Methods and Capabilities
The services we provide rest on a common set of psychometric and statistical methods. This page describes the technical toolkit rather than any particular engagement.
Which methods apply depends on the question, the structure of the data, and the level of precision the decision requires. Selecting appropriately among them is a substantial part of the work.
Test Theory and Measurement
Methods used to build assessments, evaluate how items perform, and establish that scores mean what they are intended to mean.
Classical Test Theory Analysis
Item difficulty, item discrimination, point-biserial correlations, and distractor analysis statistics used to evaluate item quality and inform test assembly decisions.
Item Response Theory Modeling
One, two, and three-parameter logistic models as well as polytomous IRT models, applied to calibrate items, estimate person ability, and support adaptive testing applications.
Test Form Equating
Equipercentile, IRT-based, and other equating methods applied when multiple test forms must be interchangeable, ensuring that scores from different forms are on the same scale.
Differential Item Functioning Analysis
Evaluation of whether items perform differently for subgroups defined by gender, ethnicity, or other characteristics, identifying and resolving sources of measurement bias.
Item Writing and Review
Development of test items that are technically accurate, clearly worded, and appropriately targeted to the relevant knowledge or skill domain, with systematic editorial and sensitivity review before field testing.
Standard Setting and Cut Scores
Structured studies using Angoff, Modified Angoff, Bookmark, and other procedures appropriate to the assessment purpose, documented to withstand legal and regulatory scrutiny.
Test Blueprint Development
Definition of the content domain, proportional representation of content areas, and the specification framework that guides test assembly.
Reliability Estimation
Internal consistency, test-retest, and inter-rater reliability analysis appropriate to the instrument and the decisions it supports.
Statistical Methods
Inferential and descriptive techniques applied to determine whether observed patterns are real, what predicts what, and how much confidence a finding supports.
Hypothesis Testing and Significance Analysis
Tests of whether observed effects or differences reflect true underlying patterns or random variation, including t-tests, z-tests, chi-square tests, and procedures tailored to data type and study design.
Multiple Regression and Correlation
Quantification of relationships between variables, control for confounding factors, and predictive models that explain outcomes in terms of measurable inputs.
ANOVA, MANOVA, and Factorial Designs
Comparison of means across multiple groups or conditions simultaneously, accounting for interaction effects when multiple factors are at play.
Factor Analysis and Principal Component Analysis
Reduction of complex, high-dimensional data into interpretable underlying constructs, revealing the latent structure in surveys, assessments, and behavioral datasets.
Structural Equation Modeling
Theoretically informed path models that capture both direct and indirect relationships among multiple variables.
Non-Parametric Methods
Distribution-free techniques applied when distributional assumptions cannot be met, preserving analytical rigor without requiring data transformations.
Power Analysis and Sample Size Determination
Calculation of the sample sizes needed to detect effects of a specified magnitude, ensuring that studies are adequately powered before data collection begins.
Adverse Impact Statistics
Selection rate comparisons, the four-fifths rule as a practical screen, and appropriate tests of statistical significance for evaluating differences in outcomes across groups.
Data Analysis
Techniques for finding structure in large or unstructured datasets, applied where the question is exploratory rather than confirmatory.
Clustering and Segmentation
Partitioning of datasets into meaningful groups based on similarity, revealing natural subpopulations within applicant, employee, or response data.
Classification and Decision Trees
Models that assign new records to predefined categories based on patterns learned from historical data.
Text Analysis of Open-Ended Responses
Structured insight drawn from unstructured text sources including open-ended survey items and written comments.
Anomaly and Outlier Detection
Identification of records that deviate significantly from expected patterns, relevant to data quality review and response validity screening.
Dimensionality Reduction
Compression of high-dimensional data into lower-dimensional representations that preserve the most important structure, improving interpretability.
Time Series and Trend Analysis
Analysis of data collected over time to identify trends, seasonal patterns, and cyclical behavior across repeated administrations.
Have a question these methods could answer?
Selecting the right approach depends on the question and the data available. Get in touch to discuss what you are working with.
Contact Us Today