Skip to content

Methodology & science

Same assessment. Different format.

The construct, items and dimensions of an assessment stay the same with Kolloqo. What is new is the format: instead of ticking items, participants describe their experiences in their own words – and the AI traces their answers back to the same items.

  1. 01Construct
  2. 02Items
  3. 03Pilot
  4. 04Validation
  5. 05Norming
  6. 06Data collectionNew: in conversation

Test development – unchanged

Who is behind it

The assessments come from IR&C’s portfolio.

The content behind Kolloqo comes from psychologists who develop assessments for professional use.

Under the direction of Dr. Jörg Prieler, IR&C develops psychological assessments for professional use. IR&C’s portfolio comprises over 25 assessments – from workplace health (HOT, BODI) and leadership diagnostics (TUF) to security-critical applications (Triple Eye, ITR).

Interviews based on a validated assessment use its content, dimensions and scoring logic from IR&C’s portfolio. Scientific responsibility stays with the assessment developers – the technology offers a format that is closer to how people naturally talk than a questionnaire.

All assessments →

over 25

validated assessments in IR&C’s portfolio

Dr. Jörg Prieler’s core expertise is IRT models and test construction; he has developed numerous performance and personality assessments.

  1. 2000Dissertation: a new mathematical model for measuring change and training effects.
  2. 2002Talk and chair, International Conference on Computer-Based Testing and the Internet, Winchester (UK).
  3. 2003With John Raven: British Journal of Clinical Psychology.
All publications
  1. 2007With John Raven: The Psychologist.
  2. 2009With John Raven: International Journal of Educational and Psychological Assessment.

Questionnaire and conversation

The same question – once ticked, once told.

The AI interview delivers two levels: scores on the assessment’s scale – each with its reasoning and confidence. And the context that participants share of their own accord in conversation.

Conversation
Kolloqo asks

How do you approach new tasks?

Ideally with a plan. I think the steps through first, then get going.

Follow-up

And when the plan stalls?

Since the reorganisation, it’s often unclear who takes the next step.

Illustrative example · score and context from the same interview

Evaluation

Step by step, an answer becomes a score on the scale.

No black box. Every score can be traced back to specific passages of text – transparent for experts and compliance.

Illustrative example · simplified

  1. Step 1 of 5

    1. Break down the answer

    The free-text answer is broken down into sections. The original text is kept.

  2. Step 2 of 5

    2. Assign dimensions

    Each section is assigned to the assessment’s dimensions. Anything unrelated to the question is left out of the scoring.

  3. Step 3 of 5

    3. Spot cues in the text

    How certain, how often, since when? Cues like these feed into the interpretation.

  4. Step 4 of 5

    4. A score on the assessment’s scale

    The answer is traced back to the same scale as in the questionnaire.

  5. Step 5 of 5

    5. Confidence and reasoning

    Every score comes with a confidence level and the passage it is based on. If the evidence is not sufficient, it says “not assessable”.

Report

What you have in hand at the end.

With a validated assessment

For questions covered by an assessment from IR&C’s portfolio.

You receive

  • The complete classic evaluation, as specified in the manual
  • The report from the conversation
  • Depending on the assessment, e.g. a traffic-light system, a percentile profile or an anonymous group report

With your own question set

For new questions without a validated questionnaire.

You receive

  • The report from the conversation: summary, score per statement, reasoning and confidence
  • The transcript – whether and in what form is agreed per project

Always included

  • Score per statement
  • Reasoning with the text passage
  • Confidence
  • Classification by experts
Example: security report →Get your own report: try the AI interview →

Evidence

Measured, not claimed.

Kolloqo doesn’t replace scientific validation – it builds on it. The quality of the results is reviewed systematically.

4.31out of 5

Participants’ satisfaction with the conversation11 Source: Pilot 2026, participant ratings

r =0.83

Range 0.64–0.96

Agreement between conversation and questionnaire of the same assessment22 Source: BODI comparative study, n = 85, analysed by Dr Benesch

From the pilot

People tell you more than you ask.

Finding from the pilot

In the AI interview, participants shared considerably more than in the classic questionnaire – specific situations, suggestions for improvement and problems that no item covers.

IR&CWorkplace health pilot study, January 2026

Questions about the methodology?

We’re happy to explain in detail how the evaluation works, which validation results are available and what quality assurance looks like for your use case.

Book a call