SOCIAL media surveys are not representative of voters, because Twitter users in Malaysia are limited in number, said the Merdeka Centre for Opinion Research.

“Comparisons with social media driven online surveys may result in misleading conclusions due to the difficulty in controlling self-selection bias, which means certain groups of people may be motivated to respond compared to others.

“There is also the question as to whether controls for representativeness of the sample were put in place,” said the pollsters in a statement posted on Facebook today.

Furthermore, Twitter users in Malaysia only account for less than 5% of the population, the centre said.

The centre was responding to criticisms that its latest survey on Prime Minister Muhyiddin Yassin was unrealistic.

Despite recent complaints about the government, the centre, in its latest survey, said that Muhyiddin’s approval ratings are stable at 67%.

According to the centre, its surveys were conducted over the telephone, so allowing it to control the representation of the respondents.

“In the context of this survey, respondents matching the age, gender and ethnic criteria – based on the profile of the Election Commission’s electoral roll – were selected and interviewed from every parliamentary constituency in Peninsular Malaysia.

“Although all surveys are not free from bias, such as social desirability, our survey places great rigour on ensuring reliability by controlling various variables such as age groups and location, that is urban v rural, along with socio-economic background, so as to have as representative a profile as possible of the electorate,” said the centre.

Although the survey was just on Muhyiddin’s popularity, the centre said, the public should not be confused or transposed on other questions such as for which party people should vote. – April 26, 2021.