How to Conduct Factor Analysis to Identify Underlying Psychological Constructs

Factor analysis is a powerful statistical method used by psychologists, researchers, and data scientists to uncover the hidden structure within complex datasets. By identifying underlying constructs—often called latent variables or factors—that explain patterns of correlations among observed variables, this technique has become indispensable in psychology, social sciences, marketing, and many other fields. Whether you're developing a psychological assessment tool, validating a measurement scale, or exploring the dimensional structure of personality traits, understanding how to properly conduct factor analysis is essential for rigorous research.

This comprehensive guide will walk you through everything you need to know about conducting factor analysis to identify underlying psychological constructs, from the theoretical foundations to practical implementation steps, best practices, and common pitfalls to avoid.

What Is Factor Analysis and Why Does It Matter?

Factor analysis is commonly used in psychometrics, personality psychology, biology, marketing, product management, operations research, finance, and machine learning. At its core, this statistical technique helps researchers make sense of large, complex datasets by reducing many observed variables into a smaller set of meaningful factors that represent core psychological constructs.

Psychologists frequently use factor analysis because many of their factors are inherently unobservable because they exist inside the human brain. For example, depression is a condition inside the mind that researchers can't directly observe. However, they can ask questions and make observations about different behaviors and attitudes. Depression is an invisible driver that affects many outcomes we can measure. Similarly, constructs like intelligence, anxiety, extraversion, self-esteem, and motivation cannot be directly measured but can be inferred from observable indicators.

It may help to deal with data sets where there are large numbers of observed variables that are thought to reflect a smaller number of underlying/latent variables. This data reduction capability makes factor analysis particularly valuable when working with surveys, questionnaires, or assessment batteries that contain numerous items designed to measure related psychological phenomena.

Historical Development and Theoretical Foundations

Charles Spearman was the first psychologist to discuss common factor analysis and did so in his 1904 paper. He discovered that school children's scores on a wide variety of seemingly unrelated subjects were positively correlated, which led him to postulate that a single general mental ability, or g, underlies and shapes human cognitive performance. This groundbreaking work laid the foundation for modern intelligence testing and factor analytic methods.

The initial development of common factor analysis with multiple factors was given by Louis Thurstone in two papers in the early 1930s, summarized in his 1935 book, The Vector of Mind. Thurstone introduced several important factor analysis concepts, including communality, uniqueness, and rotation. He advocated for "simple structure", and developed methods of rotation that could be used as a way to achieve such structure.

In fact, no other statistical tool has been as integral to the creation, development, and refinement of psychological measurement and theories as factor analysis. From Spearman's theory of general intelligence to the development of the Five-Factor Model of personality, factor analysis has shaped our understanding of human psychology in profound ways.

Understanding the Basics of Factor Analysis

Factor analysis simplifies a complex dataset by taking a larger number of observed variables and reducing them to a smaller set of unobserved factors. Anytime you simplify something, you're trading off exactness with ease of understanding. The goal is to identify the minimum number of factors that can adequately explain the correlations among your observed variables while maintaining interpretability.

The Factor Analysis Model

The basic premise of factor analysis is that observed variables are linear combinations of underlying factors plus some unique variance (error). Each observed variable can be expressed as a weighted sum of common factors plus a unique factor specific to that variable. The weights, called factor loadings, indicate how strongly each variable is associated with each factor.

It helps identify clusters or groups of related items on psychological tests. Variables that load highly on the same factor are presumed to measure aspects of the same underlying construct. For instance, if you're analyzing responses to a personality questionnaire, items about sociability, talkativeness, and assertiveness might all load highly on a factor you could label "extraversion."

Types of Factor Analysis

There are two primary approaches to factor analysis, each serving different research purposes:

Exploratory Factor Analysis (EFA)

Exploratory Factor Analysis is used when you don't have strong prior hypotheses about the underlying factor structure. This approach is appropriate in the early stages of research when you're trying to discover the latent structure in your data. EFA allows the data to "speak for itself" by identifying patterns of correlations without imposing predetermined constraints.

Common applications of EFA include:

Developing new psychological measures or scales
Exploring the dimensional structure of a construct
Reducing a large set of variables to a more manageable number
Identifying which items cluster together in survey data
Generating hypotheses about underlying psychological constructs

Confirmatory Factor Analysis (CFA)

Confirmatory Factor Analysis is used when you have specific hypotheses about the factor structure based on theory or previous research. CFA allows you to test whether your data fits a predetermined model, providing statistical evidence for or against your theoretical expectations.

Current Psychology focuses on Clinical psychology but the discussions also offer insight into other areas such as Psychological intervention, Confirmatory factor analysis, Association (psychology), Mental health and Depression (differential diagnoses). Current Psychology explores research in Confirmatory factor analysis and the adjacent study of Exploratory factor analysis.

CFA is particularly useful for:

Validating the structure of established psychological measures
Testing competing theoretical models
Assessing measurement invariance across different groups
Evaluating the construct validity of assessment instruments
Confirming factor structures identified in exploratory analyses

Key Concepts and Terminology

Before diving into the practical steps, it's important to understand several key concepts:

Latent Variables: Variables that are not directly observed but are inferred from other variables that are observed or measured. These are the underlying factors you're trying to identify.

Factor Loadings: The correlation coefficients between observed variables and factors. Higher loadings (typically above 0.3 or 0.4) indicate stronger relationships between variables and factors.

Communality: The proportion of variance in an observed variable that is explained by the common factors. High communalities indicate that the factors account for most of the variance in that variable.

Uniqueness: The proportion of variance in an observed variable that is not explained by the common factors, including both specific variance and measurement error.

Eigenvalues: Values that represent the amount of variance explained by each factor. Eigenvalues are used to determine how many factors to retain in your analysis.

Preparing Your Data for Factor Analysis

Proper data preparation is crucial for obtaining valid and reliable results from factor analysis. Several prerequisites must be met before proceeding with the analysis.

Sample Size Requirements

Adequate sample size is essential for stable and replicable factor solutions. While there's no universally agreed-upon minimum, several rules of thumb exist:

At least 5-10 participants per variable (some researchers recommend 10-20)
Minimum absolute sample size of 100-200 participants
Larger samples (300+) are preferable for more stable solutions
More complex factor structures require larger samples

Insufficient sample size can lead to unstable factor solutions that don't replicate in new samples, making your findings unreliable.

Data Screening and Cleaning

Before conducting factor analysis, thoroughly screen your data for:

Missing data: Decide on an appropriate strategy (deletion, imputation, or maximum likelihood estimation)
Outliers: Identify and address extreme values that might distort correlations
Normality: While factor analysis is relatively robust to violations, severe non-normality can affect results
Linearity: Factor analysis assumes linear relationships between variables
Multicollinearity: Extremely high correlations (>0.9) between variables can cause problems

Assessing Data Suitability: KMO and Bartlett's Test

Two statistical tests help determine whether your data is appropriate for factor analysis:

Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy

The Kaiser-Meyer-Olkin (KMO) statistic, which can vary from 0 to 1, indicates the degree to which each variable in a set is predicted without error by the other variables. A value of 0 indicates that the sum of partial correlations is large relative to the sum correlations, indicating factor analysis is likely to be inappropriate.

KMO values between 0.8 and 1 indicate the sampling is adequate. KMO values less than 0.6 indicate the sampling is not adequate and that remedial action should be taken. The general interpretation guidelines are:

0.90 and above: Excellent
0.80-0.89: Meritorious
0.70-0.79: Middling
0.60-0.69: Mediocre
0.50-0.59: Miserable
Below 0.50: Unacceptable

KMO is a test conducted to examine the strength of the partial correlation (how the factors explain each other) between the variables. KMO values closer to 1.0 are consider ideal while values less than 0.5 are unacceptable.

Bartlett's Test of Sphericity

Bartlett's (1951) test of sphericity tests whether a matrix (of correlations) is significantly different from an identity matrix (filled with 0). It tests whether the correlation coefficients are all 0. A significant result (p < 0.05) indicates that your variables are sufficiently correlated to proceed with factor analysis.

While it is often suggested to check whether Bartlett's test of sphericity is significant before starting with factor analysis, one needs to remember that the test is testing a pretty extreme scenario (that all correlations are non-significant). As the sample size increases, this test tends to be always significant, which makes it not particularly useful or informative in well-powered studies.

Step-by-Step Guide to Conducting Factor Analysis

Now that you understand the foundations and have prepared your data, let's walk through the essential steps for conducting a factor analysis.

Step 1: Choose Your Extraction Method

Several methods exist for extracting factors from your correlation matrix. The two most common are:

Principal Component Analysis (PCA)

Principal Component Analysis (PCA): A statistical procedure that uses orthogonal transformation to convert observations into linearly uncorrelated variables called principal components. While technically not a true factor analysis method, PCA is widely used for data reduction. It extracts components that account for maximum variance in the observed variables, including both common and unique variance.

PCA is appropriate when:

Your primary goal is data reduction rather than identifying latent constructs
You want to create composite scores from your variables
You're conducting an initial exploratory analysis

Principal Axis Factoring (PAF)

Also called common factor analysis, PAF extracts only the common variance shared among variables, excluding unique variance and error. This method is more theoretically aligned with the factor analysis model and is generally preferred when your goal is to identify underlying psychological constructs.

PAF is appropriate when:

You're interested in identifying latent constructs
You want to understand the shared variance among variables
You're developing or validating psychological measures

Other extraction methods include maximum likelihood, unweighted least squares, and generalized least squares, each with specific advantages depending on your data characteristics and research goals.

Step 2: Determine the Number of Factors to Retain

One of the most critical decisions in factor analysis is determining how many factors to extract. Multiple criteria should be used together:

Kaiser Criterion (Eigenvalues Greater Than 1)

This traditional rule suggests retaining factors with eigenvalues greater than 1.0, as these factors explain more variance than a single observed variable. While simple and widely used, this criterion often overestimates the number of factors, especially with large numbers of variables.

Scree Plot Analysis

A scree plot graphs eigenvalues against factor numbers. Look for the "elbow" where the steep decline in eigenvalues levels off. Factors before the elbow are retained. This visual method is subjective but provides valuable information about the factor structure.

Parallel Analysis

Parallel analysis compares eigenvalues from your actual data to eigenvalues from randomly generated data with the same dimensions. Retain factors whose eigenvalues from your data exceed those from random data. This method is considered one of the most accurate for determining the number of factors.

Theoretical Considerations

Don't rely solely on statistical criteria. Consider:

Theoretical expectations about the construct
Previous research findings
Interpretability of the solution
Practical utility of the factors
Proportion of variance explained (typically aim for 50-70%)

Step 3: Extract the Factors

Once you've chosen your extraction method and determined the number of factors, use statistical software to extract the factors. Most statistical packages (SPSS, R, SAS, Stata, Python) have built-in functions for factor analysis.

During extraction, the software calculates:

Factor loadings for each variable on each factor
Communalities for each variable
Eigenvalues and variance explained by each factor
The initial unrotated factor solution

The unrotated solution, while mathematically optimal, is often difficult to interpret because variables may load moderately on multiple factors without clear patterns.

Step 4: Rotate the Factors for Interpretability

Rotation transforms the factor solution to achieve simpler, more interpretable structure. The goal is to have each variable load highly on one factor and minimally on others, making it easier to identify what each factor represents.

Orthogonal Rotation (Varimax)

Orthogonal rotation methods assume factors are uncorrelated (independent) with each other. Varimax is the most popular orthogonal rotation method, maximizing the variance of squared loadings within factors.

Use orthogonal rotation when:

You have theoretical reasons to believe factors are independent
You want simpler interpretation with uncorrelated factors
You plan to use factor scores in subsequent analyses where independence is assumed

Oblique Rotation (Oblimin, Promax)

Oblique rotation methods allow factors to be correlated with each other, which is often more realistic for psychological constructs. For example, different aspects of personality or intelligence are typically related to some degree.

Use oblique rotation when:

You expect factors to be correlated (common in psychology)
You want a more realistic representation of construct relationships
You're interested in examining the correlations among factors

Oblique rotation produces two matrices: the pattern matrix (unique relationships between variables and factors) and the structure matrix (total relationships including correlations among factors). The pattern matrix is typically used for interpretation.

Step 5: Interpret and Label the Factors

Interpretation is where statistical analysis meets theoretical understanding. Examine the rotated factor loadings to identify which variables define each factor.

Guidelines for Interpretation

Loading cutoffs: Typically, loadings above 0.3 or 0.4 are considered meaningful, though some researchers use 0.5 or higher
Cross-loadings: Variables that load highly on multiple factors complicate interpretation and may need to be removed
Factor coherence: Variables loading on the same factor should make theoretical sense together
Naming factors: Choose labels that capture the essence of the variables loading on each factor

Indeed, while the analysis identifies factors, it's up to the researchers to name them! This subjective element requires deep understanding of the theoretical domain and careful consideration of what the variables have in common.

Even the act of labeling a factor constitutes an implicit theory of a construct, but this theory becomes falsifiable only if it is elaborated concretely and only when external validators are carefully selected.

Example Interpretation

Suppose you're analyzing a personality questionnaire and find that items like "I enjoy parties," "I talk to many people at social events," and "I feel comfortable in crowds" all load highly on Factor 1. You might label this factor "Sociability" or "Extraversion," depending on the theoretical framework and other items loading on the factor.

Step 6: Assess the Quality of Your Solution

Evaluate your factor solution using several criteria:

Variance explained: Do the factors explain a substantial proportion of variance (typically 50-70%)?
Communalities: Are most variables adequately explained by the factors (communalities > 0.4)?
Simple structure: Do variables load primarily on one factor with minimal cross-loadings?
Interpretability: Do the factors make theoretical sense?
Replicability: Can the solution be replicated in independent samples?

Validating Your Factor Structure

Identifying a factor structure is just the beginning. Validation is essential to ensure your findings are meaningful and generalizable.

Internal Validation

Reliability Analysis: Calculate Cronbach's alpha or other reliability coefficients for each factor. Values above 0.70 are generally considered acceptable, though higher values (0.80+) are preferable for established measures.

Item Analysis: Examine item-total correlations to ensure individual items contribute meaningfully to their factors.

External Validation

Although construct validation requires multiple steps, a particularly important component is external validity, the practice of examining a factor's relations with external correlates. Often, researchers simply correlate their latent factors (e.g., impulsive psychopathy features) with measures of other, presumably relevant constructs (e.g., substance use, number of arrests, aggression) and conclude that their factor has real, psychological meaning, and that the factor is valid if the correlations are significant (at α = .05) in their sample.

However, researchers should articulate a nomological net clearly and in advance of analysis. This means specifying expected relationships with other variables based on theory before conducting validation analyses.

External validation strategies include:

Convergent validity: Factors should correlate with measures of similar constructs
Discriminant validity: Factors should not correlate highly with measures of dissimilar constructs
Criterion validity: Factors should predict relevant outcomes or behaviors
Known-groups validity: Factors should differentiate between groups expected to differ on the construct

Cross-Validation and Replication

The gold standard for validation is replication in independent samples. Consider:

Splitting your sample and conducting EFA in one half and CFA in the other
Collecting new data to test the factor structure
Testing measurement invariance across different populations
Examining the stability of the factor structure over time

Common Challenges and How to Address Them

Dealing with Cross-Loadings

When variables load substantially on multiple factors, consider:

Removing problematic items if they don't have strong theoretical justification
Trying different rotation methods
Extracting a different number of factors
Examining whether the item is genuinely multidimensional

Low Communalities

Variables with low communalities (< 0.4) are poorly explained by the factor solution. Options include:

Removing the variable if it doesn't contribute meaningfully
Checking for data quality issues
Considering whether additional factors are needed
Examining whether the variable measures something distinct from other items

Heywood Cases

Heywood cases occur when communalities exceed 1.0 or when negative variance estimates appear. This indicates model misspecification. Solutions include:

Reducing the number of factors
Removing problematic variables
Using a different extraction method
Checking for multicollinearity

Subjective Interpretation

Interpreting factor analysis is based on using a "heuristic", which is a solution that is "convenient even if not absolutely true". More than one interpretation can be made of the same data factored the same way, and factor analysis cannot identify causality.

To minimize subjectivity:

Ground interpretations in established theory
Consult with subject matter experts
Use multiple criteria for decision-making
Be transparent about interpretive choices
Validate findings through external criteria

Software and Tools for Factor Analysis

Several statistical software packages can perform factor analysis:

SPSS

User-friendly interface with point-and-click options for factor analysis. Provides comprehensive output including KMO, Bartlett's test, scree plots, and various rotation options. Ideal for researchers without extensive programming experience.

R

The psych package is a great tool for assessing underlying latent structure. It can provide reliability statistics, do cluster analysis, principal components analysis, mediation models, and, of course factor analysis. R offers powerful, flexible options through packages like psych, lavaan (for CFA), and FactoMineR. Free and open-source with extensive documentation and community support.

Python

Libraries like scikit-learn, factor_analyzer, and statsmodels provide factor analysis capabilities. Excellent for integrating factor analysis into larger data science workflows.

SAS and Stata

Professional statistical software with robust factor analysis procedures. Commonly used in academic and corporate research settings.

Mplus

Specialized software for structural equation modeling and advanced factor analysis. Particularly strong for CFA and complex measurement models.

Best Practices and Recommendations

Planning Your Analysis

Start with theory: Have clear theoretical expectations about potential constructs before analyzing data
Design quality measures: Ensure your variables are reliable and valid measures of the constructs of interest
Plan adequate sample size: Collect sufficient data for stable factor solutions
Preregister when possible: Specify your analysis plan in advance to reduce researcher degrees of freedom

Conducting the Analysis

Use multiple criteria: Don't rely on a single method for determining the number of factors
Try different solutions: Explore alternative numbers of factors and rotation methods
Document decisions: Keep detailed records of analytical choices and rationale
Check assumptions: Verify that your data meets the requirements for factor analysis
Examine residuals: Look at the residual correlation matrix to assess model fit

Interpreting and Reporting Results

Be transparent: Report all relevant statistics and decisions made during analysis
Provide complete information: Include factor loadings, communalities, eigenvalues, and variance explained
Acknowledge limitations: Discuss the subjective elements of interpretation
Consider alternatives: Acknowledge that other interpretations may be possible
Validate findings: Use confirmatory factor analysis or external validation when possible

Advancing Your Research

Replicate findings: Test your factor structure in new samples
Test invariance: Examine whether the factor structure holds across different groups
Examine external validity: Investigate relationships with theoretically relevant variables
Refine measures: Use factor analysis results to improve your assessment instruments
Build theory: Use findings to advance theoretical understanding of psychological constructs

Applications in Psychological Research

Personality Assessment

It also has been used to find factors in a broad range of domains such as personality, attitudes, beliefs, etc. Factor analysis has been instrumental in developing major personality models, including the Big Five (Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism). Researchers use factor analysis to identify personality dimensions, develop assessment tools, and understand the structure of individual differences.

Intelligence and Cognitive Abilities

Factor analysis in psychology is most often associated with intelligence research. From Spearman's g factor to modern theories of multiple intelligences, factor analysis continues to shape our understanding of cognitive abilities. Researchers use it to identify distinct cognitive domains, develop intelligence tests, and investigate the structure of mental abilities.

Clinical Psychology and Psychopathology

Factor analysis helps identify dimensions of mental health symptoms, validate diagnostic criteria, and develop assessment instruments for conditions like depression, anxiety, PTSD, and personality disorders. It can reveal whether symptoms cluster into distinct syndromes or exist on continua.

Attitude and Opinion Research

In social psychology and marketing, factor analysis identifies underlying dimensions of attitudes, values, and beliefs. This helps researchers understand what drives opinions and behaviors, develop persuasive interventions, and segment populations based on psychological characteristics.

Scale Development and Validation

It is linked to psychometrics, as it can assess the validity of an instrument by finding if the instrument indeed measures the postulated factors. Factor analysis is essential for developing and validating psychological measures, ensuring that assessment tools actually measure what they claim to measure.

Advanced Topics and Extensions

Hierarchical Factor Analysis

Some constructs have hierarchical structure with both lower-order and higher-order factors. For example, intelligence might have specific abilities (verbal, spatial, numerical) as first-order factors and general intelligence (g) as a second-order factor. Hierarchical factor analysis models these multilevel structures.

Bifactor Models

Bifactor models include a general factor that influences all variables plus specific factors that capture additional variance in subsets of variables. These models are useful when you have both a general construct and specific facets.

Measurement Invariance Testing

When comparing groups (e.g., different cultures, genders, or time points), it's important to establish that your measure functions equivalently across groups. Measurement invariance testing uses CFA to verify that the factor structure, loadings, and intercepts are comparable across groups.

Item Response Theory (IRT)

While factor analysis operates at the scale level, IRT provides item-level analysis of measurement properties. IRT and factor analysis are related approaches that can be used complementarily to develop and validate psychological measures.

Structural Equation Modeling (SEM)

CFA is actually a component of the broader SEM framework, which allows you to test complex relationships among latent variables. Once you've established your measurement model through CFA, you can incorporate it into structural models that test theoretical relationships among constructs.

Ethical Considerations and Responsible Use

Avoiding P-Hacking and HARKing

The flexibility in factor analysis (choosing extraction methods, number of factors, rotation methods, etc.) creates opportunities for questionable research practices. Avoid:

Running multiple analyses and reporting only the most favorable results
HARKing (Hypothesizing After Results are Known) by presenting exploratory findings as confirmatory
Selectively removing items to achieve desired factor structures
Failing to report analyses that didn't support hypotheses

Cultural Sensitivity and Bias

Factor structures may differ across cultural groups. What appears as a single construct in one culture might be multidimensional in another. Always consider:

Cultural appropriateness of items and constructs
Translation and adaptation issues
Testing measurement invariance across cultural groups
Avoiding imposing Western psychological constructs on non-Western populations

Responsible Interpretation and Application

Remember that factors are statistical constructs, not necessarily real entities. Be cautious about:

Reifying factors as if they have independent existence
Making causal claims based on correlational factor analysis
Using factor scores for high-stakes decisions without adequate validation
Overgeneralizing findings beyond the population studied

Resources for Further Learning

To deepen your understanding of factor analysis and stay current with methodological developments, consider exploring these resources:

Tabachnick and Fidell's "Using Multivariate Statistics" provides comprehensive coverage of factor analysis and related techniques
Brown's "Confirmatory Factor Analysis for Applied Research" offers detailed guidance on CFA
Costello and Osborne's articles on best practices in exploratory factor analysis
Fabrigar and Wegener's "Exploratory Factor Analysis" provides theoretical and practical guidance

Online Resources

Statistical software documentation and tutorials (SPSS, R, Python)
Academic journals publishing methodological articles on factor analysis
Online courses and workshops on multivariate statistics
Professional organizations like the American Psychological Association and Society for Multivariate Experimental Psychology

Staying Current

Factor analysis methodology continues to evolve. Stay informed about:

New developments in estimation methods and fit indices
Advances in handling missing data and non-normal distributions
Innovations in software and computational approaches
Debates about best practices and methodological controversies
Applications of machine learning to factor analysis

For additional guidance on statistical methods in psychology, you might find resources from the American Psychological Association helpful, as well as methodological articles in journals like Psychological Methods and Multivariate Behavioral Research. The Association for Psychological Science also provides valuable resources for researchers.

Conclusion: Mastering Factor Analysis for Psychological Research

Factor analysis is a powerful tool for uncovering the hidden structure underlying psychological phenomena. By systematically reducing complex datasets to interpretable factors, researchers can identify core psychological constructs, develop valid measurement instruments, and advance theoretical understanding of human behavior and mental processes.

Success in factor analysis requires a combination of statistical knowledge, theoretical understanding, and careful judgment. While the technique provides valuable insights, it's important to remember that factor solutions are models—simplified representations of reality that should be evaluated based on their usefulness, interpretability, and empirical support.

By following the steps outlined in this guide—from careful data preparation and appropriate extraction methods to thoughtful interpretation and rigorous validation—you can conduct factor analyses that contribute meaningfully to psychological science. Always use multiple criteria for decision-making, ground your interpretations in theory, validate your findings through replication and external criteria, and remain transparent about the choices made during analysis.

As you gain experience with factor analysis, you'll develop intuition about when different approaches are appropriate, how to troubleshoot common problems, and how to integrate factor analytic findings into broader research programs. Whether you're developing a new assessment tool, exploring the structure of a psychological construct, or validating an existing measure, factor analysis provides a rigorous framework for understanding the latent variables that shape human psychology.

Remember that factor analysis is not just a statistical technique—it's a way of thinking about psychological measurement and construct validity. By mastering this approach, you'll be better equipped to develop theories, create measures, and conduct research that advances our understanding of the human mind and behavior. The journey from observed variables to meaningful psychological constructs requires patience, rigor, and creativity, but the insights gained make factor analysis an indispensable tool in the psychological researcher's toolkit.