We see moderately strong, negative correlations between obesity percentage, smoking percentage, percentage lacking healthcare and median household income. As median household income increases, the smoking percentage, obesity percentage, and percentage lacking healthcare all decrease.
We see moderate, positive correlations between obesity percentage, smoking percentage, percentag lacking healthcare and the percentage under the poverty threshold. As the percentage under the poverty threshold increases, the obesity percentage, smoking percentage, and percentage lacking healthcare increase.
Finally, there is a moderate, negative correlation between age and the percentage lacking healthcare. As age increases, the percentage lacking healthcare decreases.
The data given are statewide summaries and not measurements on the individual level. Correlations computed using aggregate data such as group means are sometimes referred to as ecological correlations; they are often stronger than correlations that would be computed at the individual level. In this case, we would expect data on income, poverty, age, obesity, etc., to have a large scatter around the averages resulting in smaller correlations at the individual level. This does not discount the associations between income, age and health risks.
Spearmans' rank correlation coefficient was computed using the JavaScript module jStat: The JavaScript Statistical Library