Sunday, May 16, 2004
Network Projection
About the 2004 election night TV coverage in Taiwan, according to the latest issue of the Con-temporary magazine, every news channels except San-li engaged in fabricating the vote-counting process after the polls were closed.
Though shockingly sadden by this shameless and unethical manipulation by the already notorious TV media in Taiwan, here, I want to offer my observation of projecting the winner during the honest (that was my, and everyone’s I think, hope), gavel-to-gavel coverage of the election night.
Given the final result is so close (about 0.23% percentage points difference), if the official result from the (official) Central Election Committee already included 50% of the all polling places’ results, the margin of the error for projecting each candidate’s final standing will be
2 * sqrt (0.5 * 0.5 / 6,000,000) = 0.04%
Therefore, though the razor-thin margin, it’s still prudent to predict that Mr. Chen Shui-bian as the projected winner half way through the vote-counting process. The margin of the error in sampling, 0.04%, means, in 19 of 20 cases the percentage point difference between the two camps during only 50% votes were already counted will not exceed 0.08%. So the margin of 0.23% is large enough to be reliable.
Though shockingly sadden by this shameless and unethical manipulation by the already notorious TV media in Taiwan, here, I want to offer my observation of projecting the winner during the honest (that was my, and everyone’s I think, hope), gavel-to-gavel coverage of the election night.
Given the final result is so close (about 0.23% percentage points difference), if the official result from the (official) Central Election Committee already included 50% of the all polling places’ results, the margin of the error for projecting each candidate’s final standing will be
2 * sqrt (0.5 * 0.5 / 6,000,000) = 0.04%
Therefore, though the razor-thin margin, it’s still prudent to predict that Mr. Chen Shui-bian as the projected winner half way through the vote-counting process. The margin of the error in sampling, 0.04%, means, in 19 of 20 cases the percentage point difference between the two camps during only 50% votes were already counted will not exceed 0.08%. So the margin of 0.23% is large enough to be reliable.
# posted by Laurent Lin : 11:06 PM
