Wednesday, November 17, 2010

Polls Apart

Over the last week or two three different organisations have issued opinion polls showing possible outcomes of the November 28th election.  These were the Institute of Public Policy, CBS-AXA and the Association of Sociologists and Demographers.

Polls in Moldova tend to produce strange results and accusations are frequently made that polls have been 'bought' by one party or another to improve its image.  When one only has two polls to look at it is difficult to figure out where the truth lies.  With three polls it's a lot easier, because the polls can corroborate each other (or not), and outliers stick out like a sore thumb.

Consider the following table.

  1. In the first column you have the names of each major-ish party.
  2. In the next three columns you have the poll results produced by each organisation (as a percentage of those who have expressed a preference); the yellow cells are the ones which I think are outliers and need to be tossed out.
  3. In the column headed 'Average' I've worked out the average of the three poll results excluding the outliers and any non-existent data.
  4. The "Rescaled" column adds back in the percentages lost when the outliers were thrown out.
  5. The adjusted column takes into account voters living abroad (who were not polled but will vote).  I've give the PL and the PLDM an extra 1% each from this source.
  6. The number of seats is based on the new formula, where votes lost by parties not crossing the 4% threshold are reallocated approximately equally to parties that do make it into parliament.

There are a number of observations here, which need to be addressed urgently by the centre-right:

  1. The Liberals and Liberal Democrats between themselves do not have enough seats to form a government.  They will have to cooperate with the PD and put up with all the blackmailing that involves.  Better to push hard now for a majority.
  2. The AIE parties will fall two short of having enough seats to elect a president.  While it may be possible to convince a couple of communists to come across, they shouldn't rely on this and instead should push hard to get across the line.
  3. Three smaller centre-right parties (Our Moldova, European Action and National Liberal) will each miss the threshold but will accumulate 6.2% altogether.  Folks, I know you don't like each other, but please grow a collective brain and put together a common list so that you make it into parliament.  By my reckoning, this would reduce the communists to 40 seats in total, and they wouldn't be able to block the election of a president.
  4. The AIE parties need to focus on their real opponents, which are the communists and any of their nasty satellite parties with a chance to make it to 4%.  The recent spats between Ghimpu and Filat don't impress the public and will only serve to reduce the votes collected by the AIE, and must stop.  Now.
  5. The Liberals have a good chance of being the third largest party in parliament.  This is a prize worth having, as it will temper somewhat the ambitions of the Democrat Party (e.g the installation of Plahotniuc as prime minister)

1 comment:

  1. Good analysis, altough I'd put ASD's score for PCRM as an outlier too. ASD always underestimates their score and this is probably no exception.

    Other correction I usually make is the residual percentage - there are 39 candidates now and the last 32 of them should make together some 10%.

    It's funny how no alliance was created this time - even if now it is legally possible - Moldovan politicians really do have a goldfish memory.