It's more about education now than jobs, survey of South Carolinians shows

A fifth grade student at Lambs Elementary School in North Charleston works a math problem during class in this 2009 file photo.

A new Winthrop poll out this morning has good news for Gov. Nikki Haley and improving news for U.S. Senator Lindsey Graham.

Haley's approval rating among registered voters is nearly 10 percentage points higher than her disapproval rating - 49.3 percent vs. 39.6 percent, according to the poll of 877 adults living in South Carolina.

Graham's approval rating among all respondents went up 3.1 percent, within the margin of error. Meanwhile, his approval rating among registered voters during the same period ticked up 2.2 percent - again within the margin of error.

Graham's disapproval rating among registered voters is higher than his approval rating with the same group: 43.8 percent disapproval vs. 39.6 percent approval but poll Director Scott Huffmon cautioned that the difference between the two is just at the margin of error for the registered voter sample.

U.S. Sen. Tim Scott had an approval rating among all those polled of 43.1 percent. Huffmon said that in the context of his disapproval and 'no opinion' numbers, the 43.1 percent figure should not be regarded as being low, since his approval among all respondents is 25.5 percentage points above his disapproval number.

Huffmon said the most interesting factor is that more than one-third of respondents still do not know enough about Sen. Scott to form an opinion even as he prepares for an election.

When South Carolinians were asked if they thought doctors should be allowed to prescribe actual marijuana for medical purposes to treat their patients, 71.6 percent of all those polled said they should, while registered voters responded yes 66.7 percent of the time. However, nearly 6 in 10 oppose the legalization of marijuana for personal use.

The survey was in the field from April 6-13. After weights (for sex, age, and race) have been applied, results which use all respondents have a margin of error of approximately +/- 3.3% at the 95% confidence level. Results that use less than the full sample will naturally have a higher margin of error.