In 1998, a somewhat different trend emerged. Figure 3 offers no evidence that opinion leaders were leading the rest of the population. Other polls conducted at the time offered an explanation for this, however: According to a study by Elisabeth Noelle-Neumann, opinion leaders were themselves divided throughout the campaign—and so they did not signal any specific direction everyone else could have followed.

On considering the overall role which those persons identified by the strength-of-personality scale as opinion leaders played in the German federal election campaigns during the 1990s, then, we could not help but conclude that the idea that these respondents could, in fact, be the long-sought opinion leaders had been impressively confirmed.

Four years later, however, during the 2002 federal election campaign, a completely different trend emerged. In contrast to 1998, the 2002 campaign was marked by two clear shifts in popular opinion. As had been the case in 1990 and 1994, the first shift was preceded by a political reorientation among opinion leaders. The second, however, completely broke the mold: In the last week of August, the Christian Democrats lost a seemingly unbeatable lead over the Social Democrats among the broad majority of the population—a shift that was only detected among opinion leaders about two weeks later.

What happened? There are three plausible explanations:

• Those persons identified by the strength-of-personality scale were not actually opinion leaders, either because the scale is not a suitable tool for identifying them, or because the opinion leader theory is completely false.

• The strength-of-personality scale might have been a suitable tool for identifying opinion leaders in the past, but was no longer appropriate due to changes in linguistic usage and response behavior over time. Opinion leaders thus continued to exist but were not identified by the scale with sufficient precision. Similar cases of “scale degeneration” are certainly not unheard of in survey research.

• The 2002 election campaign differed from prior campaigns in some important way, so that opinion leaders—who still continued to exist—were unable to exert any noticeable influence, or at least not as much influence as they had previously.