The possibilities the internet provides for empirical social research resemble a double-edged sword. Especially unrestricted self-selected web surveys reveal the advantages and difficulties inherent in them: Large numbers of participants can be reached on significantly lower costs than traditional "offline" methods could af­ford, whereas most web surveys suffer from not being representative. They have to get past two major obstacles to reach participants: Firstly, in terms of _digital divide_, online-access requirements are not equally distributed in the population. Secondly, potential participants have to notice a survey before they can decide to participate. The willingness to participate is expected to be unequally distributed, too.

We analyse the inherent conflict between advantages and disadvantages of web surveys. We assume that short field times aggravate the problem of represen­tativeness. As we know from _diffusion theory- (Rodgers95), a new product is initially adopted by _innovators_, who are highly educated and use multiple in­formation sources. Gradually, they are followed by _early adopters_, the _early­_ and the _late majority_, until finally the laggards_ accept an innovation. The disposition to adopt an innovation depends on the adopters' interest, awareness and experience.

Translated into the logic of web surveys, many innovators can be reached quickly. However, to reach a broader part of the population, longer field times have to be granted. The way the knowledge about the existence of a survey spreads is of vital importance as the types of adopters also differ with regard to sociodemogra­phic variables. Hence, the diffusion of web surveys is highly connected with their substantial results .