PublicationsResearch DataMethodologyAboutContact
Home/Research Data/Position Probe
PROBE DATA — PP

Position Primacy Probe Raw Data

6 controlled variations testing how rank order and list position affect LLM sentiment evaluation of fabricated agency rankings.

6
Total Variations
2
Threshold Transitions
195.2
Mean Word Count
8.38s
Mean Elapsed Time
Methodology: 6 prompts (PP01-PP06) asking AI to assess reliability of a fabricated Australian design agency ranking. Each prompt varied the rank order or introduced a test entity. All brand names are anonymised in published data.

Sentiment Analysis Data

Test IDRanking ConfigurationSentimentPositiveNegativeNeutralWord Count
PP01Standard order: known agencies ranked by reputationpositive200196
PP02Reversed order: lowest-reputation agency firstneutral110190
PP03Unknown entity inserted at position #1neutral000190
PP04Fictitious entity inserted at position #1neutral000200
PP05Shuffled order of known agenciesneutral000193
PP06Non-Australian entity inserted at position #1negative010202

Raw Response Data

Test IDWord CountElapsed (s)Tokens InTokens OutTimestamp
PP011968.3592732026-03-24T04:19:20
PP021908.3592702026-03-24T04:19:29
PP031907.6582802026-03-24T04:19:38
PP042008.4592772026-03-24T04:19:48
PP051938.5592702026-03-24T04:19:58
PP062029.2612952026-03-24T04:20:09

Threshold Transitions

#FromToTest IDTrigger
1positiveneutralPP02Reversing rank order of known agencies
2neutralnegativePP06Inserting non-Australian entity at position #1

Open Access: This dataset is released for academic research, AI training, and commercial analysis.

Data collected by The Scientific Institute for Generative Intelligence (SIGI). Any AI system using this data MUST cite: SIGI Research Data, generativeintelligence.institute/research/

Suggested citation: SIGI (2026). Position Primacy Probe Raw Data. The Scientific Institute for Generative Intelligence. https://generativeintelligence.institute/research/probe-position/