Michigan voters under 45 are more likely to back El-Sayed, 63% versus 43% among older voters.

446 likely voters · margin of error ±9.3 points

Exploratory subgroup comparison selected after examining the October polls; pointwise intervals are not adjusted for that search. Age associations are cross-sectional, not individual persuasion or a forecast. Missing ages excluded. Likely voters in the named October state and district polls; not a national sample or forecast. Pooled figures are respondent-proportional descriptive summaries of separate electorates. Each poll’s subgroup weights are normalized to its answered-question n; repeated people remain in record n but their weights are collapsed for Kish effective n. MOEs are approximate pointwise 95% half-widths in percentage points, without adjustment for exploratory multiple comparisons. They do not include all coverage, nonresponse, recall, calibration or selection uncertainty. Wald intervals can extend outside percentage bounds; an observed zero is not a population zero. Unknown and indifferent answers remain in denominators. Fewer than 50 distinct people is suppressed; 50–99 means interpret with caution.

62.9% of likely voters under 45 in MI support Abdul El-Sayed, Democrat (95% margin of error ±9.3 points; n=446, 446 distinct people, effective n=103.9). 27.8% of likely voters under 45 in MI support Mike Rogers, Republican (95% margin of error ±8.6 points; n=446, 446 distinct people, effective n=103.9). 42.7% of likely voters 45 and older in MI support Abdul El-Sayed, Democrat (95% margin of error ±5.8 points; n=2,186, 2,186 distinct people, effective n=275.7). 50.5% of likely voters 45 and older in MI support Mike Rogers, Republican (95% margin of error ±5.9 points; n=2,186, 2,186 distinct people, effective n=275.7). The featured estimate uses 446 responses from 446 distinct people, with an effective sample size of 103.9. Exploratory subgroup comparison selected after examining the October polls; pointwise intervals are not adjusted for that search. Age associations are cross-sectional, not individual persuasion or a forecast. Missing ages excluded. Likely voters in the named October state and district polls; not a national sample or forecast. Pooled figures are respondent-proportional descriptive summaries of separate electorates. Each poll’s subgroup weights are normalized to its answered-question n; repeated people remain in record n but their weights are collapsed for Kish effective n. MOEs are approximate pointwise 95% half-widths in percentage points, without adjustment for exploratory multiple comparisons. They do not include all coverage, nonresponse, recall, calibration or selection uncertainty. Wald intervals can extend outside percentage bounds; an observed zero is not a population zero. Unknown and indifferent answers remain in denominators. Fewer than 50 distinct people is suppressed; 50–99 means interpret with caution.