Obama “Declared” Elected! - Measuring Voting Intentions in Social Media
When I watch as-it-happens election coverage, I am always impressed to see the statisticians declare a candidate elected with only a small fraction of the votes counted. With a few polls reporting and, in some cases, 1% of the vote counted, they have enough data to confidently declare the winner elected. Of course, it depends on the vote spread; close races are harder to declare. If a candidate is leading AND their relative share of votes is measured to be increasing or “pulling away” from their opponent, then there is strong trend data to draw statistically valid conclusions.
Based on my super-scientific statistical analysis below (not really), I am declaring Obama elected.
Yes, most polls have been showing Obama in the lead, but there is still a lot of discussion about the race being close. A Washington Post article, posted on Wednesday, discusses Accuracy Of Polls a Question In Itself, and points to an “undercurrent of worry among some polling professionals and acedemics” as a result of the wide variations in results. A TMP Election Central post last night says, “New Polls Show Tight Race In Key Swing States“. A Wired article argues that provisional ballots could decide election this year. CNN questions whether or not the polling results will suffer from the Bradley Effect. Finally, the Washington Times, this morning, states that Sen. John McCain “vaulted up in the latest nation and battelground state polls, closing to within the margin of error nationwide against Sen. Barack Obama…” in this article: McCain reduces Obama Advantage.
Ok, so the polls are always being questioned… but what about the social web? There are currently between 40 and 50 thousand posts, articles, videos, tweets, etc. posted each day about the presidential race or one of the candidates. Can we determine the next president with confidence and declare a winner?
Disclosure: ok, this next section is just for fun. I’m not a statistician or a political analyst.
For fun, I decided to pick a single expression of voting intent and then measure the results and look for trends, if any. I wondered, in all the conversations taking place online, how many people explicitly declared their voting intent by saying “I am voting for Obama” (or Barack Obama) versus “I am voting for McCain” (or John McCain). Yes, there are plenty of additional ways to articulate voting intentions, but I am mostly interested in the % split, than the overall numbers. For interest, I also added the phrase “I am undecided” in the context of a discussion around the political race.
Here is what I found for results in the past 30 days:
This isn’t a poll result. This is a measure of unsolicited expressions of voting intent on the social web (blogs, twitter, etc.). Clearly, this favors Obama with 67.6% of the declarations versus 28.2% for McCain.
So now let’s see if there is a trend and compare these recent (October Results) to those in August and September (see below).
There is a significant change from September to October with Obama’s share increasing by a whopping 13.2% month-over-month and McCain’s share seeing a -10.7% decrease. The trend is also evident from August to September with Obama support increasing 4.1% month-over-month while McCain support declined 2.0%. What is most significant to me is that there is a trend and the rate of change has increased significantly in recent weeks showing that voting intentions for Obama are increasing faster than voting intentions for McCain. It is also interesting to see that relative expressions of “I am undecided” are slowly decrementing which is what we would expect to see as the election date approaches.
Just to be safe, I thought I would check one more thing in the spirit of Net Promoter Score. Companies measure this because promoters or advocates are viewed as valuable influencers for a brand. The score is determine by asking the single question, “How likely is it that you would recommend us to a friend or colleague?”. Instead of asking this question, however, I wanted to measure how many people ACTUALLY recommended a particular candidate to someone else. So I measured how many people advocated, “You should vote for Obama” (or Barack Obama) versus “You should vote for McCain” (or John McCain). The results were consistent with Obama showing 60.1% share of the promoters versus 39.9% for McCain as shown in the chart below.
Can you shoot holes in my theory? Absolutely. It is unscientific. Still, I am declaring Obama elected. What do you think? Oh, last thing (also for fun)… I thought I would also measure how many other people uttered, “Obama declared elected”. I’m the first!
October 31st, 2008 - Posted in Social Media | |
on October 31st, 2008 at 2:52 pm
You discount all of the people who are voting for either @VoteDoom or @Richards08!
Just kidding.
Really interesting and fun post. My only worry is how many of these people won’t or can’t vote.
on November 2nd, 2008 at 7:09 pm
opinion polling2.0 ?…
2 days to go and Barack Obama stays ahead of John McCain in all opinion polls, even with the notorious margin of error factored in - more or less 3% for evenly split voting intentions in polls carried out on 1,000 respondents. Although some polls carri…
on November 3rd, 2008 at 1:25 am
[…] The Media Philosopher reveals who could win the election based on sampling key phrases in social media - http://www.mediaphilosopher.com/2008/10/31/obama-declared-elected/ […]