Astrid, Luka and Me

I’ve used up another 15 seconds of fame, this time in the Monday, September 29, Vitebski Kurier:

Vitebsk Election News Report

Translation:

Did expectations come true?..

Our attempt to have a conversation with one of the long-time observers of OSCE, coordinator Gary Ouellet from Canada were not successful. He providently refused by phone or other means to make any evaluations or comments on organization or process of the elections. His interpreter politely advised us to address all the questions to the headquarters of OSCE in Minsk. Nor did we manage to meet any other observers at polling stations to ask for their impressions. Nevertheless we were lucky to have a brief conversation with short-time observers Astrid Ganterer and Kevin Deegan-Krause (see photo) before the meeting of the District Election Commission of Vitebsk’s DEC #19 in the regional state administration building. It is their first time in Belarus, and they saw some parts of Vitebsk and Minsk. The representatives of OSCE mission persistently avoided all the questions connected to the elections. They kept silence, not revealing if their expectations came true or not. One could see either exhaustion or disappointment in their faces. Still, our foreign guests promised to answer all questions at the press-conference in Minsk…” (Thanks to our translator and several friends for clarifying the Russian).

Actually I am not allowed to clarify what emotion was on my face that day, but if it was reflective of the faces of those who reported for the OSCE in Minsk that day,  then “disappointment” would seem to be an appropriate choice (see http://www.osce.org/minsk/).

P.S. Search the picture above closely for the Hidden Lukashenka. If you enjoy that, you might want to try this Hidden Picture.

P.P.S. More election observation pictures (taken by others) online at:

Conservative Democrats of Slovakia

I saw the logo of the Conservative Democrats of Slovakia (KDS) for the first time today (apparently I live a sheltered life; or maybe it’s the party).

KDS
My first thought was that it was a surprisingly attractive and iconic design for a party that seemed to take so long to get its act together.  My second, and rather less charitable thought was that it is less than ideal for a party with less than 1% support in the polls to choose the asterisk as its symbol.  Or perhaps this, too, is just a clever way to sneak into the public opinion polls:

Smer, 44%
SDKU, 13%

Other*

My third thought related to the design.  Am I mistaken in thinking that this is merely the existing KDH (Christian Democratic Movement) logo
KDH logo
With an X through it?
You decide:
KDH and KDS Logos

Poll Analysis: UVVM Volatility

Also available in Slovak, here

I have occasionally expressed concern about the recent numbers coming out of UVVM, particularly those for the Party of the Hungarian Coalition.  Something is wrong there. In the spirit of my heroes at http://www.fivethirtyeight.com, I decided to take a look at the data.  The chart below calculates the total month-to-month differences in party results (up, or down) and then measures these as a percentage of a party’s average support during that period (i.e. a party with an average support of 20 percentage points and average monthly volatility of 1 percentage points would be listed at 5%).  I have done the calculations for the overall average (all of the polls within a given month) and for UVVM and for the two electoral periods for which we have good data: 2002-2006 and 2006-present.

Party 2002-2006 2006-2008
  Overall UVVM Difference Overall UVVM Difference
Smer 7.0% 9.2% 2.2% 8.9% 7.8% -1.1%
SDKU 12.5% 16.4% 3.8% 13.2% 16.2% 3.1%
SNS 14.7% 18% 3.3% 12.1% 12.1% 0.0%
MK 8.0% 12.0% 4.0% 10.1% 16.1% 6.0%
HZDS 9.9% 12.4% 2.5% 12.2% 14.1% 2.0%
KDH 10.4% 16.1% 5.8% 10.2% 10.9% 0.7%
Average 10.4% 14% 3.6% 11.1% 12.9% 1.8%

What is not surprising here is that UVVM alone has much higher levels of volatility than the overall average which includes multiple polls that smooth out the monthly variability.  What is surprising here is the fact that the the difference in volatility of SMK between UVVM and the overall average grew (from 4% to 6%) even as the differences between UVVM and other surveys declined.  Between 2002 and 2006 the volatility of SMK was second to the bottom (after Smer) in both UVVM and overall averages.  Between 2006 and 2008, the SMK volatility stayed the second lowest in the overall averages but rose to highest in UVVM by a wide margin.

This is all particularly surprising since the overall volatility of the Hungarian Coalition’s electorate have remained remarkably stable over time, as the following table demonstrates:

Party (elections in sample) Volatility as a share of average party support Raw volatility
ANO (2002,2006) 103.2% 4.9%
SNS (1994,1998,2002,2006) 80.5% 5.9%
Smer (2002,2006) 73.6% 15.7%
KSS (1994,1998,2002,2006) 51.2% 2%
SDKU (2002,2006) 45.1% 7.5%
HZDS (1994,1998,2002,2006) 38.7% 8.7%
MK (1994,1998,2002,2006) 11.4% 1.2%
KDH (1994,2002,2006) 10.7% 0.9%

There is an easy explanation for this:  UVVM has had difficulty maintaining its Hungarian sample.  This is understandable–this is a difficult task–but it is important to keep this potential problem in mind rather than to assume that all polls actually reflect representative samples.  This should be a question of analysis rather than assumption.  It also raises questions about the representativeness of this sample (and that of other pollsters) for other parties.

I am deeply grateful to UVVM for all the support they have provided to me over time. This is a problem that needs attention but I am hopeful that the institute can return to its previous levels of excellence.

October 2008: UVVM

UVVM Monthly Report: October 2008

Also available in Slovak, here

UVVM reverted to the mean this month, its poll numbers pulling back from extremes for most of the major parties and toward the results obtained by other pollsters.  There are gains to be made from a comparative analysis of various polls, but as usual the major papers (http://spravy.pravda.sk/opozicne-strany-si-mierne-polepsili-dvg-/sk_domace.asp?c=A081015_151947_sk_domace_p04)
present these without a hint of analysis from ČTK  which itself takes these numbers as givens rather than as a particular (and with regard to SMK, particularly troubling) survey.

Otherwise (and maybe this is why the papers do not bother to put more energy into it) there is not much to see here that is new:

Smer leads, but not as much as before.  SDKU moves slightly upward into clearer (but far distant) second place while SNS sinks toward the other three major parties:

UVVM+poll+data+ for +all+parties+ for the most recent +24+months+ in Slovakia
The movement of the mid-size parties is clearer here.  Particularly troubling is the movement for SMK.  A party which has by far the most stable overall electorate and the most stable historical voting patterns has the highest level of volatility in month-to-month poll numbers of any major party in UVVM surveys (see a more in-depth post here: http://www.pozorblog.com/#/p=110).  (HZDS has also been rather volatile, and while it is possible that this party’s support is actually a bit more volatile, its major swings also seem to be more the result of survey networks than actual shifts in public opinion.)

UVVM+poll+data+ for +all+parties+except+Smer+ for the most recent +4+months+ in Slovakia
Aggregated according to blocs, the shifts above mean a slightly closer gap between opposition and coalition, but no different than the overall average for the period.

UVVM+poll+data+ for +coalition+support for the most recent +24+months+ in Slovakia

UVVM+poll+data+ for +coalition+support for the most recent +4+months+ in Slovakia

Vote blocs, too, stay roughly the same.

UVVM+poll+data+ for +party+'blocs'+ for the most recent +24+months+ in Slovakia

UVVM+poll+data+ for +party+'blocs'+ for the most recent +4+months+ in Slovakia

And Smer is still likely to be able to build a coalition with a single party of its own choosing.

UVVM+poll+data+ for +estimated+party+seat+distribution for the most recent +1+month+ in Slovakia

UVVM+poll+data+ for +estimated+party+seat+distribution for the most recent +24+months+ in Slovakia

As always, the actual polling numbers are available online at Google Docs:

http://spreadsheets.google.com/pub?key=pdhlCClsiyAMi39bLFpY_Zg

And the most recent three months are below in tabular format (using “iframe” which may not work on all browsers).

<br />

pozorblog po slovensky

Amid constant news of contraction and decline this week, there is (slightly old) news of yet another small but significant expansion: machine translation to and from Slovak.   Machine translation is old news these days (though it is no less remarkable for our acceptance of it) and is no panacea (it is still bad enough that without a basic understanding of the language it is easy to be deeply confused or seriously misled), but it is nice to see Slovak make the list of Google Translate languages.

For this blog it makes little difference–the population of those who are interested in what goes on here but cannot read English is infinitesimal–but it is nice to be able to make the gesture and all subsequent posts here will include a link like the one below allowing automatic translation.

http://74.125.93.104/translate_c?hl=en&sl=en&tl=sk&u=http://www.pozorblog.com/%3Fp%3D108&usg=ALkJrhgjlN7kuRnEQTuh4dfXjTjpiKFfUw

I hope in future to go one step further and install a plugin that makes that process even easier.

More Pozorblog

It has occurred to me that it might be useful for me (and maybe for others) to keep track of articles in the Slovak press that are interesting or in some way provocative and that the easiest way to do this through the social-bookmarking service called del.icio.us

I’ve therefore begun to use the tag ‘pozorblog’ in del.icio.us and all the articles with that tag are listed on a specific page:

http://delicious.com/kdecay/pozorblog

Thanks to del.icio.us this has its own RSS feed so anybody so obsessed can keep track of postings in this area. We shall see if this experiment goes anywhere.

Not To Minsk Words

I returned two days ago from Belarus where I was an observer in the 2008 parliamentary elections.

Lukashenka and Flag Courtesy of Wikipedia

OSCE rules request that we leave assessments to the organization’s main report and remain silent.  Fortunately, in this instance, the OSCE judgement so closely corresponds to my own experience that I can refrain from my own judgments and express my own  empressions exclusively through quotations from the initial report.  That report can be found online at http://www.osce.org/item/33272.html.  Unless otherwise noted, the quotations reflect my own personal experience.

The Scope Of The Mission:

As OSCE notes, “On Election Day, 449 observers were deployed to observe the opening of polling stations, the process of voting, the vote count, and the tabulation of the votes at DECs. This included 76 specially designated teams to observe the tabulation process.”

What Went Relatively Well: The Voting Process

The OSCE report perfectly mirrors my own experience regarding the experience of voters in the voting process:

  • “On election day, observers reported that voting was well conducted overall in those polling stations visited.”
  • “Observers generally evaluated the opening procedures as good or very good in 100 per cent of the 93 cases observed.”
  • “The voting procedures were also positively assessed by observers, with 95.4 per cent of cases evaluated as good or very good.”

With a relatively minor exception:

  • “Campaign materials were displayed inside polling stations in 3 per cent of cases.”

What Did Not Go Quite As Badly: The District Election Commission Process

Again, the OSCE report mirrors my own experience regarding the conduct of the aggregation of the count at the district level:

  • “While some opposition candidates claimed to have been the subject of pressure on the part of local administrations, other candidates, including from the opposition, declared that the attitude of DECs was friendlier and more open than in the past and that the pre-election climate was improved.”

Nevertheless, as the OSCE reports,

  • “In 54 per cent of tabulations observed, they were not able to observe the figures being entered into the spreadsheet tables.”

What Went Badly:  The Counting Process

As the OSCE report concludes, “Voting was generally well conducted, but the process deteriorated considerably during the vote count. Promises to ensure transparency of the vote count were not implemented….” The process deteriorated considerably during the count and tabulation, violating paragraph 7.4 of the Copenhagen commitments of the OSCE.

  • “The integrity of the process was undermined by the vote count which was assessed by observers as bad or very bad in 48 per cent of observations [including me]. “
  • “37 per cent of observers, including some of those who noted hindrances [i.e. including me], reported not having a full view of the vote count proceeding, thus compromising the transparency of this fundamental element of the election process.”
  • “OSCE monitors were prevented or hindered from observing the vote count in 35 per cent of cases. This compromised the transparency of this fundamental element of the election process.”
  • “In 50 per cent of cases, early votes were not compared with the number of entries in the voter lists.”
  •  “Observers could not see the voters’ mark in 53 per cent of cases.
  • “Numerous cases were noted of counting procedures taking place in complete silence with small slips of paper being passed between commission members; this significantly undermined any transparency in the count.”

Reading Between the Lines:

What to make of an election that combines relatively smooth voting and district-level counting with comprised transparency at the polling-station counting stage?  The OSCE report makes certain suggestions about the possible reasons.  These conform to my own experience.

  • “From observers comments, in some instances it was noted that there were significant discrepancies between turnout observed and the number of votes noted in PEC protocols…. A high incidence of mobile voting was noted in some cases.”
  • “Where access was possible, several cases of deliberate falsification of results were observed…. Deliberate falsification was observed in 5 cases by observers.

If, indeed, the non-transparent counting procedures reflect discrepancies, this requires an explanation of how the aforementioned discrepancies could be combined with a relatively transparent voting process.  The OSCE report offers several potential explanations that correspond to my own experience:

  • “Lack of clear detailed regulations on the printing of ballots, the number of ballots to be printed, the percentage of extra ballots, and security features”
  • “Lacking instructions on observation, each PEC was free to decide on how observation would be dealt with.”
  • “The Electoral Code does not provide any clear mechanism for securely keeping the ballot boxes after the start of early voting, nor does it provide specific regulations for enhancing the integrity of the ballot.”
  • “The lack of any official protocols to document the record of voting on each day of early voting remains a concern.  These outstanding issues allow the possibility of electoral malfeasance.”

And, perhaps most significant of all,

  • In nearly all cases in which OSCE/ODIHR EOM observers had access to such information, they reported that PECs were composed of staff from the same place of work, such as enterprises or schools. Existing hierarchical relationships seem to have been transferred to the PEC, i.e. heads or deputy heads of such work places became PEC chairpersons, with their staff as the PEC members. This further contributed to the lack of independence of individuals in the commissions…. The composition of election commissions diminished stakeholders’ confidence in the process.

Most interesting of all, is the question of why.  If, as OSCE notes, “The election took place in a strictly controlled environment with a barely visible campaign,” why should engineered discrepancies and falsifications be necessary at all?  The need to make minimum turnout requirements? The desire for overwhelming margins of victory?  Simple habit?  I will leave answers to these questions for other experts and other venues.

September 2008: Poll Comparisons

Trends and comparisons monthly report

Inside the averages are multiple polls using slightly different methods and samples.  When put together these yield a surprisingly smooth and consistent picture that is the most useful for assessing levels and trends, but it is useful to take them apart from time to time to discern any internal shifts and to assess the work of the pollsters whose efforts combine to make the overall average.

Between June and September we had one month with only one poll (July) and one with only two (August), so the most useful baseline comparison here are the first and last entries for each poll.

For Smer the story is consistent over time.  Among FOCUS respondents Smer achieved a near-record high in the June survey; among UVVM respondents the high came in the July survey and as always it was considerably higher than Smer numbers in the other surveys. We do not have MVK for this period but Smer almost invariably polls lower for MVK (nearer to FOCUS-levels) than for UVVM.  Over time UVVM numbers show stability around 46%, while FOCUS shows a decline from around 44% to around 39% and MVK shows a rise from 39% to 41%, meaning that the overall shift is slightly downward.

Multiple-poll+average+ for +Smer+ for the most recent +4+months+ in Slovakia

The SDKU graph is rather striking and shows an unusual pattern: near unanimous agreement between the three pollsters in June and August followed by almost the same mean in September composed of divergent estimates (FOCUS, as usual on the high side, UVVM as usual on the low side and MVK, as usual (for both SDKU and Smer) near the mean.

Multiple-poll+average+ for +SDKU+ for the most recent +4+months+ in Slovakia

Results for SNS show a shocking degree of consistenc.  All three polls remained within a single point and showed the exact same pattern: a rise between June and August followed by a decline below June levels in September.  This kind of univocality is really quite.

Multiple-poll+average+ for +SNS+ for the most recent +4+months+ in Slovakia

But just when the polls were looking consistent, results for MK give us reason for taking any estimate with a large grain of salt:  Of all parties it is the Hungarian Party’s almost exclusively ethnic base that should yield the most consistent answers and yet UVVM numbers for September tend to wander around more than for other, non-ethnic parties, this month dropping below 7%.  This is highly unlikely and appears to reflect a long-term problem on the part of UVVM to get the ethnic samples consistent.  The Hungarian ethnic population is stable, and the overall averages for MK are quite stable (though showing a long term pattern of decline) so MK’s numbers should be relatively stable as well and for UVVM they simply are not.  This, of course, raises questions about how other sampling questions could be affecting other parties’ results (for any of the polls, not just UVVM) and suggests the importance of looking at multiple polls.  As it is averaged here, MK remains stable at just above 9%.

Multiple-poll+average+ for +MK+ for the most recent +4+months+ in Slovakia

Similar problems emerge with regard to HZDS.  All three polls agreed on a level of 8%-9% in June but have since diverged, with FOCUS and UVVM show a drop to around 7% while UVVM alone shows a rise to arund11%.  This is a divergence of nearly 50% on an average support of 8% and something is clearly wrong here.  It is not clear what, but a facile decision by majority rule suggests that the problem may lie with HZDS.  Given the inconsistencies, it would not be surprising to see the party’s preferences in the UVVM survey fall back to around 9% in October.

Multiple-poll+average+ for +HZDS+ for the most recent +4+months+ in Slovakia

With KDH there is simply not much to say.  As with SNS, the polls tend to agree and show the same patterns over time.  Here, however, the pattern is absolute stability around 9%. This is what one would expect of a “demographic” population rather than the variations in MK seen above.

Multiple-poll+average+ for +KDH+ for the most recent +4+months+ in Slovakia

For the smaller parties it is always more difficult to gauge, but KSS remains stable around 2% (FOCUS and UVVM show a slight overall rise, MVK shows a drop, but the June MVK number was a recent-record-high for MVK and may have been an outlier in the first place.

Multiple-poll+average+ for +KSS+ for the most recent +4+months+ in Slovakia

SF, too, hovers around 2%, and from June to September shows almost no change in any of the three polls, though they disagree among themselves whether the party is closer to 1% (UVVM) or 3% (FOCUS)

Multiple-poll+average+ for +SF+ for the most recent +4+months+ in Slovakia

There are some issues that I realize I am not dealing with in these analyses, particularly support for the Green Party (SZ) which is included in the questioner list only by MVK, and the question of non-voters, which is worthy of attention and should become a topic of future monthly and quarterly graphs.

September 2008: Poll Averages

Overall Monthly Report

Both FOCUS and MVK released their numbers in the last 48 hours, and while it is difficult to talk about bad news for a coalition with over 60% of the likely vote, September was not a particularly good month.

As the overall graph below shows, Smer dropped slightly from its plateau, though it still remains well above past plateaus and shows a slightly upward trendline over the past 2 years. The bigger changes came within SNS, which posted its biggest monthly drop in years, down to just above 10%. There are also some questions about HZDS’s slight gain that I will deal with in the next post (the party’s gains in the UVVM survey are sharply at odds with its losses in FOCUS and MVK surveys.)

2:||||||||||||||||||||||||

For the current opposition, there was no particularly encouraging news but SDKU did turn up slightly after nearly a year of downward trend (of course it has turned up slightly in other months during the overall decline) and both the Christian Democrats and the Hungarian Coalition remained stable, as has been their habit.

Multiple-poll+average+ for +all+parties+ for the most recent +4+months+ in Slovakia

The overall news is therefore reversion toward the two year mean for both coalition (dropping to just over 60%) and opposition (rising above 30%). The remaining 9% or so remains divided among parties across the party spectrum and both Slobdne Forum and the Communist Party posted slight gains this month but remain at the 2% mark. HZD and ANO, not pictured here, face an even less hopeful prospect, with numbers around 1% and perfectly flat trendlines.

Multiple-poll+average+ for +coalition+support for the most recent +24+months+ in Slovakia

Within the framework of (rather artificial) party blocs, these changes put the parties of the “Right” ahead of the “Slovak National” bloc, which has fallen by this standard to its lowest level ever (which is not to say that “Slovak National” ideas are at their lowest ebb since these have clearly been incorporated into the rhetoric of Smer.
Multiple-poll+average+ for +party+blocs+ for the most recent +24+months+ in Slovakia

Multiple-poll+average+ for +party+blocs+ for the most recent +4+months+ in Slovakia

In terms of the overall consequences for electoral politics, the report of these results remains a broken record: Smer remains in a strong position to chose a single party as its post-election coalition partner.

Multiple-poll+average+ for +estimated+party+seat+distribution for the most recent +1+month+ in Slovakia

Multiple-poll+average+ for +estimated+party+seat+distribution for the most recent +24+months+ in Slovakia

As always, the actual polling numbers are available online at Google Docs:
http://spreadsheets.google.com/pub?key=pdhlCClsiyAMi39bLFpY_Zg

And the most recent three months are below in tabular format (using “iframe” which may not work on all browsers).

<br />