LY ELECTION OVERVIEW: THE POLLSTERS WEIGH IN

C O N F I D E N T I A L SECTION 01 OF 03 TAIPEI 003771

SIPDIS

STATE PASS AIT/W

E.O. 12958: DECL: 11/22/2014
TAGS: PGOV TW
SUBJECT: LY ELECTION OVERVIEW: THE POLLSTERS WEIGH IN

REF: TAIPEI 03340

Classified By: AIT Director Douglas Paal, Reason 1.4 (B/D)


1. (C) Summary: Taiwan's top polling centers are predicting a
close election race between the Pan-Green and Pan-Blue camps
for control of the Legislative Yuan (LY),with neither side
currently expected to gain a majority. While pollsters agree
that a majority for either side is unlikely, their estimates
span a wide spectrum. The polling estimates range from a
three seat margin of victory for the Pan-Green over the
Pan-Blue, with the Pan-Green four seats shy of a controlling
majority, to a two seat victory for the Pan-Blue. The
pollsters consulted by AIT cautioned that, although they
employ rigorous polling methodologies, their polls cannot
fully capture Taiwan's complex political landscape. End
Summary.

Taiwan's Top Polling Centers: The Estimates
--------------


2. (C) The directors of some of Taiwan's top polling centers
say their polling data point to a close LY race with only a
slim margin of victory for either the Pan-Green or Pan-Blue
camp. Almost all media and party surveys predict that
neither side appears likely to gain a majority. In the 2001
legislative election the Pan-Blue camp, made up of the
Kuomintang (KMT),the People First Party (PFP),and the New
Party (NP),gained a majority with 115 out of 225 LY seats,
compared to 101 seats for the Pan-Green alliance, composed of
the Democratic Progressive Party (DPP) and the Taiwan
Solidarity Union (TSU). Of those 225 seats, 168 are elected
from local election districts, 8 from aborigine districts, 8
from overseas Chinese constituencies, and 41 are at-large
candidates (Reftel).


3. (C) DPP Survey Center Director Pan Yi-shuan told AIT that
internal party polls showed the Pan-Green leading the
Pan-Blue 109 to 106 seats -- an eight seat gain for the
Pan-Green and a nine seat reduction for the Pan-Blue. Pan
said the largest gains in the Pan-Green camp would come from
the DPP, whose overall seats she projected would rise to 94
from the 87. She said the TSU is expected to make marginal
gains, increasing its seats from 13 to 15. On the Pan-Blue
side, Pan predicted the KMT would consolidate its position
and pick up seats at the expense of the PFP. Pan warned,
however, that the DPP needs to be concerned that overly
optimistic predictions that the Pan-Green will gain a
majority could weaken Pan-Green-leaning voter turnout and
spur apathetic Blues to flock to the polls to defend their
endangered legislators.


4. (C) Era Survey Research Center Director Tai Li-an
estimates a more modest Pan-Green showing based on his polls,
projecting a four seat gain to 105 seats versus a twelve seat
loss to 103 for the Pan-Blue. Tai believes the remaining 17
seats will go to independents, many of whom lean Pan-Blue but
could, he said, easily be swayed by selective government
largesse to support Pan-Green initiatives. According to Tai,
his polls show that the weakness within the Pan-Blue camp is
coming from PFP candidates, especially those in the southern
and central districts of the island. Tai remarked that PFP
candidates are being dragged down by the declining reputation
of their Chairman James Song and the "ethnic-Mainlander"
brand of their party in heavy ethnic Taiwanese areas.


5. (C) TVBS Polling Center Director Wang Yeh-ting is the only
one to say the Pan-Blue are in the lead (Note: TVBS
consistently showed the Pan-Blue ahead by large margins
during the presidential campaign, with its March 20 exit poll
projecting a 6 percentage point win by the Lien-Soong
ticket). According to Wang, the Pan-Blue will win 105 seats
versus 103 for the Pan-Green. Like Pan and Tai, Wang says
his data show the DPP will make the biggest gains and take 91
seats while the TSU will gain only 12--far below the 20-plus
seats the TSU claims they hope to gain. Wang also says the
distribution of seats within the Pan-Blue alliance will shift
heavily in favor of the KMT, which will take 71 seats, to the
detriment of the PFP, which is projected to get 33 seats--a
loss of 13 seats from its 2001 showing. Wang's estimates,
like Tai's, project 17 seats for independents with members of
the Non-Party Solidarity Union (NSU) picking up 10 seats.
Wang said six of the of the remaining seven independents lean
Pan-Blue, which could help the Pan-Blue retain control over
the LY.

How They Do It: A Look at Polling Methods
--------------


6. (C) All three polling center directors described their
polling operations, which are generally similar in size and
methods. Each of the three centers employ 50 part-time
callers and poll randomly selected households during the
evenings between Monday and Friday. Their polling samples
are average 1,000 people -- the three directors agreed that
samples of 400 to 800 people, used by most other polling
centers on Taiwan, are too small and create large sampling
errors. The centers began LY district-by-district polls in
early October and completed their first roundabout of the
island in mid November. Pan said the DPP's efforts will
continue, focusing on the more hotly contested districts.
TVBS's Wang said his center will re-poll contested areas,
focusing on places such as Taipei city and county and
Kaohsiung city that have candidates with "star power." The
three centers' polls have relatively high response rates,
close to 80 percent for the DPP's Survey Center and around 70
percent for ERA and TVBS. Pan and Wang attributed the high
numbers to the use of "warm-up" questions and "bi-lingual"
callers who can conduct their interviews in Mandarin Chinese
and Fujianese (minnanhua) or Hakka dialects, depending on the
respondent's language preference.


7. (C) The three centers, however, have taken different
approaches to compiling their district polls into an overall
estimate of the LY outcome, which perhaps accounts for a
large measure of their variation. Pan said the DPP Survey
Center took its polling numbers as a basis, but also factored
in the results of projected "vote distribution" (peipiao)
into its calculations. Wang and Tai said their estimates are
more conservative, tallying up polling numbers without
heavily factoring vote distribution, which all parties are
likely to attempt. Wang admitted that TVBS had yet to
conduct polls in Penghu, Hualien, Taitung, Kinmen, and Matsu
and was basing its projections in those areas, as well as for
the aborigine and overseas Chinese constituencies, on "past
experience."

What They Can't Do: The Limits of Polling
--------------


8. (C) The polling directors acknowledged that the LY
election outcome -- even more so than the 2004 presidential
contest -- is difficult to predict because of political
factors and local variables that their polls cannot fully
capture. Tai said that voter turnout and vote distribution,
in particular, are likely to play a determining role.
Attempts by both camps in the final week of the election to
distribute votes from strong to weak candidates also add to
the unpredictability of the race. The distribution of votes
is determined by last-minute maneuvering and party
calculations that are difficult to gauge in advance through
public opinion surveys, said Tai. The 20 to 30 percent
non-response rate of those polled further compromises the
accuracy of election projections.


9. (C) Pan and Wang warned that some polling conducted on
Taiwan is plagued by other "soft" factors. In particular,
Pan said that local politicians may manipulate polls, even
commission their own data, for various propaganda purposes,
such as showing they are a viable contender in their
district. Smaller polling centers and companies are known to
skew their sampling to deliver results desirable to their
paymaster, according to Wang. Emile Sheng, an academic
polling expert who has consulted with both TVBS and ERA, told
AIT that he was recently threatened with a lawsuit by PFP
incumbent Pang Chien-kuo. Sheng said that Pang has alleged
that Sheng has intentionally skewed polls to show Pang out of
contention for re-election. Sheng noted, however, that all
media polls put Pang in exactly the same position (number 14
in the competition for a ten-seat district) in the hotly
contested Taipei City South district.


10. (C) Wang further pointed out that his center's polls,
like all household-based polling on Taiwan, are more likely
to miss young, unmarried professionals, who rely exclusively
on cellphones. Wang said that, on the opposite side of the
spectrum, some polling companies rely heavily on cellphone
interviews, which skews samples against people over 60 years
of age who, in the south, tend to be strong DPP supporters.
Another group that tends to shy away from participating in
polls, especially those perceived as "pro-Blue," are
Pan-Green loyalists. Many observers attributed the
inaccuracy of TVBS presidential campaign polling, especially
their March 20 exit poll, to this factor.

Comment: A Clouded Crystal Ball
--------------


11. (C) AIT's discussion with these polling directors
suggests that Taiwan polls are at best imperfect soothsayers
of the LY election outcome. Election projections,
nevertheless, appear consistent in pointing to a
consolidation and expansion of the two major parties, the DPP
and the KMT, that will come largely at the expense of their
partners, the TSU and the PFP respectively. The expectations
of an overall weaker Pan-Blue position track with what AIT is
seeing in various districts around the island and point to a
new post-election landscape for Taiwan. While the exact
numbers may be impossible to predict, it appears increasingly
likely that the Pan-Blue will lose it's LY majority and
present the Pan-Green, if it makes significant gains, with an
opening to woo independents and form a working majority.
PAAL