Being below the national average in MHI doesn't necessarily make a county poor, especially if the cost of living in those places is also lower or households are smaller than the national average. The real question is how many of those counties had a higher percentage of folks below the poverty line. I haven't looked at data for all 14 counties, but the census bureau's map shows most Vermont counties are around average in that category.
You're rejecting MHI/PCI in favor of poverty line as a measure? Poverty line is calculated pretty ineffectively, IMHO. PCI is my favorite measure, in areas of comparable family/individual population. Poverty rates oftentimes can be hugely, and as Al says, this is one such instance.
Read a paper on this...remind me to drag it up?