A year after the Fairfax papers ended a four-decade partnership with Nielsen, Australian opinion polling is set to undergo another upheaval as News Corporation pulls the plug on the market research operation behind the biggest name in the game. The Australian announced yesterday that the Newspoll organisation, which has provided it with its polling since 1985, will be wound up, with the loss of 26 jobs. Polling will continue to appear in the paper under the Newspoll banner, but what was once a company will now be merely a brand. As of July, the actual work will be conducted by Galaxy Research, which was established in 2004 by Newspoll alumnus David Briggs and has long conducted polling for the News Corporation tabloids. Newspoll has played a substantial role in promoting The Australian's cherished status as an agenda-setter, which has been achieved through a regular fortnightly publication schedule, assiduous promotion on the part of the paper, and a formidable record of accuracy in pre-election polling. Garden-variety political junkies have long been accustomed to waiting up every second Monday to see which way its pendulum swings, however much they might hate themselves for doing so. Higher up the tree, there has been more than one occasion where a political leader's future appeared to hang on the vagaries of its latest result. Writing in Quarterly Essay shortly after the national fiasco that was the 2010 federal election, George Megalogenis went so far as to complain of a "dizzying turnover of political leadership talent since Newspoll went fortnightly" -- something that happened in 1992 -- which he held partly responsible for the demise of an era of reform that had run from the Hawke government to the early Howard years. But for all the fuss surrounding it, it has always been an open question as to whether Newspoll really offers anything its rivals don't, particularly in recent times. Certainly its election eve polls have almost invariably scrubbed up well the morning after, but the pollster goes to special efforts at these time, most visibly with respect to sample size. A particular source of controversy of late has been its exclusive targeting of landline phones, despite the intuitively compelling argument that this no longer obtains it a representative sample, particularly among younger voters. Pollsters' hesitancy to include mobile phones in their surveys reflects their need to have some idea of a prospective respondent's location. This can be achieved at no cost with landlines, as the phone numbers come with geographic prefixes, but to do the same thing with mobiles requires expensive licensing from directory services. Since the interviewer-administered polling conducted by Newspoll was quite expensive enough in any case, this was hard to justify when post-election analysis continued to show that landline polling was hitting the mark. However, it has appeared on recent form that Newspoll has been paying a price in volatility between one poll and the next, even if there hasn't been particular bias evident in favour of one party or the other. Since the Abbott government came to power, the average fortnightly movement on two-party preferred in Newspoll has been 1.9%, compared with 1.4% from Morgan, which uses a mix of door-to-door and SMS polling, and 0.9% from Essential Research's stable-to-a-fault online series. A possible explanation is that Newspoll has been compelled to inflate the value of a limited number of responses from the 18-to-34 cohort, thereby increasing the effective margin of error. For close observers of the pollsters' form, the better performer of late has actually been Galaxy Research, which has expanded the methods used in its national polling to encompass online surveying as well as a component of mobile phones. While its polling has been considerably less frequent than Newspoll's, the evidence suggests it has been steadily achieving results well in line aggregated polling trends. However, Galaxy's recruitment by The Australian does not entirely guarantee that we are soon to be treated to polling of this calibre on a fortnightly basis. In a pointer to the cost pressures that led to yesterday's announcement, Briggs says the telephone component will be conducted through "robopolling", as the use of live interviewers would be "prohibitively expensive". Given the high non-response rates, robopolling can only be regarded as a second-best option. Galaxy's robopolling has nonetheless performed well in the difficult endeavour of targeting individual electorates, and the national polling of dedicated robopolling firm ReachTEL has produced consistently credible results, despite some wayward performances in its electorate polls. Even so, it remains to be established whether the cost-cutting can be achieved without compromising Newspoll's reputation as the gold standard in Australian opinion polling.