Soaring past the skeptics

by Stephen Fenerty

Ski jumps. Luge tracks. Hockey rinks. Public opinion surveys. Text mining software. That's right. The city of Turin knows there's more to hosting the 2006 Winter Olympics than building the facilities. There's also the challenge of understanding your citizens and winning their enthusiasm.

In a project developed by the city of Turin, Italy, and Turin University, research agency Metis Ricerche is using SAS to monitor citizens' attitudes to the 2006 Winter Olympics due to be held in Italy's Alpine region.

The challenge is to understand if residents are in favor of the games and the opportunities the Olympics will bring. In order to gain that necessary understanding, text mining was used to analyze survey results. Text mining is a process consisting of various steps, such as finding terms and reducing them to their root form (stemming and parsing) and applying analytical procedures such as clustering and classification.

"Since 2001, we've been tracking citizens' attitudes," explains Dr. Gianluca Bo, the project manager at Metis Ricerche. "Our data comes from Internet-based surveys carried out using SAS, and text mining and further analyses are then performed also with SAS. In fact, we've been using SAS in our work since 1980."

The results enable the city's public administration to understand how citizens view their city and illustrate their "acceptance" or "refusal" of the Olympics. Five surveys will be conducted in total; the first, in 2002, covered 1,000 people in the city and 400 in the surrounding valleys. The second, in 2003, involved 1,400 respondents and focused on an "open" question about perceptions of Turin itself. "The results are very important, especially in targeting campaigns to communicate the benefits of the games and generate the widest popular support," says Bo. "Previous studies have shown that a deep-rooted involvement from citizens of Olympic host cities is key in ensuring a satisfactory and enduring outcome, not least in economic regeneration.

"The first stage is using data mining to identify our target group, based on answers to closed questions that explore attitudes to the games. Second, we use text mining to explore views on Turin itself. The goal here is to better describe the target group and understand the features and channels that would maximize the impact of an information campaign."

Analyzing 'closed' questions
Citizens were asked questions involving their degree of approval for the choice of Turin, their level of pride, whether the games will bring personal advantages, and whether they will stimulate economic recovery. In 2002 and 2003, more than 95 percent of people were "quite" or "very" favorable to the Olympics, although more detailed analysis showed variations in the overall level of agreement.

"Delving deeper, we've singled out particular people who need more and specific information to engender greater enthusiasm," says Bo. "Using SAS for text mining, we've described our target in a more precise way. And we've discovered the type of content such a group will be particularly sensitive to, and the channels through which they'd like to receive information. These factors provide the right ingredients for successful information campaigns."

Citizens were initially classified as "well informed," "not informed," "exacting" and "uninformed" with regard to the games. They were also classified based on interest shown in other events held in Turin such as the International Book Fair; this enabled groups to be classified as "fan," "participant," "indifferent" or "hostile."

"Correspondence analyses showed relationships between these two typologies," says Bo. Using cluster analysis, Metis Ricerche mapped the different groups. Results showed that 5 percent of respondents had an "adverse" attitude to the games, while respondents who were deemed "enthusiastic" made up 38 percent. Those respondents in the "critical" group (people expressing a favorable attitude tempered with some criticism) were the main target for communications.

Text mining: Analyzing 'open' questions
"The question, 'Thinking about Turin, what first comes to mind?' was asked before any Olympic questions, and so was conditioning-free. We used text mining to analyze the results obtained earlier," says Bo. Analyzing these responses enabled Metis Ricerche to further understand differences between the groups defined as either adverse, enthusiastic or critical. After text parsing, 104 words were extracted from a start list of 202. The most frequently used words included: chaos, beautiful, building sites, public works, Olympic games, Fiat, big, traffic and dynamism. Using multiple correspondence analyses, these words were attached to the three main groups already identified, with the words overlaid onto the earlier cluster analysis.

"This showed how people in the adverse area, in terms of attitude to the games, also had a mainly negative view of the city," says Bo. This group included words such as: disaster, pollution, negative and crisis. "This confirmed this as the area of total refusal: in Torinese dialect, we'd say these are the 'bastian contrari' – people who will disagree in principle. It follows they are likely to maintain their convictions, so any attempt to gain buy-in would mean wasted time and resources." Those categorized as enthusiastic (in complete agreement with the Olympics) also showed a positive image of Turin, using words such as pride, transformation, willingness, enjoyable and evolution.

The most value from this SAS analysis comes with the critical group. "Here, people complained about problems of everyday life, such as transport, building sites and crime," says Bo. "This is the best target for information campaigns. More in-depth analysis of the attitude of this group towards the games means it was also possible to identify the most appropriate content to include in campaigns, to convince them of the value of the Olympics."

Changing attitudes
Analysis showed that the critical group was predominantly male, over 51 years of age, with a low level of education, and living in the Alpine region where skiing competitions would take place. "When we mapped percentage responses to questions about advantages arising from the games, we found the critical group's answers had the same pattern as the enthusiastic group, albeit at a lower level." These advantages included infrastructure, sports facilities, international visibility, tourism, employment and personal profit. Within the critical and enthusiastic groups, results showed a marked closeness in two particular areas – employment and personal profit - which suggested they were critical issues for the target group. "This led us to recommend that an information campaign was needed to highlight all the positive aspects associated with the games and especially job opportunities," says Bo.

The same approach was applied to possible problems caused by the games such as traffic, crowds, public costs, private costs, environmental damage and corruption. Here, results from the adverse group showed a completely different pattern to the enthusiastic and critical groups, which lay very close together. However, these two patterns showed small differences in areas such as traffic and public costs, as well as larger differences in areas including unnecessary facilities and corruption. "The information campaign should address these themes," says Bo. "Moreover, association between different groups and looking at favored sources of information is also guiding local administrators in the right channels and communication methods to use."

With the process continuing in the run-up to the games, it is hoped that the critical group can be moved closer to the enthusiastic, with knock-on effects in the ultimate success of the 2006 games and in the longer-term economic benefits for Turin and its surrounding region.

Bio: Stephen Fenerty, a freelance writer based in the UK, has been writing about SAS for more than 10 years.