Idiot that I am, I couldn’t figure out how to insert a thumbnail in WordPress, which is what we use in Inquirer Current. To think that all I wanted to do was upload this.
Monthly Archives: March 2007
Well, at least the four on senatorial preferences conducted by Pulse Asia since July 2006. I uploaded the following crude Excel worksheet I put together to Inquirer Current (under a different heading: Looking for trends), with a view to understanding some of the trending that marks the 2007 Senate race. By crunching the numbers, perhaps later tonight, I might be able to answer even Benj’s important question. As always, please click on the image to read it.
In the Philippines, what do the Right and the Left have in common? They both think the United States, effectively, still runs the country.
I don’t know if the rest of the piece is as coherent as it seemed to me, though, when I was in the thick of writing.
I’m happy to see that Manolo Quezon, the country’s most influential political blogger, has finally come to believe in the gospel of “net dissatisfaction.”
I’ve come to be convinced by the view of Philippine Commentary that “net satisfaction” ratings are at best, misleading, and in general, a bogus figure. Better to focus on the numbers expressing satisfaction, and then the numbers expressing dissatisfaction, but don’t subtract one from the other.
I preached the same gospel 15 months ago, in a post titled “Net Dissatisfaction.”
One thing I’m sure of: The net rating, as an input for analysis, can be misleading. (Is this perhaps the reason why countries with a longer tradition of opinion polling, such as the US and the UK, do not use net scores?) ….
But the emphasis on the net scores (GMA’s net satisfaction rating had gone down to -30, her second-worst ever) makes it harder to understand what it is people really are saying. The real story lies in the actual satisfaction and dissatisfaction numbers, not in the (extrapolated) differential.
Philippine Commentary (Dean Bocobo, of course, the prophet of numeracy) was kind enough to leave the following comment, after reading that post:
The NET SATISFACTION RATING (positive minus negative) is a perfectly good statistic. But what most media don’t seem to know is that as a statistic it has TWICE THE ERROR as the individual components. So if you are CALCULATING NSR from a survey with 1200 respondents, the raw data has an error margin of about plus or minus 3% (actually its one over the square root of 1200, but never mind), the MARGIN OF ERROR in the NSR is PLUS OR MINUS 6%
Okay. That makes three of us. Plus or minus.
The various analyses of the latest Pulse Asia survey (some of them by bloggers I had read for the first time, others colleagues I look up to) leave me somewhat dissatisfied.
For one thing, I think no one has sufficiently acknowledged the essential difference between the February 2007 survey by Pulse Asia and other Pulse Asia and even SWS surveys before it: The conduct of the survey simulates the act of voting. (This change is a Pulse Asia tradition, if I’m not mistaken. In the 2004 elections, pre-election surveys up to January 2004 were conducted in the old way, while those beginning with the February 2004 poll were simulated electoral exercises.)
Does the mere act of filling up a “ballot” and then inserting it in a pretend ballot box change everything? Well, no. But it does change certain things. More specifically, it makes certain conditions improbable.
Let’s look at the contention (raised by the indefatigable Ellen Tordesillas) that Joselito Cayetano’s nuisance votes ought to be added to Alan Peter Cayetano’s. I would agree, immediately, if the survey respondents had been simply asked to choose their preferences from a given list. But the February Pulse was conducted differently. Respondents were asked to write their choices on a ballot-like piece of paper. Now I hold to the assumption that the electorate, by and large, knows what it is doing. Wouldn’t it be just as fair to assume that those who actually took the time to write down Joselito Cayetano’s name did so knowingly, as it is to assume, as the redoubtable Ellen does, that those who did did so under the mistaken assumption that they were voting for Alan Peter Cayetano? Perhaps some were merely having fun, or testing the limits of the survey, or actually registering a kind of protest against the voluble oppositionist, or voting for a KBL candidate. There are other possibilities; how can we readily assume that it was all a mistake? (And, oh, the 5 percent of “spoiled” votes are in fact already included in Alan Peter Cayetano’s rating; read the Pulse Asia press release again.)
My point is: This kind of survey, conducted this way, may actually give us a better snapshot of the voting public at this time, than the other kind of checklist survey.
To be sure, I am wary of accepting the random sample of “representative adults 18 years old and above” as a random sample of registered voters. (They may all be of voting age, but are they registered to vote?) Unlike a few other surveys it has done in the past, Pulse Asia does not explicitly state that all the respondents in the February 2007 survey are registered voters; but the language of its press statement leads us to think that they all are. (So until I check this with Pulse Asia, I will tentatively assume we are talking about registered voters.)
Does it make a difference? Yes, I would think so. In more survey-savvy polities like that of the United States, more weight is given to surveys of registered likely voters, for the basic reason that an unregistered voting-age respondent, or a registered voter who is disinclined to vote, will not materially affect the outcome of a given election.
If we assume that the February Pulse random sample is 100-percent registered voters, then a comparison with the SWS February survey, taken just a few days before Pulse started its field work, should be instructive. SWS scrupulously notes that, of its sample, only 88 percent are registered voters. That should mean that the views of the remaining 12 percent are just so much static, yes? I think, at the very least, we should factor that survey noise into our reading.
Thanks to Manolo Quezon, I was able to read atheista.net’s take on the surveys. I’m afraid, however, that this must-read blogger confuses the volatility of the rankings with the candidates’ ratings. Admittedly, Koko Pimentel is a good friend of mine, but precisely for that reason I follow his campaign closely. Now simply because Koko has gone down to 13 (he’s one of five candidates with a statistical chance of landing in 12th place), Atheista assumes that voters have come to a realization about him: He’s not Nene. But I take a look at the actual ratings (24.4 in January, 23.1 in February), and what do I see? Virtually no change, because of the survey’s margin of error of plus or minus 2. So what happened? Other candidacies exploded (and I really do think TV exposure and the lack of “negatives” is the key factor here: thus, Noynoy, Chiz, Joker, Edong).
With a fill-up rate of seven (that is, the average respondent named seven candidates to vote for), it seems to me that the race for the last six slots or so is, relatively speaking, wide open. The next two months’ campaigning will tell, especially since 10.6 percent of registered voters in the Pulse Asia survey (and some 16 percent in the latest SWS poll) are still undecided. They didn’t name even a single candidate.
The latest Pulse Asia survey, conducted between February 28 and March 5, should gladden Loren Legarda’s heart. It shows her leading all senatorial candidates with a vote-for index of 56.8 percent. Ping Lacson and Kiko Pangilinan, the topnotcher in the Social Weather Stations February 2007 survey, are distant also-rans, with ratings of 41.1 and 39.4.
A quick read of the results (I hope to return to this topic a little later): Alan Cayetano’s candidacy seems to have stalled, at (a still high) 30.9. The campaigns of Vic Magsaysay and Sonia Roco have improved their fortunes the most (they are now among the 17 candidates with a shot at landing in the Top 12). Ed Angara’s TV campaign is working (the ad featuring Sarah Geronimo, who seems to me to be a rather bland media-manufactured celebrity, is especially effective); now, at 32.1, he shares 4th to 10th place. And actor Richard Gomez continues to languish in the low double-digits.
The latest SWS survey (conducted a couple of weeks ago) is now available on the SWS website.
Kiko Pangilinan has taken over first place (although to be sure Loren Legarda is both within the margin of error and striking distance). Are those expensive ads for Mike Defensor, Migs Zubiri, and Butch Pichay really working? I suppose they are, but at markedly different "speed."
I’m afraid, though, that good friend Koko Pimentel (now at 14, still statistically within the "Magic 12" orbit) has become a target for junking. Ed Angara, who has obviously more money to spend and a bigger network to help him spend it, is marginally better-situated, at 13.