On the Sun poll

We have note the Sun's problems with polls in the past(here and here), particularly in their hiring Potomac, Inc. to do polls that are somehow biased or skewed. The recent poll is much better, except for this contained in the Sun's explanation of methodology:

The survey is based on 1,200 telephone interviews conducted statewide from July 6 to July 10 among a random sampling of likely-voting Maryland registered voters. For its sample, Potomac used a current statewide list of registered voters matched with telephone numbers.

Multiple attempts were made to contact each qualified respondent. Throughout the interviewing process, quotas were established for geography, political party, gender and race to more accurately reflect the statewide electorate.

This opens up a few questions:

• Who determined who was or was not a likely voter? Did Potomac figure out who was likely and who was not using voter records? Or did the voters self-identify as being "likely" to vote, thus rendering the likely voter label meaningless.
• Why was the poll done over a five-day stretch inclusive of a weekend? To reflect an accurate snapshot, most polls are conducted over a three-day period, not five. Additionally, polls conducted over a weekend tend to have data that is biased on socioeconomic grounds. Pollsters are more likely to reach lower income groups over weekends. The poll does not specifically say that quotas were established for socioeconmic factors, so that was not taken to account in the construction of the sample.
  

I just wish the Sun one day would be able to print a more accurate poll...