Soccernomics (28 page)

The viewing figure Alavy uses for any given game or event is the

“average program audience.” That is the average number of viewers during the entire event, rather than the peak figure, which would typically be one and a half times higher, or the “reach,” which includes anyone who caught at least three minutes.

Alavy reports the viewing figure for any game as a percentage of all the people in each country who live in households with TV sets. In most European nations almost every household has owned a set since the 1970s. However, that is not true in less developed nations. Only about a quarter of India’s population of 1.1 billion, for instance, lives in households with TVs.

Alavy’s collection is a treasure chest: thousands of TV ratings covering hundreds of games between national teams screened in fifty-four countries since 1998. However, there is an obvious problem: Germans probably watch a lot of games involving Germany, and Germany has a pretty good record; how to compare the viewing figures of the successful Germans with the not-quite-so-successful English? Indeed, how to compare German viewing of Euro 2008, where their team made the final, to viewing by the English, whose team didn’t even qualify? Or to cite another complication: a casual fan who watches only his own national team is surely not to be compared to the fanatic who glues herself T H E C O U N T R Y T H AT L O V E S S O C C E R M O S T

193

to the sofa for every match of the tournament. Going one step further, viewers in Malaysia or Colombia who follow the Euro must surely be considered to have reached the acme of perfection in fandom.

There are other problems. The World Cup in Japan and Korea had disastrous TV ratings in Europe, because most games were played in the European late morning or at lunchtime. That is why FIFA was so keen to keep the World Cup of 2010 in South Africa despite all the problems there: Johannesburg is usually just one hour ahead of central European time, making it ideal for European viewers. It’s less conve -

nient for Americans. In all the agonizing over the refusal of Americans to watch soccer on TV (as opposed to playing it, which American children do in great numbers), an obvious but overlooked explanation is that most of the best games are played in Europe at times when most Americans are working.

We also need to control for the significance of each match. Obviously, the final will draw more viewers than a group match, but how many more?

In short, no two of our thousands of ratings are strictly comparable.

Happily, there is a way to control for the differences, and so to strip down to the essence of TV fandom. We can expect viewing levels in every country to rise for the final of a tournament. We can also expect an uplift when the country’s own national team plays, or when a game is shown in prime time. Once again, we need to wheel out the technique of multiple regression. A reminder: multiple regression is the mathematical formula for finding the closest statistical fit between one thing (in this case TV viewing figures) and any other collection of things (here time of match, who is playing, and so forth). When you have many observations—like the thousands we have here—the regression technique is extremely powerful. Not only does it give you a precise figure for the influence of one factor on another, but it can also tell you how reliable that statistical estimate is. For example, the regression can show that playing a game in prime time will add, say, 4 percent to the size of the audience, and it can also say that the probability that this estimate is mistaken is less than 1 percent.

194

Running a regression will allow us to isolate each contributory factor—such as kickoff time—and measure its impact on attendance.

Not only will this tell us a great deal about what attracts viewers to tele-vized soccer, but the final figure that we are left with after these regressions consists of two parts. One part is what is called a “random error”: that which we cannot explain because we know too little about what happened on the day. The other part, though, is a “fixed effect”: an estimate of the viewing share that a soccer match will achieve in a given country, stripped of all other characteristics like kickoff time, identity of teams, stage in the tournament, and so on. The “fixed effect” is what we are after. The size of the fixed effect in each country will tell us how eager that country is to watch international soccer on TV.

The method works like this: Our data show that the “fixed effect”

for the UK is 6.98. This means that just under 7 percent of all British households with TV sets will watch a World Cup or European championship game, regardless of its characteristics. Seven percent is the proportion of Brits that we can expect to have watched, say, Tunisia versus Saudi Arabia at the World Cup of 2006. However, if the game is played in the middle of prime time (meaning a kickoff between 7 p.m. and 8:45 p.m.), the total of British viewers rises by about 4.3 percent. If the game is a semifinal, viewing goes up another 5.6 percent.

So our regressions would lead us to estimate the British audience for each World Cup semifinal in 2006 at 16.9 percent: the core audience of 7 percent would have risen by 4.3 percent because each game kicked off in prime time, and by another 5.6 percent because it was a semifinal.

The actual viewing figures for each game were within 2 percentage points of our estimates.

By no means perfect, but over the hundreds of games seen in fifty-four countries, the errors are relatively small. The average rating figure for each game taken across all fifty-four countries is 8.6 percent. Again: that is the proportion of households with TV sets that watched the game. For more than three-quarters of our estimates, the error was smaller than half this number. So while we cannot place too much weight on any single estimate for a particular match seen in a specific country, overall the regres-T H E C O U N T R Y T H AT L O V E S S O C C E R M O S T

195

sion summarizes the data pretty well. The regression captures just over two-thirds of the variation in viewing figures. The remaining third of the variation is due to unexplained random factors.

Before we ask which countries are the biggest soccer couch potatoes, let’s pick apart the other regression effects for what they tell us about fandom. We found that only two time-of-day effects mattered: games played during “sleep time” (midnight to 5:40 a.m.) reduced the average audience by 3.6 percentage points, while prime time added about 4 percentage points depending on the exact time.

The significance of a match was crucial. In the group stages, if one or both teams playing in the third match were already knocked out, the TV rating fell. If both teams were out, the game lost 4.7 percentage points of its rating—a giant loss, since a rating of 4.7 percent is usually enough to keep a series on TV in the first place.

Yet when one or both teams in a third group game had nothing much to play for because they had already qualified for the next round, viewing figures barely fell. Soccer is soap opera. We watch it because of the story that unfolds after the game, not just because of the game itself.

As long as a team is still in the tournament, its story continues.

Viewing figures rise steadily for each round of a tournament. The World Cup’s round of sixteen adds 1.4 percent to the audience, while quarter-finals add 2.5 percent, semifinals 5.6 percent, and the final itself 10.1 percent. Even the match for third place, often mocked as mean-ingless, raises audiences by 4.9 percent—only slightly less than the semifinal effect. The UEFA dropped the third place play-off from the European championship in 1984 because the game was thought to be boring. Perhaps the UEFA should reconsider.

Besides time and significance of the match, a third effect sometimes matters: who is playing. Brazil is everybody’s “second team.” Despite getting duller since 1970, it still commands a viewing premium of 2.2

percent whenever it plays. Weirdly, only one other team consistently enhanced viewing figures across all fifty-four nations in Alavy’s database: not Italy or Holland or Argentina but, yes, England. Its games boost global TV ratings by 1.4 percent.

196

To sum up, imagine a World Cup final played in mid–prime time between England and Brazil. This would add 10.1 (the final effect) + 4

(prime time effect) + 1.4 (England effect) + 2.2 (Brazil effect) = 17.7

percent to average viewership expected in any country.

There is one last factor to consider: nationalism. Its effect is enormous. For any game, for any country, broadcasting the national team added an average of 17.9 percent to the audience. It appears that for the average person on earth, simply watching his own national team in a humdrum group match is more attractive than watching England versus Brazil in the World Cup final. People who love soccer are vastly outnumbered by nationalists who tune in only for “our boys.”

If England did play Brazil in a World Cup final in prime time, how many Britons would watch? Our model would predict 7 percent (the core audience) + 17.9 (the nationalist effect) + 16.3 (the prime time, final, and “Brazil” effects) = 41.2 percent. That would equal about 25

million viewers. In fact, this is almost certainly an underestimate, partly because it wouldn’t count the people watching in pubs and other public places. Predictions about events that are both extremely rare and important are inevitably subject to wide margins of error. The same problem afflicts forecasting in financial markets: no statistical model can accurately predict a crash in the stock market. We have a good idea of how regular “small” factors like the time of day will affect viewing figures. We are much less precise about the effect of big one-shot deals like the country’s own national team playing a final.

Now we can finally open the last door: after correcting for all incidental factors, which country has the highest TV ratings for World Cups and European championships?

And the winner is: Croatia! Alavy notes that Croat passion goes beyond soccer. “They also have very high ratings in other major sporting events,” he says. For instance, when Croatia and Spain met in the world handball final in 2005, average ratings in Croatia were nearly four times higher than in Spain. This enthusiasm may stem from having recently fought for independence in a nationalist war, or it may just be because Croatia has excellent sports teams.

T H E C O U N T R Y T H AT L O V E S S O C C E R M O S T

197

F I G U R E 9 . 4
The biggest soccer couch potatoes

Core TV viewing rating for soccer as

Team

percentage of households with TV sets

Croatia

12.4

Norway

11.9

Netherlands

11.5

Uruguay

10.7

Denmark

9.1

Serbia

9.0

Ecuador

9.0

Brazil

8.8

Korea

8.7

Sweden

8.6

Germany

8.6

Hungary

7.9

Italy

7.8

Argentina

7.4

Indonesia

7.3

Singapore

7.1

Romania

7.1

UK

7.0

Lebanon

7.0

The other European countries that watch most international soccer—

the Norwegians, Dutch, Danes, and Serbs—also have relatively small populations and relatively big soccer reputations. Quite likely, small nations score high in our rankings because they tend to be more interested in what is going on outside their borders than big nations like France or Mexico. Alavy notes, “To have high average ratings, it’s all about having a high tendency to watch when your local heroes are
not
playing.”

Britain, for all its claims of soccer obsession, is stuck in what Alavy calls “a kind of midtable mediocrity,” sandwiched in eighteenth place between Romania and Lebanon. Germany has the most eager couch potatoes among Europe’s big countries. Eleventh in the table, they would be even higher but for their new custom of “Public Viewing,” as they call it in their impenetrable language. Since 2006, Germans have 198

taken to watching games in huge crowds on big screens outside. About 12 million Germans are estimated to have chosen this method to watch their team lose to Italy in that year’s World Cup semifinal. The figure was particularly impressive given that the 30 million other Germans who watched at home were by themselves the country’s largest-ever TV

audience. At Euro 2008, the “Fan Mile” by the Brandenburg Gate in Berlin had to be closed nearly three hours before the final because it was already packed with nearly half a million singing and dancing

“Public Viewers.” Yet these people do not show up in Germany’s “people meter” data.

Generally, European countries dominate at the top of Alavy’s table.

Partly this is because most World Cup games are played at times designed to suit them. There is a vicious (or virtuous) cycle here: because Europe has high viewing figures for soccer, World Cups are often held in Europe, and so European teams do well, and so European viewing figures are high. Only as Asians become richer and more enthusiastic about soccer might this change. Chinese prime time could be a very lucrative market one day.

The Asian figures in our table need a pinch of salt. The Korean figures are for 2002 to 2008, so they benefited from the “halo effect” created by South Korea’s run to the semifinal in 2002. Moreover, in Korea and Indonesia the World Cup is hard to escape because it is often shown “wall-to-wall” on several channels simultaneously. By contrast, Singapore scores high partly because only the best World Cup matches are shown there—and partly because Singaporeans love betting on soccer the way other people love soccer.

Figure 9.5 shows Alavy’s least-enthusiastic nations.

The Taiwanese must really not like soccer. Of course, even they cannot achieve a negative rating in practice, but our data combine the actual viewing figures with the specific characteristics of each game (time of day, group match, and so forth) to produce a final figure. In real life the Taiwanese numbers do not come out far from zero. Nor do the ratings for the US and India. An oddity of the US is that it is the only country where the nominal national team does not draw the highest ratings. At T H E C O U N T R Y T H AT L O V E S S O C C E R M O S T