The science behind our trait assessments
What it is
Our trait assessment feature provides high-quality candidate signals on job-related traits you define upfront, with optional support from our AI agent.
The feature leverages interview transcripts to analyze candidates’ language use and is heavily grounded in scientific theories from the fields of psychology and linguistics. These theories attribute language as a key behavior that contains powerful insights into our unique psychology - how we think, decide, and communicate.
Powered by the world’s most psychometrically validated language analysis method6, our feature uses a sophisticated analysis to generate numeric scores for each candidate on the job-specific traits you select. These scores are surfaced for you alongside practical insights and actionable take-aways for each candidate, all of which is intended for use in the next stage of your hiring process (typically, the first in-person interview).
This feature ultimately helps our customers quantify qualitative data and translate it to actionable insights. Backed by scientific intelligence and rooted in measurement science, it enhances each of your candidate profiles with more objective, data-driven insights you can trust to power your hiring decisions.
How traits have traditionally been measured & used in organizations
Companies have been interested in measuring traits - both in the pre-employment and employment contexts - for a long time. Traditionally, trait measurement has often been done with the use of psychometrically validated inventories. Most consist of multiple choice questionnaires administered either via self-reporting or having leaders/colleagues complete them on behalf of others (i.e., other-reporting).
At face value, these lists of questions generally seem like they measure the traits of interest. They might also seem like anyone could simply write, package, and sell them as an ‘inventory’. But, in reality, these inventories are heavily based on principles from the field of psychometrics, a specific application of assessment science.
Why is this important? Because although questions might seem like they’re measuring a trait, it doesn’t mean they actually are. The steps followed to psychometrically validate an inventory are guided by a long list of criteria, two of which include answering the following: 1) to what extent are the questions actually measuring the trait of interest (i.e., content validity), and 2) to what extent does this test explain future outcomes we would expect from the trait (i.e., predictive validity)? Once an inventory meets the criteria outlined in the process, it can be labelled as ‘psychometrically validated’ and then packaged and sold to companies.
While the actual validation process is extensive, what’s important for context is understanding one fundamental principle in trait measurement. This principle lays the foundation for why measuring traits - which are abstract and latent (i.e., not observable) - is even possible:
Behaviors are the unit of measurement we can use to make valid inferences about people when the things we want to measure are not directly observable (i.e., traits, versus a hard skill such as coding).
The good news is, we can observe and measure behavior in many different ways:
- We can ask people about their past behaviors to make inferences about their future behaviours,
- We can ask them about their intentions to behave in a certain way in the future
- We can observe how people behave in the present, either through directly observing actions or through the language they use
Many traditional assessments use #1 and #2 to measure behaviour.
Hirevue’s traits assessment feature uses #3.
Following this approach, behavior captured via any of the methods listed above goes through a measurement lens that allows us to convert observations into trait scores.
Hirevue’s trait assessment feature
Hirevue’s trait assessment feature builds on the core measurement principle described above: behaviour is the observable unit that allows us to infer traits. In our case, language is the behavior, and we sample that behavior via candidates’ language patterns in interviews.
What we can learn from language: Decades of research in psychology and linguistics demonstrate that language reliably reflects our unique individual psychology. One key finding is that both content words (what we talk about) and function words (how we structure language) reveal meaningful information about personality, motivations, and cognitive style7. Through systematically analyzing these word patterns, it becomes possible to infer traits that would otherwise require lengthy psychometric questionnaires.
Importantly, the trait frameworks we leverage in our tool were carefully selected to represent the most important job-relevant traits. Our selection is largely consistent with those commonly found in traditional psychometrics tests such as the Big 5 Personality Inventory, Drives and Motivations, and Teamwork style. In general, extensive research has focused on the use of language for explaining individual traits such as these and have cumulatively demonstrated the empirical validity of the model8-11.
The tech behind the feature: The model powering our trait assessment feature is also heavily grounded in scientific intelligence. It leverages the LIWC, one of the most widely validated approaches to psychological language analysis and has been cited in over 27,000 research papers. This model translates patterns in interview transcripts into structured measurements that align with established frameworks in both personality psychology and psychometrics. By combining advances in linguistics with the rigor of psychometric measurement, our trait assessment feature converts conversational interview data into scientifically grounded trait insights, bringing measurement science directly into your interview process.
The benefits of language-based trait assessments
Traditional trait assessments have played an important role in the past, but they come with practical limitations that are compounded in modern hiring environments.
Most conventional inventories rely on self-report questionnaires, which introduce several well-documented sources of bias. There are also operational challenges. Many assessments are expensive, difficult to administer at scale, and disruptive to the candidate experience. Timing is also a concern in the pre-employment context: lengthy assessments early in the process may discourage applicants, while administering them later can slow down decision-making.
Hirevue’s trait assessment addresses these challenges by embedding measurement directly into the interview process. Instead of asking candidates to complete an additional test, our system analyzes language already produced during interviews. This approach removes the need for separate questionnaires while reducing susceptibility to impression management and response bias.
For hiring teams, the result is a more efficient and objective signal early in the hiring process. Recruiters gain quantitative insights derived from real candidate behavior, which can help prioritize applicants, guide follow-up questions, and structure in-person interviews.
Best practices for incorporating trait assessment in the hiring process
Trait insights are most powerful when used as one component of a multi-measure hiring strategy. After screening for minimum qualifications and narrowing down your candidate pool to those who have the most job-relevant skills, trait assessments can provide additional high-signal information about which candidates are best fit for the role.
Combined with structured interviews, work samples, and skill evaluations, language-based trait scores contribute to a more multidimensional understanding of candidates. This helps organizations identify strengths that may not be obvious from resumes alone and predict how individuals are likely to perform in a given role or environment.
Ultimately, the goal is not to replace human judgment, but to augment it with scientifically grounded insights you can trust - helping teams make more consistent, data-informed hiring decisions while saving time and resources.
FAQs
Frequently asked questions
How does the AI know what the hiring bar is for a specific role?
The system starts by converting a job description into an I/O informed interview script that defines the questions and scoring standards for the role. Hiring teams can review and adjust this rubric, so the AI evaluates responses using the same criteria your team would use.
What is the Trust Score, and why does it matter?
Trust Score measures how closely the AI’s evaluations match the hiring team’s own scoring decisions. During calibration, the system compares human and AI evaluations and highlights disagreements. As alignment improves, teams gain confidence that the AI is applying their hiring standards consistently at scale.
How does the system avoid being a “black box” AI?
Every recommendation from the AI includes an explanation of how the candidate response was scored and references moments in the interview that informed the score. This allows reviewers to verify the reasoning behind each evaluation.
Do humans still make the final decision?
Yes, always. The AI Interviewer produces structured evidence and recommendations, but hiring teams stay in control of the decision.
What kind of evidence can recruiters see?
Reviewers can see how answers were scored against the role criteria, the rationale behind scores, and references to the candidate’s interview responses so decisions are easier to review and stand behind.
