Perceived accuracy of automated risk assessment tools
Full Question Text
In some places, automated tools make assessments about public safety risks that in the past were made by people. For example, keyword tracking software may be used to monitor social media for events where a crime might happen. In your opinion, are the assessments made by these tools generally more or less accurate than those made by humans?
| Total | |
| More accurate | 9% |
| About the same | 16% |
| Less accurate | 20% |
| Don't know | 55% |
| County | Township | City | Village | Total | |
| More accurate | 11% | 7% | 14% | 10% | 9% |
| About the same | 18% | 14% | 16% | 19% | 16% |
| Less accurate | 15% | 23% | 12% | 18% | 20% |
| Don't know | 57% | 55% | 57% | 53% | 55% |
| <1500 | 1500-5000 | 5001-10000 | 10001-30000 | >30000 | Total | |
| More accurate | 8% | 9% | 5% | 14% | 17% | 9% |
| About the same | 17% | 14% | 19% | 12% | 17% | 16% |
| Less accurate | 24% | 19% | 19% | 15% | 12% | 20% |
| Don't know | 51% | 59% | 56% | 59% | 53% | 55% |
| Upper Peninsula | Northern Lower Peninsula | West Central Lower Peninsula | East Central Lower Peninsula | Southwest Lower Peninsula | Southeast Lower Peninsula | Total | |
| More accurate | 9% | 7% | 9% | 7% | 10% | 12% | 9% |
| About the same | 12% | 20% | 15% | 15% | 13% | 18% | 16% |
| Less accurate | 23% | 20% | 19% | 22% | 18% | 19% | 20% |
| Don't know | 56% | 53% | 57% | 57% | 58% | 51% | 55% |
| Rural | Mostly rural | Mostly urban | Urban | Don't Know | Total | |
| More accurate | 7% | 11% | 12% | 19% | 0% | 9% |
| About the same | 14% | 18% | 19% | 12% | 6% | 16% |
| Less accurate | 23% | 20% | 13% | 10% | 11% | 20% |
| Don't know | 55% | 51% | 56% | 59% | 83% | 55% |