Discredited instruments — Organizational Development

A catalog of popular organizational tools that the evidence does not support, collected in one place so that they can be recognized and declined.

These are not merely weakly supported, which describes much of this collection and is not disqualifying. These are instruments where the central claim has been tested and not found, and where using them for a consequential decision — hiring, promotion, team assignment, or training design — misallocates money and can be unfair to individuals.

Evidence: Very weak to refuted. Each entry below states its own status. The common pattern: an intuitively appealing typology, an appealing self-recognition experience, commercial incentives to sustain it, and either no predictive validity or a directly failed test of the mechanism it claims.

Use when: someone proposes one of these, and you need the specific objection rather than a general sense of unease.

Do not use when: as a licence to dismiss all assessment. Several instruments in this repository are well validated — see psychological-safety, westrum-organizational-culture-typology, and job-characteristics-model. Scepticism is a tool, not a personality.

Learning styles

Claim. People learn better when instruction matches their preferred style — visual, auditory, kinaesthetic, or one of some seventy other published schemes.

Status: refuted. The claim requires a meshing interaction: people labelled visual should outperform when taught visually, and underperform when taught auditorily. Studies designed to test that interaction — rather than merely asking people their preference — consistently fail to find it. Reviews have repeatedly concluded there is no adequate evidence base for using learning styles in instruction.

What is true underneath. People do have preferences, and preferences are real. Content also has an inherent best modality: teach geography with maps and music with sound, regardless of who is in the room. Prior knowledge predicts learning far better than style ever did.

What to do instead. Design for the material. Use spacing, retrieval practice, interleaving, and worked examples — all well supported.

Myers-Briggs Type Indicator

Claim. People fall into sixteen types, and the type predicts suitability for roles, teams, and communication approaches.

Status: weak, and inappropriate for consequential decisions. Three specific problems. Test–retest reliability is poor; a substantial proportion of people get a different type on retaking within weeks. The underlying traits are continuously distributed, so dichotomizing at the mean assigns two nearly identical people to opposite types. And there is little evidence of predictive validity for job performance.

What is true underneath. Some of the dimensions correlate with better- supported personality traits, particularly extraversion. The Big Five model has far better psychometric properties.

What to do instead. If you need personality assessment for selection, use a validated Big Five instrument, and know that structured interviews and work sample tests predict job performance better than personality does. If you want a team conversation about difference, say that is what it is, do not use the result for any decision, and pick something cheaper. See https://github.com/joelparkerhenderson/big-five-personality-traits.

DISC and colour-based type systems

Claim. Four types — variously dominance/influence/steadiness/conscientious- ness, or red/blue/green/yellow — describe behavioral style and predict fit.

Status: weak. The same dichotomization problem as MBTI, generally with less published psychometric work behind it. Commercial variants proliferate without independent validation. Predictive validity for job performance is not established.

What to do instead. As above. If the goal is a shared vocabulary for discussing working preferences, any framework will do that, and none of them should touch a hiring decision.

Generational categories

Claim. Baby Boomers, Generation X, Millennials, and Generation Z have distinct workplace values requiring different management.

Status: weak. Studies attempting to isolate generation from age and period effects find small and inconsistent differences, typically far smaller than within-generation variation. Much of what is attributed to generation is life stage — a 25-year-old in any decade wants different things from a 55-year-old — or the economic conditions of the moment.

What to do instead. Ask people what they want. Segment by role, life stage, and caring responsibility, which are observable and actionable.

Others, briefly

Graphology — handwriting analysis for selection. No predictive validity; still occasionally used in some markets.
Neuro-linguistic programming — the specific claims about eye movements and representational systems have not been supported.
Left-brain versus right-brain personality — hemispheric specialization is real for some functions; the personality typology built on it is not.
Birth order effects on personality — large modern studies find effects at or near zero.

Why they persist

Understanding the mechanism helps you argue against them without insulting the person who proposed them, who is usually acting in good faith.

The Barnum effect. Descriptions vague enough to fit anyone produce strong recognition. "You are outgoing but value time alone" fits nearly everybody.
Non-judgmental framing. Every type is presented as equally good, so nobody is offended and everyone participates. This is genuinely pleasant and is not evidence.
Commercial reinforcement. Certification, licensing, and consultancy revenue create a supply of confident advocates.
Real experience, wrong explanation. The workshop is useful — people talk about how they work together, often for the first time. The value comes from the conversation, and it gets attributed to the instrument.

How to decline one

Ask what decision the result will inform. If the answer is a hiring, promotion, or assignment decision, that is the strongest objection and the easiest to make.
Ask what evidence would show it was not working. Instruments in this list rarely have an answer.
Offer the substitute rather than only the refusal: a structured interview, a work sample, a validated instrument, or simply a facilitated conversation with no typology attached.
Do not attack the colleague who proposed it. They had a good workshop once. That experience was real; the explanation was wrong.

Examples by audience

Health care: personality typing is common in team-building days. It becomes a patient safety issue only if it is used to justify who may challenge whom — "that is just how reds communicate" is a way of excusing behavior that a just-culture approach would examine.
Software: learning styles appear in internal training design, where they consume budget that spacing and retrieval practice would spend better. Personality typing appears in hiring, where structured interviews and work samples predict performance better and are more defensible if challenged.
Executive: the most expensive version is a leadership development programme built on a typology, priced per participant, with no outcome measure. Ask for the theory of change and the evidence before approving it. See theory-of-change and evidence-based-practice.

Limitations

This list is not exhaustive, and new instruments appear continuously.
"Discredited" applies to the specific claim tested. Some of these tools have benign uses as conversation starters, and this file should not be cited to suggest otherwise.
Evidence changes. If a claim here has been supported by newer research, that is a defect worth reporting — see ../../CONTRIBUTING.md.