[ad_1]
‘All established margins of error are assumed to have a confidence level of ± 95%, which means that sampling errors are accepted in five out of every hundred cases. We mentioned that the deviations are calculated at the percentage level, respectively to the score obtained by each candidate, person or party and not to the distance between the percentages, as wrongly invoked in the media, although the margin of error of difference is calculated in a different. Such an approach means that instead of talking, for example, of a deviation of ± 2-3%, exaggerated predictions are invoked with differences of up to 10 percent, although the deviations that matter are those that are outside the margin of error for each candidate, or outside the margin of error calculated at the level of gap or distance, a procedure that is sometimes used in the US where only two candidates count, while in Romania they appear second in a hierarchy with several candidates “, shown in a CURS-Avangarde point of view.
Thus, the poll conducted at the polls had an assumed margin of error of 2.6%, and in the case of Nicușor Dan a deviation of 2.4%. Thus, decreasing from 47.8% to five percent, the result is finally recorded by the BEC, respectively 42.8%.
“The winner of the office of mayor general, mayor of Craiova and mayor of Constanța was estimated correctly. The mayor general was estimated with a significant deviation, but the mayors of Constanța and Craiova were estimated more accurately, practically, even taking into account the score. Invoking the distance between candidates is irrelevant if the BEC results and the established margin of error are not taken into account. In fact, practically, the distance between 1st and 2nd place is reduced to 2.4%, if the 5 percent of the 1st place, according to data from the BEC and 2.6 percent of the margin of error ”, shows the cited source.
A significant deviation, on the margin of error, also appeared in section 1, where the exit poll credited Clotilde Armand with 48%. In this case, the deviation was 4.5%.
CURS-Avangarde states that this was due to the high percentage of refusals to answer the questionnaires, of more than 25% of all respondents, but also to a “spiral of silence”.
“We do not rule out the operation of a certain spiral of silence that would put the PSD candidates at a disadvantage, knowing that this party was attacked with more force in the media, so some of its voters do not show who they voted with, especially in the sector 1, with a more intellectualized electorate (50% with higher education) and older. This explains, at least in part, the underestimation of the possibility of mayor for the PSD candidate for mayor of sector 1 and the overestimation of USR candidates -Plus, young people responding with request to the interviewers ”, affirms the cited source.
The conclusion of the two institutes is that the results of the exit polls confirm “the BEC data especially in the case of performance indicators or criteria related to the identification of the winning candidate, the establishment of the order or hierarchy and integrating, in most cases, the maximum margins of error established “.
Read here the full explanations of CURS-Avangarde
[ad_2]
