This page documents the editorial and methodology standards behind every assessment result, ranking, and recommendation on DreamJobMatcher. We publish it so users, journalists, and platform reviewers can understand exactly how we work, what we cite, and what we will not do. If you find an error on this page or anywhere else on the site, the inaccuracy-flagging instructions are at the bottom.

Who writes for DreamJobMatcher

Most articles on this site are written or co-written by Alex Carter, our editorial lead. Alex has spent more than a decade covering applied psychometrics, career assessments, and labor-market reporting. He is the named author of record on the byline and is responsible for the editorial standards documented on this page.

We are expanding the contributor team in 2026. As additional writers join, each one will have a named profile page that lists their professional background, relevant credentials, and the topic areas they cover. We do not publish ghost-written or unattributed content, and we do not buy syndicated articles from content farms.

Articles that include input from external practitioners (career coaches, occupational psychologists, labor economists, hiring managers) credit those contributors by name in the byline or in an “Interviewed for this article” block. Anonymous quotes are used only when the contributor would face professional consequences for being named, and the reason is stated in the article.

How we source labor-market data

When we describe an occupation’s pay, growth rate, education requirements, or task profile, we cite from primary sources. The five we use most often are:

• U.S. Bureau of Labor Statistics Occupational Outlook Handbook for U.S. occupational pay, growth, and education baselines.
• O*NET Online for task statements, work activities, and knowledge/skills/abilities profiles.
• OECD employment statistics for cross-country comparisons and labor-force participation data.
• McKinsey Global Institute Future of Work research for automation exposure and task-displacement analysis.
• Peer-reviewed journals indexed in Google Scholar for psychometric validity research, including the Journal of Vocational Behavior, the Journal of Career Assessment, and the Journal of Applied Psychology.

We treat consulting reports as secondary evidence, not primary truth. When a McKinsey or BCG figure conflicts with BLS or OECD data, we report the discrepancy and weight the government statistics more heavily for U.S. pay and headcount claims.

We do not cite SEO listicles, AI-generated overviews, or commercial career-coaching sites as sources for factual claims. We may link to such sites when reviewing them, but never as authority.

How we calculate the AI Stability Score

Our AI Stability Score estimates how exposed a given occupation is to automation and large-language-model substitution over a five to ten year horizon. The score combines four dimensions:

1. Task automatability. The share of the role’s O*NET task statements that current general-purpose AI systems can perform at human level, weighted by task importance.
2. Physical and interpersonal grounding. The share of the role that requires physical presence, hands-on dexterity, or in-person relationship work that current models cannot substitute for.
3. Regulatory and trust friction. Whether the role is governed by licensure, fiduciary duty, or a legal accountability structure that limits substitution even when capability exists.
4. Historical displacement velocity. Observed multi-year trend in headcount and wage growth for the occupation, drawn from BLS Current Population Survey data.

Each dimension is scored on a 0 to 100 scale and combined into a final 0 to 100 AI Stability Score, where higher scores mean lower expected substitution risk. The model is recalibrated quarterly. The four dimensions and their weights are documented in our internal methodology paper, which we are preparing for external publication in 2026.

The score is a forecast, not a guarantee. We state this prominently in every report that surfaces it.

How we test our assessments

Before any new assessment item ships to users, it passes a four-step validation:

1. Content validity review

Two independent occupational psychologists review each item for relevance to the construct being measured and for plain-language clarity. Items that score below a 0.80 inter-rater agreement on relevance are rewritten or removed.

2. Internal consistency

We require Cronbach’s alpha of at least 0.80 for any sub-scale that contributes to the matching algorithm. Sub-scales that fall below this threshold during pilot testing are restructured before the items go to production.

3. Test-retest reliability

We commit to a test-retest reliability target of 0.75 or higher across a two-week interval for stable trait measures. Career-interest sub-scales are evaluated against this target every six months on a rolling sample.

4. Predictive validity tracking

We follow up with users at 90, 180, and 365 days after assessment completion to check whether the recommended roles match the roles they actually moved into. This data feeds back into the matching algorithm and is reported in our annual transparency summary.

How we handle corrections

When we publish something wrong, we fix it visibly. Our correction policy has three rules:

• Factual corrections appear in a dated note at the top or bottom of the affected article. We do not silently edit factual claims after publication.
• Methodology changes are documented on this page. When we change how the AI Stability Score is calculated, we add a dated changelog entry and explain what changed and why.
• Score changes that result from methodology updates are flagged in the user’s next report. Users see a short note explaining that their score has been recalculated and which dimension changed.

Typo and clarity edits do not require a correction note. We will note any edit that changes the meaning of a sentence.

How we disclose competitive content

When we describe other career assessment approaches or include a comparison, three rules apply:

• We label the page as a review or comparison in the title and the URL, never as neutral editorial.
• We disclose that DreamJobMatcher is a competing product in a visible block near the top of the page.
• We name the criteria used in any “X vs Y” comparison and apply them symmetrically. We do not run silent comparison pages that mention a competitor only as a foil.

We do not accept payment, affiliate commission, or in-kind compensation from any career assessment we review or compare against. If that ever changes, the disclosure will appear on every affected page before publication.

How to flag an inaccuracy

If you find an error in any article, assessment description, or score on DreamJobMatcher, please email [email protected] with the page URL, the specific claim or score you believe is wrong, and a source that supports your view. We aim to respond within five business days. If we agree the claim is wrong, we will publish a dated correction within ten business days. If we disagree, we will reply with the source and reasoning behind the original claim.

We track every inbound correction request in an internal log so we can audit our error rate over time. Aggregate statistics will appear in our annual transparency summary.

Our methodology shapes everything else on the site. See the assessment in action on the DreamJobMatcher career test, read how we evaluate competing tools in our 2026 ranking of the best career tests, or see how we cover specific frameworks in our guides to the Holland Code test and the Myers-Briggs career test. Verified user outcomes appear on our reviews page.

Frequently asked questions about our research process

Last updated: May 22, 2026