```json { "table_name": "polling_on_ai_usage_jul_2026", "table_type_tags": [ {"tag": "dimension", "confidence": 0.90} ], "primary_type": "dimension", "insufficient_evidence": false, "entity": { "columns": ["question_number", "response_verbatim"], "unique_per_row": false }, "time_axis": null, "reasoning": "The composite entity (question_number, response_verbatim) identifies what each row is about: a specific survey response option for a specific question. question_number has 15 unique values (RPUV 7.00), response_verbatim has 65 unique values (RPUV 1.62), and their product (975) far exceeds the 105 rows, indicating the composite grain is unique per row. Each row holds demographic breakdowns (percentages across gender, age, income, etc.) as attributes of that question-response combination. The _ingestion_timestamp is a provenance column (ignored per rules). No time axis orders actions or changes — this is a cross-tabulated survey result table where each row describes one response option's distribution across demographic segments." } ```