Regular polls gauging the popularity of the chief executive would be the last to be axed if the budget for the University of Hong Kong's public opinion programme (POP) were cut, its director has vowed.
Dr Robert Chung Ting-yiu has come under attack in recent weeks from government supporters and pro-Beijing media unhappy with the programme.
"If the POP is to face resource constraints and have to cut projects, the poll on the chief executive will be the last one to be axed," Chung told an HKU forum on academic freedom.
"A society's chief must be subject to scrutiny by scientific social research."
Earlier this month, Peter Lee Ka-kit, son of property tycoon Lee Shau-kee, complained about Chung's polls during a session in Beijing attended by National People's Congress chairman Zhang Dejiang .
Lee, a Standing Committee member of the Chinese People's Political Consultative Conference, suggested pro-establishment organisations finance and commission their own surveys.
Central Policy Unit chief Shiu Sin-por and Executive Council member Cheung Chi-kong joined the chorus questioning Chung's polling methodology.
"Some businessmen think money can solve problems … this is unacceptable," said Chung, without naming Lee. "All polls have to be conducted scientifically.
"If the data yielded scientifically is unsatisfactory politically, leading to surveys conducted with an unscientific approach, the issue is no longer academic but political."
HKU's programme asks respondents to rate the extent to which they support Chief Executive Leung Chun-ying on a scale of 0 to 100 and whether they approve or disapprove of his performance.
Shiu, a key government adviser, argued that the 0-100 scale left too much room for interpretation and that Leung's score, based on the average of responses, could be distorted by the extremely low marks possible using such a scale.
"Asking respondents to assign a [0-100] mark to politicians is a common polling technique," Chung said. "Foreign polls also ask … if respondents would vote for a certain politician if an election is to be held the next day." He suggested that technique is better suited to democratic countries.
Chung said the methodology for every poll conducted by the HKU programme since 1991 was fair and freely available to examine.
Extreme marks - both positive and negative - tended to cancel each other out, he said, meaning they had little effect on the final mark given to Leung.