Cheap shot from Golding: No political bias in polls

[Karl]

Cheap shot from Golding: No political bias in Gleaner-commissioned polls
published: Sunday | May 27, 2007

Byron Buckley, Associate Editor

Bruce Golding, leader of the Jamaica Labour Party (centre), addresses supporters on the campaign trail in Savanna-la-Mar, Westmoreland, earlier this month. He is flanked by party colleagues Dr. Horace Chang (left) and Russell Hammond. Golding has implied that the Gleaner-Bill Johnson polls are biased towards the People's National Party.- photo by Dalton Laing

Stunned by last Sunday Gleaner's poll report that his Jamaica Labour Party (JLP) was trailing the rival People's National Party (PNP) by seven percentage point, Bruce Golding attacked the bearers of the bad news.

According to the JLP leader, Bill Johnson is the PNP's pollster on loan to the Gleaner Company. Cheap shot. The inference to be drawn is that the series of Bill Johnson polls commissioned by The Gleaner are biased in favour of the PNP.

A drowning man grasps at straws. Golding, understandably, might have experienced a sinking feeling on seeing his party drop five points since January when the JLP held a one-point lead over the PNP in Bill Johnson polls done for The Gleaner. Then, there was no PNP labelling of the pollster or The Gleaner.

All honky dory
Everything was, apparently, honky dory with Johnson's poll findings when The Sunday Gleaner's lead story of April 22, screamed, 'Tough to beat - Tufton poised to take St. Elizabeth South West for JLP'. In that story we reported that the JLP, with 52 per cent popular support, was ahead of the PNP at 25 per cent.

The previous week, The Sunday Gleaner reported in a front-page story, 'Portmore seat wide open', that the JLP (29 per cent) was leading the PNP (28 per cent) in St. Catherine South East. Johnson found that the JLP's Arthur Williams (28 per cent) was ahead of the PNP's Colin ************an (24 per cent). There was no complaint from Golding about polling bias.

On May 6, The Sunday Gleaner lead story was headlined, 'PNP ship sinking: JLP says Palmer's resignation indicative of a party in trouble.' The poll findings showed the parties tied at 33 per cent in voter support, but with the JLP's Sally Porteous at 42 per cent leading the PNP's Vando Palmer at 30 per cent. Palmer resigned as candidate two days before the poll results were published.

Furthermore, the May 13 edition of The Sunday Gleaner carried a front-page story, 'JLP has a shot at St. Andrew South East'. We reported the party standings, with the JLP at 35 per cent ahead of the PNP at 30 per cent. Golding must have been pleased; there was no protestation from him.


A tally of the five constituency polls carried by The Gleaner as at last Sunday shows the JLP ahead in three seats and the parties deadlocked in two. A sixth poll result carried today shows the JLP with 29 per cent and the PNP 26 per cent of popular support in St. James West Central. So, where is the PNP bias in The Gleaner's poll reports to which Golding alludes? Could it be that in an effort to steady himself from the impact of the poll findings, he accidentally threw a blow below the belt?

Accuracy of polls
Golding's reaction aside, the publishing of the parties' standings last week - PNP at 34 per cent and JLP at 27 per cent - ignited a firestorm of criticism about the accuracy or validity of Johnson's polls. People have questioned how the national poll findings did not reflect the trend from the constituency polls - a JLP lead. But is this a valid expectation? Can poll findings in five constituencies out of 60 reflect the national political situation?

The answer is no, because these five constituencies are not a representative sample of the entire 60. No attempts should be made to generalise a particular constituency poll finding. There is an exception to this rule which we will address later in this article.

Furthermore, based on the rules of research, the size of the sample (480) used in the constituency polls carries a larger margin of error - five per cent - compared to three per cent for the sample of 822 registered votes used in the national poll. This difference in error margin means the constituency polls are less accurate than the national poll with its smaller error margin. So, any comparison of the constituency poll findings with the national poll results is like comparing apples and breadfruit.

On the matter of margin of error, readers need to interpret poll results bearing this statistic in mind. Let me explain. When the margin of error of three per cent is applied to the PNP's popular standing of 34 per cent as captured by Johnson's recent national poll, the figure could increase or decrease by as much as three per cent in an actual election.

So, conceivably, the poll findings in the constituencies in which Johnson found the JLP leading could shift up or down by the five per cent margin of error in a real election as illustrated in the table.

Of course, movements within the error margin are not uniformed. Factors can affect the PNP and JLP simultaneously, and to different degrees, thus resulting in either party gaining or losing an advantage. In the cases of St. Andrew East, and Manchester Central, where both parties were in deadlock, any movement within the error margin could shift the balance of power.

Part of the difficulty readers expressed in response to the release of the party standings last Sunday was that the national poll results did not reflect the findings at the constituency levels. This is a misunderstanding on the part of some readers.

Pollster's credibility
The national poll results are not the aggregate of 60 mini-polls in constituencies. For the purpose of conducting a national survey, the island becomes one big constituency with respondents drawn from a wide cross-section of areas to ensure the sample is representative. This approach helps to minimise bias from, for example, garrison constituencies or political enclaves.

The credibility of pollster Bill Johnson was also called into question last week. Critics pointed to the variance in his poll findings related to the outcome of the general election in St. Lucia last year.

Close to election day, in an attempt to influence supporters, the ruling party of then Prime Minister Kenny Anthony released poll findings by Johnson indicating that the party was ahead of the opposition party in popular support. But, the ruling party subsequently lost the election six seats to the opposition's 11, while both parties tied for popular support.

Johnson explains that he had conducted a series of polls for the ruling party; and while the first of these showed the ruling party ahead of the opposition in popular support, the latter polls showed the party rapidly losing voter support. In fact, Johnson says, his last poll had accurately predicted a tie in popular votes for each party.

One lesson from the St. Lucian election for us is that the popular vote split does not directly reflect the pattern of voting in constituencies.
As mentioned earlier, there are instances where poll findings in particular constituencies are used to predict generalised outcomes. But, in fact, the prediction is based on historical patterns from actual election outcomes.

For example, St. Elizabeth SW has always gone to the party that forms the government after an election since Independence. Political scientists describe this as a weather-vane seat because it signals the direction of the political winds of change.

Hence, the importance of The Gleaner's probing of the electoral situation in a seat like St. Elizabeth SW. The findings there and in similar-type seats point to the likely national winner of the next general election.

[Karl]

Golding needs to stop talking

Golding needs to stop talking
published: Sunday | May 27, 2007

Dawn Ritch, Contributor
Opposition Leader Bruce Golding says he's not perturbed by polls, but has a press conference nevertheless.

"According to the Jamaica Labour Party (JLP) leader, Johnson's latest poll does not reflect what his constituency polls have been reflecting over the last four or five weeks. Those polls, Golding said, showed the JLP leading in all constituencies except one, and showed swings to the JLP of up to 8.7 per cent in one instance." (Daily Observer, May 21)

Somebody needs to advise the Opposition leader to stop talking. Because every time he opens his mouth he incriminates himself. The soundest image advice possible would be to cease and desist.

There are 60 seats, and Bruce is saying at a press conference he has 59 of them. Then whose does he not have in the governing People's National Party (PNP)? Is it only the Prime Minister's?

According to the JLP leader, this is what his scientists have told him. They have told him that the 'Portia factor' is non-existent. It will only be played out in that she is being allowed to keep her seat. Now, I don't know who is counting, but it can't be a national poll.

Try as hard as you can to find every little pocket of support for the Jamaica Labour Party, the Prime Minister and her People's National Party just sail away with national approval.

The latest Bill Johnson Poll shows Mrs. Simpson Miller's poll figures on their way up again. She has established a 19 per cent personal lead over Bruce, and her party is seven per cent ahead of his. That's with the PNP not yet campaigning in the field, and no money.

It shows a rejection of Bruce Golding as having any electoral possibility at all. And it also demonstrates that the Jamaican people believe what she says. A different poll, the CVM Market Research Services survey, shows the two parties in deadlock still.

The irony is that the PNP train has not yet started to roll. It has not built up any momentum yet. They're not out in the field as a party organisation, and have no ads running on electronic media. Unlike the JLP.

Bruce is hysterically calling for general election, and the Prime Minister is completely ignoring him. Only she can decide the date under the Constitution. Bruce should go settle himself and have a drink, because nobody wants to see him. Nobody's turning out.

Nevertheless, at his recent press conference, the Opposition leader said the JLP would win the next election "comfortably". Thus, has his projected landslide on the order of 1980 which he'd publicly promised earlier, dwindled to a mere comfortable margin. All this from the same lips, and before a general election has even been announced. Only a fool or a rank opportunist could take a word he says on faith.

That's a bad quality to have in politics. You need fixed reference points, persistence, ability and the capability not to turn off people, either singly or in droves. So it was clear to me long ago that Bruce Golding, as leader of the JLP and the Opposition, was not even a choice made by a committee. It was made by a computer.

Personal history
Somebody can't just reimage him or herself. Everybody has a personal history and record that seeps like water into the stone of a people's consciousness. Any party that tries to spend money changing an ounce of that is politically insane.

Worse yet, the JLP is causing no end of public controversy with the Endimites Bruce brought in and to whom he gave seats. They all listen to their navels, nothing else. They have a political tin ear. This is demonstrated by Joan Gordon-Webley with her 'Jezebel' remarks about Portia, and Chris Tufton complaining that Buchanan said 'Satan' tempted him.

This only calls public attention to the exemplary life of the country's first female Prime Minister, and casts an unflattering spotlight upon her accusers. Then, Tufton nonsensically reminds the whole country by making a fuss about 'Satan', that in the run-up to the last general election on a JLP platform, he told the assembled crowd of thousands to vote PNP. None of this is helpful to the JLP.

Besides which they use these descriptions as though they were mere swear words and had no other meaning. When Mrs. Simpson Miller quotes from the Bible, or believes in God, this is not done for political effect. That's just the way she is. Myself, I think no sermon should last longer than five minutes, with the sole exception of Anglican parsons, who can speak no more than 15. It's unfair to the human race. Any point can be made memorably in five.
This means there's no time on television for faking controlled anger, or rocking back and forth while stroking your lip. It's the waste of a good sound bite, which is usually only two minutes. Bruce Golding should remember this when next he spends so much time sneering on camera. Their chances are best at the polls if he is retired quickly.

Recovered some lost ground
Mrs. Simpson Miller has just established a commanding lead for the People's National Party, and recovered some lost ground for herself.

Another poll shows the statistical deadlock of the past six months in the country remaining unchanged. Both hugely divergent polls have to be viewed against the fact that the PNP party organisation has not yet cranked up in the field. On the other hand, the JLP has been campaigning non-stop for the duration, and plans to give their donors medals.

The only conclusion, therefore, is that when Bruce goes into the field, he digs a deeper hole for himself and his party, and when she goes, she builds credits for herself and hers.

This is a demonstration of the 'Portia factor' in action. Considering the PNP's relative inactivity, both polls give her credibility a boost, and offer no comfort whatsoever to the JLP. So there seems little to be gained from claiming and hoping that her public persona is not an asset.

Above all, and this is where Bruce fails repeatedly and utterly, she inspires trust. That is the currency of politics, and Mrs. Simpson Miller has her very own printing press. All she needs to worry about is inflation, particularly since I hope she will become Prime Minister and Finance Minister.

Years of economic decline have meant that people would sooner have someone to trust than eat food. Such a person represents hope and the future possibility of good order and a prosperous life. There is no substitute for trust.

Whether it's constituency polls, national polls, or the Budget speech, we either trust you or we don't. Those who don't are not even inclined to listen to what you say. Such is the value of trust. On this overwhelmingly important matter Bruce Golding is bankrupt. Even if a computer selected him, he should never have agreed.

[Assasin]

It seems like the attack dogs are out. Not one thing of substance or policies in these two articles. Nothing but political nonesense which is not important.

They could find some substance to put inna the articles and these suppose to be respected Journalists? What if Golding say him nuh believe the polls because it nuh suit him? Nuh dat every politician do?

Now wonder Jamaica is where it is, if we can talk about foolishness and not one substance whether it be Portia dress or this foolishness, don't tell me them can't find a better story.

[Muadib]

Karl.. are you related to Dawn Ritch ?

[Karl]

Karl.. are you related to Dawn Ritch ?

No!
Actually I wonder what is Dawn's real motive. Dawn is a die-hard JLPite! Why the nice things about Portia? Is she angling for a job?

[Karl]

Polls, prophecies and predictions
Claude Robinson
Sunday, May 27, 2007


Bill Johnson's poll finding that the ruling People's National Party (PNP) has apparently surged ahead of the Jamaica Labour Party (JLP) seemed to have spooked the Opposition, with key officials doing their best to discredit the numbers and pooh-pooh their consequences.
Claude Robinson

Opposition Leader Bruce Golding dismissed Johnson as a fraud and the findings as fraudulent, while the party's general-secretary Karl Samuda was equally derisive as he predicted in a radio interview that the JLP would not only retain the 26 seats won in 2002 but would "comfortably" take nine seats now held by the PNP.

On the same Breakfast Club radio programme (Tuesday) PNP general-secretary and minister of information and development, Donald Buchanan, asserted that the results mirrored the party's own private polls, conducted almost weekly, and other soundings of voter preference.

Buchanan went further, predicting that the PNP would take four seats from the JLP. He identified these as Western St Thomas (held by JLP deputy leader James Robertson); North-East Manchester, held by JLP heavyweight Audley Shaw; North-Central Clarendon held by Pearnel Charles and North-West St Elizabeth where he says Ann Marie Warburton will whip JC Hutchinson. Samuda brushed aside the prediction, saying Buchanan was living in "la-la land".

The media frenzy was fuelled by Johnson's latest poll, published in the Sunday Gleaner of May 20 which found that if an election was held at the time of the poll (May 4-5), 34 per cent of persons polled would have voted PNP, compared to 27 per cent for the JLP.

The finding appeared to be in sharp contrast with Johnson's poll in January which found the parties in a statistical dead heat: PNP, 31 per cent and JLP, 32 per cent although the pollster was careful to say that it was too early to tell whether the lead was a temporary surge or a more lasting change that will hold. This could only be determined by future polls.

As reported by the Gleaner, the poll result showed that the JLP has gained just one percentage point after nearly 10 months of campaigning, moving from 26 per cent in March 2006 to 27 per cent in May 2007. The party hit the campaign trail on July 26 last year.

This could hardly be comforting news to an Opposition party which has been lambasting underachievement and corruption in the PNP Government that has been in office for 18 years.

The Johnson finding was also at odds with the pollster and colleague Mark Wignall who was asserting at the same time (Sunday Observer May 20) that his own polling and analysis of other data feel on the ground led him to conclude the PNP will lose the next election.

Meanwhile, the polling pool was further muddied Wednesday night with Don Anderson reporting on CVM-TV (Direct) that his findings, based on a survey also done in early May, still had the two major parties in a dead heat at 25 per cent each.

These findings are about the same as Anderson's earlier poll (February 2007) in which the JLP polled 27 per cent, compared to 26 per cent for the PNP. His new poll also showed an increase in the number of persons who are undecided about voting or which party to vote for.

So, part of the interest in the Johnson poll was the fact that it was the first national poll to break the apparent deadlock in voter support among the major parties and, if correct, could change the dynamics of the campaign which, as I said before, could be career-ending for the losing party leader.

Interestingly, the intensity of the debate over the Johnson poll findings has only been matched in recent months by the controversy surrounding the prophecy of Bishop Phillip Phinn, senior pastor of Word of Life Ministries who prophesied that Prime Minister Portia Simpson Miller will win the next general elections which are constitutionally due by October.

Predictive power of polls

Behind the media frenzy and disclaimers among some politicians that polls are of no consequence, the reality is that they are taken seriously, and for good reasons, including the predictive power and influence of political polling in democratic politics built on the free choice of the electorate from among competing candidates, parties and ideologies.

It is generally agreed that political polling, as practised today, began only in 1944 in the presidential election in the United States when George Gallup established scientific survey methodology to measure voter intention by using a small, but scientifically selected, sample of voters to represent the opinion of the whole population.

Gallup used the technique to predict that Franklin D Roosevelt would defeat the Republican Alf Landon, who was the favourite among many others using other techniques, including interviewing millions of voters, to predict the outcome.

Gallup repeated his success a year later in the United Kingdom when he predicted that the British Labour Party would prevail over the Conservatives led, at the time, by the popular war-time and victorious Prime Minister Winston Churchill.

The later Professor Carl Stone earned an enviable reputation, still unmatched by his successors, for accurately predicting the outcome of several elections in Jamaica and indeed, his name still lives on in the Stone Poll Organisation which continues to poll for this newspaper and other entities.

Historically, and across different cultures and political environments, polls have had mixed results as predictors of election results. Since Carl Stone, the results among Jamaican pollsters have been mixed and will be the subject of a separate analysis.

One of the international examples of the pollsters getting it badly wrong was the 1992 UK general election in which polls leading up to the election and the 'exit polls' of persons after they actually voted showed a lead for the Labour Party but, in reality, the Conservatives were the big winners.

As happens in such cases, there is a lot of soul searching such as 'late swing', in which voters make up their minds at the last minute based on some new reality that has entered the picture, or that voters did not give their true preferences to the pollsters.

Predictive power apart, polling can also affect the electoral process and election outcome in several ways, including the so-called 'bandwagon' effect and the corresponding 'underdog' effect.

The first theory assumes that voters will be propelled to vote for the party or candidate perceived as the 'winner', while the 'underdog' theory assumes that voters will support the losing party or candidate out of sympathy.

While the evidence in support of these theories appears to be inconclusive, political parties whose main reason for being is to win and hold state power, do not take chances.

Those who find themselves in the lead go to great pains to warn their supporters against complacency. The worry is that the notion that the party has 'already' won may lead some supporters to stay away from the polls and could redouble the efforts of others to come out in support of
the underdog.

In addition, polls also impact on politicians in that the results may guide them to shift policy positions or priorities in line with the prevailing mood of the electorate. They help to define strategy.

Further, polls help to shape the media agenda and the way the media frame the issues and events that are covered in the news. Public opinion surveys in many countries over several decades have affirmed that the media are very good at telling us what to think about, that is, what to consider important.

If the media report frequently and prominently on poll stories showing particular results, voters will begin to attach great importance to such stories and use them to help make up their minds about voting.

Against that background, it is not surprising that the JLP has been seeking to undermine the poll result by attacking Johnson's credibility, pointing out that he is the PNP pollster and that he was wrong about the St Lucia elections in which his published polls showed the (then) governing St Lucia Labour Party ahead of the United Workers Party which was the eventual winner.

Interestingly, the JLP did not seek to discredit some of Johnson's other poll findings (reported in the Gleaner) showing the JLP ahead in some PNP-held constituencies, even where Simpson Miller was way ahead of Golding in terms of personal popularity, leadership ability and capacity to solve the key problems facing the country.

Johnson pointed out that he had done several polls for the SLP, including one just before the election showing the race tightening, but the client declined to publish those results. Furthermore, he said he was only wrong in one constituency in 2002.

While the firestorm about poll credibility raged, one underlying reality seemed to have escaped debate: The PNP is still below where it was a year ago, the JLP is exactly where it was a year ago. And all of this is after the parties have spent the better part of the year doing nothing but campaigning. Is anybody listening?

Claude Robinson is senior research fellow, Mona School of Business at UWI
kcr@cwjamaica.com