Possible Clues:

• Sampling of voters after voting
• Way-out head tries to forecast voting result
• Way-out head tried to forecast voting result
• Election day benchmark
• Survey of voters just having voted
• Voting-site survey
• Aspect of psephology that's on the way out
• Pixel lot (anag.)
• Way-out method of making predictions?
• Mad exploit overshadowing learner's study of voting trends