Correcting data ‘wobble’
by Peter Wickham
The last NATION/CADRES poll of public political opinion in Barbados was conducted between February 8 and 11, 2013 and reported on in the Sunday Sun on February 17. In that report, CADRES noted that “the data presented a major departure from earlier trends and as such could be colloquially assessed as a “wobble” or a “bump” in the campaign of one party which, it was argued, could have been either temporary or permanent.
As such a second report was promised, reflecting public opinion in the second week of the campaign (February 15 to 18); CADRES now presents the data which was tabulated within the last 24 hours.
Since the last report is less than one week old, it is perhaps not necessary to repeat the methodological outlines presented in that report. It is nonetheless necessary to note that the Polling Divisions (PDs) used in this most recent poll are no different to those used in the first February poll. In this instance, however, CADRES reassigned interviewers to reduce the likelihood that respondents would be re-interviewed.
Regarding considerations, the point needs to be made that this second poll captured a fully evolved election campaign and the launch of both manifestos, which were not available when the last poll was conducted. As such, this poll in our opinion presents a far better podium from which we can assess the likely outcome of the 2013 general election.
Interestingly enough, the only noteworthy consideration would be the controversy created by the publication of the first poll and the significantly reduced swing against the DLP which was reported. Naturally, CADRES is unable to assess any impact on respondents which the knowledge of a second, new poll may have had; however we did note the uncharacteristically low level of “non-response” in this instance.
In the first February poll the non-response was 33 per cent, which was the lowest we ever recorded in a CADRES poll. On this occasion; however the non-response was 24 per cent and implies that the public took a greater interest in this exercise.
Satisfaction with Government
In this survey the public’s level of satisfaction with Government was measured by reference to two specific questions which are presented in Figure 01. The data indicates that a majority of Barbadians are not satisfied that “we” are on the right track, and a similarly marginal majority believe that it is time for a change of government.
These two responses indicate a high level of indecision on the part of the electorate, which is clearly divided about the extent to which the Government should be changed, and whether it has performed well. It is interesting to note comparisons with the first February survey in which three per cent less persons believed that is necessary to change the Government, but a similar quantity (41 per cent) expressed the view that a change of Government was not required.
This growth in support for the “change” view came from the undecided group over the past week. The response to the “Right Track/Wrong Track” question also suggests some similarity of thought over the past week since two per cent more were convinced that Barbados was on the “Wrong Track”, while one per cent more said “Right Track” with the uncommitted cohort also shrinking on this occasion.
On this occasion CADRES used similar techniques to assess leadership and here comparisons are also presented for the convenience of readers of this report, who might also be familiar with the previous one. Figure 02 presents the responses to the “Yes” or “No” question which sought to establish whether persons approved of the Prime Minister or not.
It is interesting to note that a majority of respondents did not approve of either Leader of the Opposition Owen Arthur or Prime Minister Freundel Stuart, although in both instances the difference in quantities of persons who approve and disapprove are marginal and well within the margin of error of the survey.
On this occasion, Stuart’s approval rating has risen from 39 per cent to 41 per cent while Arthur’s has remained constant at exactly 37 per cent. The disapproval rates for both leaders have increased marginally with Stuart’s moving from 40 per cent to 41 per cent and Arthur’s moving from 38 per cent to 41 per cent. In both instances the movement came from the uncommitted cohort which is somewhat smaller for both leaders.
Table 01 presents another perspective on the issue of leadership which is in this instance assessed from the perspective of a numerical rating on a scale ranging from “1” to “10” with “1” relating to the lowest or worst performance and “10” relating to the highest or best performance.
These tables are presented with adjacent scores for the month of February for the convenience of readers. This demonstrates that the leadership ratings for Stuart have fallen marginally while those for Arthur have risen marginally within the intervening week and in both instances the changes are statistically insignificant.
Table 01: Leadership Ratings
Table 02 presents information on the competitive leadership preference question and gives the perspective of committed and uncommitted voters. In this instance also all movements within the week have been within the margin of error of the poll, but on the outer limits of that margin.
Hence it can be seen that Arthur moves back to the top of the list with 37 per cent (from 33 per cent) while Stuart now moves slightly downward with 32 per cent (from 36 per cent). Although the list appears to present an order of precedence, it should be noted that neither leader has moved significantly within this time period and both are still in a statistical “dead heat”.
Table 02: Leadership Preferences
There is a similar consistency with respect to the leadership preferences when viewed from the perspective of party support, which is presented in figure 03. It should be noted that the “Uncertain Voters” still favour Stuart most, while both leaders remain popular within their respective parties.
Figure 3: Leadership Preferences
Figures 04 and 05 speak to party support and the related swing analysis, and these are presumably self-explanatory. On this occasion there have been significant changes in the party support statistics and the most notable change has been the sharp reduction in the quantity of persons who were uncertain about their voting intention or refused to answer this question. This is presumably consistent with the fact that the election is now days away and persons should by now be better able to decide for whom they will vote.
The BLP has clearly benefited most from the past week of the campaign since their support has moved from 34 per cent to 42 per cent, which is a statically significant shift, while the DLP’s support moved from 32 per cent to 34 per cent, which it should be noted is well within the margin of error. This means that DLP support has remained fundamentally unchanged in the last week, while the BLP has been able to grow its support among the “Uncertain Voters”.
The foregoing party support statistics facilitate the swing analysis presented in figure 05 and project a swing towards the BLP which CADRES has computed at seven per cent, with a swing away from the DLP of six per cent. This uneven swing pattern implies that the independent candidates can cumulatively expect to receive one per cent of the national vote, which would directly impact on the DLP’s support base. Table 03 updates the likely swing impact on the DLP Government’s life cycle.
This report should be read in the context of the first February report which clearly spoke to a political campaign in evolution, while this one appears to speak to a more mature political campaign.
The main noteworthy comparative observation is that the BLP was leading in the polls one week ago and continues to lead on this occasion and appears to have widened its lead.
It is also clear to CADRES that this will be one of the most marginal and indeed competitive campaigns ever endured in this nation’s history.
This marginality is evidenced by the fact that a shift of four per cent to five per cent in support over the last week has resulted in a presumptive loss of five more seats for the DLP.
CADRES has repeatedly made the point that this DLP Government appears strong in terms of numbers but is in reality comparatively weaker than any other first term government that the DLP has had previously. As such the slightest deterioration in support will have a devastating impact on seats.
It is against this background that CADRES has argued that the party better organised on Election Day in all of the marginal constituencies has what would be called a better “fighting chance” to win those “seats”.
Notwithstanding the possibility of DLP “surprises” in some instances, CADRES is at this time of the opinion that the BLP will win the 2013 election on Thursday with between 17 and 20 seats, while we believe that the DLP will occupy the Opposition benches with between 10 and 13 seats.
The appended Table 04 presents a colour coded chart which identifies the strengths and weakness of each seat in contest based on an identification of factors identified by CADRES as significant in the election process. To the right of each constituency is the CADRES assessment and the colour represents our opinion on the likely outcome while “green” signifies marginality.