birth?
• Does the effect of pasture stocking rate differ between two grass varieties?
• Does the growth response to a feed additive depend on the protein level of the base ration?

Define the response variable precisely. Weight gain, feed conversion ratio, mortality, milk yield, egg production, and disease incidence are common farm responses. Decide how and when you will measure it. If you measure weight gain, specify the start and end weights, the scale used, and the timing of measurements.

Step 2: Choose Factors and Levels

Select factors that are relevant to your production system and that you can manipulate. Levels should represent practically meaningful contrasts. For a continuous factor like stocking rate, choose two levels that differ enough to produce a detectable effect but are both within the range of normal farm practice.

Avoid choosing levels so close together that the response difference is too small to detect, or so far apart that one level is impractical or unethical. If you are testing a drug or vaccine, the levels must follow the approved label directions, and any off-label use requires veterinary oversight.

Step 3: Determine the Number of Replicates

The number of replicates per treatment combination depends on the variability of your response and the effect size you want to detect. More variable responses and smaller effect sizes require more replicates. A statistical power analysis, which you can perform with standard software or with the Experimental Design Assistant, helps you determine the minimum number of animals or plots needed.

For a 2x2 design with four treatment combinations and n replicates per combination, the total number of experimental units is 4n. If you have 40 animals available, you can assign 10 to each combination. If you have only 20 animals, you can assign 5 to each combination, but your ability to detect small effects will be limited.

Step 4: Randomize and Blind

Assign animals or plots to treatment combinations using a random process. The Experimental Design Assistant can generate randomized treatment schedules. If possible, keep the people measuring outcomes unaware of treatment assignments. Blinding prevents bias in measurement, especially for subjective outcomes like body condition scoring or disease severity assessment.

Step 5: Conduct the Experiment and Record Data

Run the experiment according to your protocol. Record all data promptly and completely. Include the date, animal identification, treatment assignment, and the measured response. Also record any deviations from the protocol, such as an animal that became ill or a pen that flooded, because these events can affect your results.

Calculating Main Effects and Interactions

The 2x2 Calculation Table

For a 2x2 factorial design with equal replication, you can calculate main effects and interactions using a simple table of treatment means. Let the four treatment means be:

• Mean A1B1 = response when factor A is at level 1 and factor B is at level 1
• Mean A1B2 = response when factor A is at level 1 and factor B is at level 2
• Mean A2B1 = response when factor A is at level 2 and factor B is at level 1
• Mean A2B2 = response when factor A is at level 2 and factor B is at level 2

The main effect of factor A is the average of the A2 means minus the average of the A1 means:

Main effect of A = [(A2B1 + A2B2) / 2] - [(A1B1 + A1B2) / 2]

The main effect of factor B is the average of the B2 means minus the average of the B1 means:

Main effect of B = [(A1B2 + A2B2) / 2] - [(A1B1 + A2B1) / 2]

The interaction effect is half the difference between the effect of A at B2 and the effect of A at B1:

Interaction = [(A2B2 - A1B2) - (A2B1 - A1B1)] / 2

This interaction formula can also be written as:

Interaction = [(A2B2 + A1B1) - (A2B1 + A1B2)] / 2

Worked Example with a Worksheet

Suppose you test two factors in a beef cattle trial. Factor A is implant use (no implant versus implant) and factor B is protein supplement (low versus high). Average daily gain in kilograms per day for each treatment combination is:

Main effect of implant = [(1.30 + 1.55) / 2] - [(1.10 + 1.25) / 2] = 1.425 - 1.175 = 0.25 kg/day

Main effect of protein = [(1.25 + 1.55) / 2] - [(1.10 + 1.30) / 2] = 1.40 - 1.20 = 0.20 kg/day

Interaction = [(1.55 - 1.25) - (1.30 - 1.10)] / 2 = [0.30 - 0.20] / 2 = 0.05 kg/day

The positive interaction indicates that the combined effect of implant and high protein is slightly larger than the sum of the individual effects. The implant effect is 0.20 kg/day at low protein (1.30 - 1.10) and 0.30 kg/day at high protein (1.55 - 1.25). The response to implant is larger when protein is high.

Using Statistical Software

The hand calculations above work for balanced designs with equal replication. For unbalanced designs, or when you need formal significance tests, use statistical software. Analysis of variance for a factorial design partitions the total variation into components for factor A, factor B, the A by B interaction, and error. The F-tests for each component tell you whether the observed effects are statistically significant.

The GFD package for R provides tools for analyzing general factorial designs, including situations where assumptions of normality and homogeneity of variance are violated. This package is freely available and can handle complex designs that standard analysis of variance procedures cannot.

Options and Tradeoffs in Factorial Design

Full Factorial Versus Fractional Factorial

A full factorial design tests every combination of factor levels. This provides complete information about all main effects and interactions but requires more experimental units as the number of factors grows. A 2x3 design has 6 combinations, a 2x4 design has 8, and a 3x3 design has 9. With many factors, the number of combinations grows rapidly.

A fractional factorial design tests only a subset of combinations. This reduces the number of experimental units needed but confounds some effects, meaning you cannot separately estimate certain interactions. Fractional designs are most useful in initial screening experiments where you want to identify which of many factors have important effects before running a more detailed follow-up experiment.

Evidence from maize breeding shows that sparse factorial designs can be as efficient as tester designs for genomic prediction, and for some traits they outperform tester designs. A 363-hybrid factorial design obtained by crossing 90 dent and flint inbred lines was evaluated for silage in eight environments and used to predict independent performances of a 951-hybrid factorial design. At the same number of hybrids and lines, the factorial design was as efficient as the tester designs, and for some traits outperformed them (Genomic prediction of hybrid performance).

Response Surface Designs

When factors are continuous, such as temperature, pH, or concentration, response surface designs allow you to model the response as a smooth function of the factors. These designs use more than two levels per factor and can identify optimal factor combinations. A central composite design, for example, combines a factorial design with additional axial and center points.

In a study of the bacterium Ralstonia solanacearum, a central composite rotational design was used to evaluate culture media with different concentrations of sucrose and glucose and different pH values in the inoculum phase. The independent variables were pH and sugar concentration, and the dependent variables were optical density, dry cell weight, and polymer yield. The highest cell growth was obtained when sucrose was used at a concentration above 35 g per liter in combination with an acidic pH (Complete factorial design to adjust pH and sugar concentrations).

For animal production research, response surface designs are useful when you want to find the optimal level of a continuous factor, such as the ideal protein concentration in a ration or the optimal stocking density. The tradeoff is that these designs require more treatment combinations and more experimental units than a simple 2x2 design.

Factorial Designs in Complex Systems

Factorial designs are not limited to two factors. Multi-factorial designs can include three, four, or more factors. The complexity of interpretation increases with each additional factor, especially when higher-order interactions are significant.

In pharmacokinetics, multi-factorial interactions can cause additive, synergistic, or opposing effects on drug exposure. Factors such as genetics, disease, polypharmacy, and natural product use can each alter drug metabolism, and their combined effects are not always predictable from single-factor studies. Determining the magnitude and direction of these complex multi-factorial effects requires understanding the rate-limiting processes for each drug (Multi-factorial pharmacokinetic interactions).

For farm research, this means that a feed additive that works well in healthy animals might behave differently in animals with subclinical disease, or an interaction between genetics and nutrition might only appear when both factors are varied together. Factorial designs are the only experimental approach that can reveal these dependencies.

Observations and Measurements in Factorial Experiments

What to Measure

The response variables you measure should directly address your research question. Common measurements in animal research include:

• Growth performance: body weight, average daily gain, feed intake, feed conversion ratio
• Reproductive outcomes: conception rate, litter size, weaning weight
• Health outcomes: mortality, morbidity, treatment incidence, lesion scores at slaughter
• Product quality: carcass weight, fat depth, milk composition, egg quality
• Behavioral outcomes: activity levels, aggression, stereotypic behaviors

Measure all response variables on every experimental unit. Missing data complicates the analysis and reduces statistical power. Plan your measurement schedule in advance and ensure that all personnel are trained in the measurement procedures.

Recording Data

Use a standardized data recording form, either paper or electronic. Each record should include:

• Animal or plot identification
• Treatment assignment
• Date and time of measurement
• The measured value
• Initials of the person who made the measurement
• Any notes about deviations or unusual observations

Electronic data capture reduces transcription errors and makes the data easier to analyze. If you use paper forms, check them for completeness before entering data into a computer.

Quality Control

Quality control in a factorial experiment involves verifying that treatments were applied correctly, measurements were made accurately, and data were recorded without errors. Check treatment application records against the randomization schedule. Verify that scales and other measurement equipment were calibrated. Review data entry for obvious errors, such as values outside the plausible range.

The Research Data Framework from the National Institute of Standards and Technology provides guidance on managing research data throughout its lifecycle. Good data management practices, including documentation of methods and storage of raw data, support the credibility and reproducibility of your findings.

Common Failure Patterns in Factorial Experiments

Confounding Due to Poor Randomization

If treatment groups are not randomized, an uncontrolled variable can be confounded with the treatment. For example, if all animals receiving treatment A are housed in the pens closest to the feed alley, and those pens happen to have better drainage, you cannot tell whether differences in response are due to the treatment or the pen location.

Prevention: Use a formal randomization procedure, such as a random number generator or the Experimental Design Assistant, to assign treatments to experimental units. Record the randomization schedule and verify that it was followed.

Insufficient Replication

With too few replicates, the experiment lacks power to detect real effects. A non-significant result may simply mean the experiment was too small, not that the treatment has no effect. This is a common failure in farm research where animal numbers are limited by cost or availability.

Prevention: Conduct a power analysis before starting the experiment. If you cannot achieve the required number of replicates, consider using a fractional factorial design or reducing the number of factors studied.

Ignoring Interactions

Analyzing only main effects when a significant interaction is present can lead to incorrect conclusions. If the effect of factor A depends on the level of factor B, the main effect of A averaged across B levels may not apply to any specific B level.

Prevention: Always test for interactions before interpreting main effects. If the interaction is significant, describe the effect of each factor within each level of the other factor instead of reporting only the main effects.

Pseudoreplication

Pseudoreplication occurs when treatments are applied to groups but the analysis treats individual animals as independent units. If a treatment is applied to a pen of animals, the pen is the experimental unit, not the individual animal. Analyzing individual animals as if they were independent inflates the apparent sample size and can produce false positive results.

Prevention: Identify the experimental unit before the experiment. The experimental unit is the smallest unit to which a treatment is independently applied. Analyze the data at the level of the experimental unit.

Violation of Statistical Assumptions

The classical factorial F-test depends on normality and homogeneity of variance assumptions. When these assumptions are violated, the type I error rate can be inflated and the power of the test decreased. Nonparametric tests have been proposed to analyze interaction effects in factorial designs when assumptions are violated. Simulation studies show that the adjusted rank transform and adjusted median transform tests can be effectively used to analyze interaction effects in 2x2 factorial designs when normality assumptions are not met (The comparison of nonparametric statistical tests for interaction effects in factorial design).

Prevention: Check the distribution of your response variable and the equality of variances across treatment groups before relying on standard analysis of variance. If assumptions are violated, use appropriate alternative methods or transformations.

Welfare and Safety Context

Animal Welfare Considerations

Factorial designs can reduce the total number of animals needed for research by allowing multiple factors to be studied in one experiment. This aligns with the principle of reduction in animal research, which seeks to minimize the number of animals used while maintaining scientific validity. Factorial designs have been used successfully to reduce the numbers of animals used in routine drug screening experiments (Reduction in laboratory animal use by factorial design).

However, factorial designs can also create treatment combinations that raise welfare concerns. A combination of the lowest nutrition level and the highest stocking density might produce unacceptable welfare outcomes, even if each factor alone is acceptable. Review all treatment combinations for welfare implications before starting the experiment.

Safety Considerations

If your experiment involves drugs, vaccines, or other regulated products, follow all label directions and applicable regulations. Any use outside the approved label requires veterinary oversight and appropriate documentation. Withdrawal periods for meat, milk, or eggs must be observed if treated animals will enter the food supply.

The EQUATOR Network provides reporting guidelines for health research, including animal studies. Following these guidelines improves the completeness and transparency of your research reporting.

Professional Escalation Criteria

Consult a veterinarian or animal scientist if you observe any of the following during a factorial experiment:

• Unexpected mortality or morbidity in any treatment group
• Signs of pain, distress, or poor welfare that may be related to a treatment combination
• Treatment effects that are much larger or smaller than expected
• Data patterns that suggest a problem with treatment application or measurement

A professional can help you determine whether the experiment should be modified or stopped, and whether animals need treatment or removal from the study.

Limitations of Factorial Designs

Interpretation Complexity

As the number of factors increases, the interpretation of results becomes more complex. A significant three-way interaction means the effect of factor A depends on both factor B and factor C, which is difficult to describe and apply in practice. Higher-order interactions are often difficult to interpret biologically.

Resource Requirements

Full factorial designs require more experimental units than single-factor designs when the number of factors is large. The number of treatment combinations grows multiplicatively, and each combination needs replication. For farm research with limited animal numbers, this can be prohibitive.

Applicability to Long-Lived or Large Animals

Factorial designs are challenging to perform with long-lived organisms or at the community and ecosystem levels. The time and space required to test all combinations of factors can be impractical. In multiple-driver research, full factorial response surface and other advanced designs remain the most promising path toward linking experiments and theory, but they are difficult to implement with long-lived organisms (Designing More Informative Multiple-Driver Experiments).

Assumption Dependence

The validity of factorial analysis depends on statistical assumptions, including normality and homogeneity of variance. When these assumptions are violated, the type I error rate can be inflated and power decreased. Alternative nonparametric methods exist but may have their own limitations (The comparison of nonparametric statistical tests for interaction effects in factorial design).

Records and Documentation

What to Document

Maintain a complete record of your factorial experiment, including:

• The research question and hypotheses
• The factors, levels, and treatment combinations
• The randomization schedule
• The number of replicates per treatment combination
• The response variables and measurement methods
• The dates and times of treatment application and measurement
• All raw data and any data transformations
• Any deviations from the protocol and their reasons

Data Management

Store raw data in a format that can be accessed and analyzed. Electronic spreadsheets are common, but consider using a database or statistical software format for larger datasets. Back up your data regularly and keep a copy in a separate location.

The Research Data Framework provides guidance on data management practices, including documentation, storage, and sharing. Good data management supports the credibility of your findings and allows others to verify or build on your work.

Reporting

Report your factorial experiment according to established guidelines. The EQUATOR Network provides reporting guidelines for health research, and the Experimental Design Assistant can help you document the design decisions you made. Transparent reporting allows readers to assess the validity of your conclusions and to replicate your methods.

Frequently Asked Questions

What is the difference between a main effect and an interaction effect?

A main effect is the average effect of one factor across all levels of the other factors. An interaction effect is the extent to which the effect of one factor depends on the level of another factor. If the effect of diet on weight gain differs between breeds, there is an interaction between diet and breed. Main effects alone cannot reveal this dependency.

How many animals or plots do I need for a 2x2 factorial design?

The number depends on the variability of your response variable, the size of the effect you want to detect, and the statistical significance level you choose. A power analysis can help you determine the minimum number of replicates. With four treatment combinations and n replicates per combination, the total number of experimental units is 4n.

Can I use a factorial design if my factors have more than two levels?

Yes. A 2x3 design has one factor at two levels and one factor at three levels, creating six treatment combinations. A 3x3 design has nine combinations. The same principles of main effects and interactions apply, but the calculations are more complex and statistical software is recommended.

What is a fractional factorial design and when should I use it?

A fractional factorial design tests only a subset of all possible treatment combinations. This reduces the number of experimental units needed but confounds some effects, meaning you cannot separately estimate certain interactions. Fractional designs are useful for initial screening when many factors are being explored and the number of treatment combinations would otherwise be unmanageable.

What should I do if my interaction effect is significant?

If the interaction is significant, describe the effect of each factor within each level of the other factor. Do not rely on main effects alone, because the main effect averaged across levels may not apply to any specific level. Present the treatment means in a table or interaction plot to show how the response changes across combinations.

Can factorial designs reduce the number of animals used in research?

Yes. Factorial designs generally provide more information than the one-variable-at-a-time approach because each animal contributes information on the effect of every factor and because such designs can highlight interactions among variables. In drug discovery settings, factorial designs have effectively reduced the numbers of animals used in routine screens.

What are the common mistakes in factorial experiments?

Common mistakes include poor randomization leading to confounding, insufficient replication reducing statistical power, ignoring significant interactions, pseudoreplication where individual animals are treated as independent when treatments were applied to groups, and violation of statistical assumptions such as normality and homogeneity of variance.

When should I consult a statistician or animal scientist?

Consult a professional before starting the experiment if you are unsure about the number of replicates needed, the appropriate design for your question, or the analysis methods. Consult during the experiment if you observe unexpected mortality, signs of distress, or data patterns that suggest a problem with treatment application or measurement.

Related Articles

• biology study guide
• What Is a Buffer in Biology? Definition and Examples
• What Is a Stimulus in Biology? Definition and Examples
• What Is a Trait in Biology? Definition and Examples
• Carrying Capacity in Biology: Definition and Examples

References and Further Reading

• Research Data Framework. National Institute of Standards and Technology.
• EQUATOR Network. EQUATOR Network.
• Experimental Design Assistant. NC3Rs.
• NCBI Literature Resources. National Center for Biotechnology Information.
• PubMed. National Library of Medicine.
• Metabolomics in hypertension.. Journal of hypertension, 2014.
• Multi-factorial pharmacokinetic interactions: unraveling complexities in precision drug therapy.. Expert opinion on drug metabolism & toxicology, 2021.
• Genomic prediction of hybrid performance: comparison of the efficiency of factorial and tester designs used as training sets in a multiparental connected reciprocal design for maize silage.. TAG. Theoretical and applied genetics. Theoretische und angewandte Genetik, 2022.
• Designing More Informative Multiple-Driver Experiments.. Annual review of marine science, 2024.
• Solid lipid nanoparticles optimized by 2(2) factorial design for skin administration: Cytotoxicity in NIH3T3 fibroblasts.. Colloids and surfaces. B, Biointerfaces, 2018.
• Reduction in laboratory animal use by factorial design.. Alternatives to laboratory animals : ATLA, 2004.
• Factorial design applied to LC-ESI-QTOF mass spectrometer parameters for untargeted metabolomics.. Analytical methods : advancing methods and applications, 2023.
• Complete factorial design to adjust pH and sugar concentrations in the inoculum phase of Ralstonia solanacearum to optimize P(3HB) production.. PloS one, 2017.
• Taxonomic Restructuring of Rhizosphere Guilds is Driven By Agricultural Management and Scion Genotype in Apple.. 2026.
• Chewing modulates theta oscillation and functional connectivity of the frontocentral cortex in attention and working memory.. 2025.
• Rapid Verbal Persuasion to increase influenza vaccine uptake: protocol for a randomized hybrid type 2 effectiveness -implementation trial.. 2025.
• Effect of emotion on hippocampal-dependent associative binding through the lens of the weather prediction task.. 2026.
• Using artificial neural networks to reveal the human confidence computation.. 2025.
• The role of gamification in sustainable teacher education: Examining concept images of curriculum elements using a solomon four-group design.. 2026.
• Man vs. machine: Multi-country experimental evidence on the quality and perceptions of AI-generated research blog content.. 2026.
• The comparison of nonparametric statistical tests for interaction effects in factorial design. Decision Science Letters, 2019.
• Adsorption Mechanism of Metalaxyl Pesticide in Pesticide/Heavy Metal Sediment Using Fractional Factorial Design/Fixed Effects Models. 2020.
• Rheological investigation of synthetic-based drilling fluid containing non-ionic surfactant pentaerythritol ester using full factorial design. 2021.
• Mechanical optimization of virgin and recycled poly(ethylene terephthalate) biocomposites with sustainable biocarbon through a factorial design. 2020.
• Benefits of a factorial design focusing on inclusion of female and male animals in one experiment. Journal of molecular medicine, 2019.
• Implications of interaction effects on the interpretation of factorial experiments in biology. Ecologia Austral, 2001.
• GFD: An R package for the analysis of general factorial designs. Journal of Statistical Software, 2017.

This article is educational and does not replace institutional policy, professional advice, or applicable safety and regulatory requirements.