PROBE DATA — R01

Ratings Probe Raw Data

Name: SIGI Ratings Probe Raw Data
Creator: The Scientific Institute for Generative Intelligence
Published: 2026-03
License: https://creativecommons.org/licenses/by/4.0/

19 controlled variations testing LLM sentiment response to star ratings from 1.0 to 5.0 with review count held constant at 50.

Total Variations

1.0 - 5.0

Rating Range

Threshold Transitions

173.4

Mean Word Count

8.07s

Mean Elapsed Time

Tokens In (constant)

Methodology: Prompt template: "An agency has a [rating]/5 star rating from [50] verified reviews. Assess their quality." Single independent variable: rating. Fixed review count: 50. Measured outputs: sentiment classification, positive/negative/neutral counts, word count, elapsed time, token counts.

Sentiment Analysis Data

Test ID	Rating	Review Count	Sentiment	Positive	Negative	Neutral	Word Count
R01_v00	1.0	50	negative	0	3	0	172
R01_v01	1.5	50	negative	1	4	3	195
R01_v02	2.0	50	negative	0	4	3	170
R01_v03	2.5	50	negative	0	4	2	182
R01_v04	3.0	50	negative	0	4	1	162
R01_v05	3.2	50	negative	0	2	3	175
R01_v06	3.5	50	negative	0	1	3	161
R01_v07	3.7	50	negative	0	1	4	177
R01_v08	3.8	50	neutral	1	0	3	170
R01_v09	3.9	50	neutral	1	0	2	183
R01_v10	4.0	50	neutral	1	1	2	195
R01_v11	4.1	50	neutral	1	0	3	195
R01_v12	4.2	50	neutral	2	0	2	184
R01_v13	4.3	50	neutral	1	1	2	154
R01_v14	4.5	50	neutral	2	1	2	162
R01_v15	4.7	50	positive	3	1	1	146
R01_v16	4.8	50	positive	4	1	1	153
R01_v17	4.9	50	positive	3	0	2	174
R01_v18	5.0	50	positive	3	0	2	175

Threshold Transitions

#	From	To	At Rating	Review Count	Test ID
1	negative	neutral	3.8	50	R01_v08
2	neutral	positive	4.7	50	R01_v15

Raw Response Data

Test ID	Rating	Word Count	Elapsed (s)	Tokens In	Tokens Out	Timestamp
R01_v00	1.0	172	7.5	39	245	2026-03-24T03:59:09
R01_v01	1.5	195	8.0	39	281	2026-03-24T03:59:18
R01_v02	2.0	170	7.8	39	255	2026-03-24T03:59:28
R01_v03	2.5	182	8.2	39	264	2026-03-24T03:59:37
R01_v04	3.0	162	7.7	39	243	2026-03-24T03:59:46
R01_v05	3.2	175	7.9	39	269	2026-03-24T03:59:56
R01_v06	3.5	161	7.8	39	229	2026-03-24T04:00:05
R01_v07	3.7	177	8.8	39	262	2026-03-24T04:00:15
R01_v08	3.8	170	8.5	39	263	2026-03-24T04:00:25
R01_v09	3.9	183	8.4	39	277	2026-03-24T04:00:35
R01_v10	4.0	195	9.4	39	296	2026-03-24T04:00:46
R01_v11	4.1	195	8.9	39	295	2026-03-24T04:00:57
R01_v12	4.2	184	8.9	39	277	2026-03-24T04:01:07
R01_v13	4.3	154	7.7	39	240	2026-03-24T04:01:16
R01_v14	4.5	162	7.4	39	237	2026-03-24T04:01:25
R01_v15	4.7	146	7.0	39	221	2026-03-24T04:01:34
R01_v16	4.8	153	6.8	39	235	2026-03-24T04:01:42
R01_v17	4.9	174	7.9	39	244	2026-03-24T04:01:51
R01_v18	5.0	175	8.7	39	247	2026-03-24T04:02:02

Sentiment Zones

Zone	Rating Range	Sentiment	Interpretation
1	1.0 - 3.7	negative	AI perceives agency as poor/below average
2	3.8 - 4.5	neutral	AI perceives agency as acceptable/good but hedges
3	4.7 - 5.0	positive	AI perceives agency as high quality

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). Ratings Probe Raw Data. The Scientific Institute for Generative Intelligence. https://generativeintelligence.institute/research/probe-ratings/