Ibrahim Suffian says the main determinant of whether a poll is accurate depends on those represented
KUALA LUMPUR 2011/09/01: The Merdeka Center for Opinion Research's recent poll on Prime Minister Datuk Seri Najib Razak's popularity was 97 per cent accurate, said programme director Ibrahim Suffian.
He said the poll was conducted according to standard procedures.
"Surveys of this nature normally engage between 1,000 and 1,200 respondents, as this sample size allowed a margin of error of only about three per cent."
He said the main determinant of whether a poll was accurate depended on those represented, and not the size of the sample.
"This means that a 1,000-respondent poll that is representative of the population may be more accurate than a 50,000-respondent survey that is skewed or unrepresentative."
The poll was conducted through telephone where random numbers were dialled to engage a wide demographic.
"They were randomly selected at different stages to ensure randomness at the level of state of residence, ethnicity, gender and age groups," he told the New Straits Times.
Also, he said, it was essential for surveys to be conducted in the shortest time to convey accurate feedback from the public.
"Events occurring over the same time a poll is conducted would influence the process and results.
"In the case of this poll, the results provide a snapshot of public opinion during the period when the poll was carried out."