30 August 2006

Playing the numbers

[This is the first of a three part series looking at public opinion polling and provincial government media relations. In the first installment, we'll take a look at the broad issue of news releases output and content. In the second - The media and the message - we'll look at the other media relations activity for the provincial government. In the last installment, we'll pull the whole thing together and take a look at why government does what it does.]

Over the past decade, political watchers in Newfoundland and Labrador have come to accept the idea that the government party will adjust its media activity to coincide with polling periods.

The theory behind it is simple. Since there is only one pollster who collects data at regular, known and predictable periods, a flood of "good news" issued at the right time can reinforce positive feelings toward government or, at least, neutralize or counteract negative feelings.

Marketing researchers will do this routinely. In the wake of a flurry of advertising, they will conduct a short research program to see if the advertising had an impact on a randomly selected sample of the target audience. While marketers may have to establish a baseline - a survey or other research done before the advertising hits - political parties have the advantage of having a rolling baseline of regular polling done either privately for the party itself or publicly.

In Newfoundland and Labrador, only Corporate Research Associates (CRA) conducts quarterly polling. CRA's omnibus poll puts questions to random samples of the provincial population, on behalf of paying clients - including the provincial government - and includes a series of three standard questions on political attitudes.

Lately, CRA has taken to conducting two separate omnibus polls within the span of two to three weeks. The three political questions are CRA's own and are released shortly after the data is collected and analyzed. For them it is a simple marketing device to keep CRA's polling profile high. CRA's news releases are invariably picked up by media outlets in the province and Don Mills, CRA's president is readily available to provide some commentary and analysis.

There's absolutely nothing unusual about CRA's actions: it's just good business. Their questions are standard and, all other things being equal, they should pick up attitudes reliably.

The idea of politicians trying to influence polls got a special highlight during the most recent CRA data collection period (started 12 Aug, for about two weeks). The Liberal opposition issued a news release alleging the Progressive Conservatives were trying to goose the numbers. The Liberals should know since the practice of playing to the polls started under Brian Tobin.

For their part, the Progressive Conservatives dutifully denied any such activity. CBC reported that the Premier's Office stated that "the number of news releases issued recently is consistent with general distribution."

Don Mills defended his research work to CBC:
"It would be, in any case, a very difficult task, because people form impressions over a course or length of time on many issues," Mills said.

"[That's] not to say they couldn't be influenced by a string of good news announcements but, frankly, I doubt it."
Peaks match polling periods 11 out of 12 times

In order to examine the idea of government trying to issue a string of well-timed, "good news" releases, let's first take a look at the volume of releases coming from the provincial government.

If the volume was consistent with general distribution - whatever that is - we'd expect to see a fairly constant number of releases issued. There might be some identifiable peaks and troughs like at budget time, around Christmas and during the summer. Alternatively, we might see some random peaks and troughs driven solely by the operation of government itself. Again, we might see some identifiable peaks at times when there is a lot of news - budget time - or valleys when people are off for Christmas holidays.

CRA polls at predictable times each quarter of the year, usually varying the start and end times of its data collection by a day or two year to year. Government subscribes to CRA's omnibus so it would know well in advance when Don Mills' people will be hitting the phones. Therefore, if we have regular polling and random peaks and troughs of news release volume, we wouldn't expect to see much of a correlation between the polling periods and the news release output.

Figure 1, below, shows the number of news releases issued each week beginning in the fourth quarter of 2003 and continuing to the first part of August 2006 before recent polling started. It includes all types of news releases and media advisories on the provincial government's website.

Fig. 1. Government news releases, total by week
4th quarter 2003 to current. Green lines mark CRA polling periods.

In 11 of the 12 polling periods since the last quarter of 2003, government news release output peaks either immediately before or during the polling. In the one exception, the volume increases steadily during the period and peaks immediately after polling ended.

On the face of it, this suggests that government news releases are not being issued at random. They are not "consistent with general distribution". Rather, there is a fairly obvious connection between output and polling. Even more interesting is the fact that more often than not the peak output is immediately before polling starts.

If someone was trying to influence polling, the best time to peak "good news" would be immediately before data collection started. That way, there is the greatest likelihood that the good news will be fresh in public minds when the telephone calls start.

A little good news

Volume of news releases alone isn't proof of an effort to influence polling results, though. The results we've found so far just give ius a clue something is up that isn't random.

The next task is a content analysis - taking a look at what the releases say. The past three weeks are instructive. For the period 31 July to 06 August, government issued 27 news releases, followed by 34 the next week. During the two weeks when CRA was actively collecting data, government issued 27 releases each week.

August is normally a time when people take vacations. Yet, in the first week of the month, the provincial government announced $1.5 million in funding for the College of the North Atlantic; an expansion of family law services; changes to social assistance funding to give clients more money; almost $75, 00 in funding for the Association for New Canadians; $200,000 for silviculture projects; and, upgrading of a health care clinic in Baie d'Espoir, among other positive initiatives.

In the week immediately before polling started, the provincial government announced progress on funding for the Trans-Labrador highway; $7.0 million for a new school in Torbay; $78, 000 in violence prevention for aboriginal communities; $7500 and a speech from a cabinet to support a conference on rural development; $82 million in federal money for municipalities; $137, 000 for renovations to public housing; and, a $6.3 million sports centre in St. John's.

The provincial government also issued a news release on the proposed dismantling of the Stephenville mill. This story broke in a routine public advisory of projects needing environmental review.

Curiously, the very first release of the following week announced a successful resolution to a problem in Stephenville where used tires have been piling up as part of a problem-plagued provincial government program dating back to the Liberals. Even though government won't know until October if the solution even exists, they issued the release anyway.

Finance minister Loyola Sullivan issued a news release announcing that the provincial government's finances were in even better shape than previously announced. One of the many odd things about this release was that it contained information normally released in November once it had been reviewed by the Auditor General. Sullivan claimed he was announcing the news early, unauditted and to the general public so that government officials could have the latest information in managing their budgets.

The general pattern continued throughout the week, with the first release on Friday announcing $1.45 million for economic development in Stephenville. In the next week, the general trend continued, but this time included $5.5 million in funding for College of the North Atlantic in Goose Bay; increased fees for dentists (framed as improved dental care for children); and, new police officers in the crime intelligence field.


The provincial government news release output coincides with Corporate Research Associates' polling periods too frequently for it to be accidental.

The news release content tends toward "good news" moreso than one would expect to find if release content was random or than one would expect in a period like August when most people tend to be on vacation.

Even when bad news does appear - such as the routine Stephenville environmental notice - it was quickly counteracted with not one but two positive announcements.

There's no question it happens.

In the next installment, we'll look at why.