As was mentioned above, the study sample of Stock, von Hippel, and Gillert consisted of German householders. To ensure that we would have enough subjects for analysis at in all three innovation process stages, my colleagues and I recruited participants in two different ways. First, we used "snowball sampling” (Goodman 1961). In this method, individuals who have a rare characteristic—in our study, engagement in innovation development—are asked to identify others they may know who have the same characteristic (Welch 1975). (The utility of snowballing stems from the observation that people with rare characteristics tend to know or be aware of people similar to themselves.) In the second method, our goal was to increase the number of individuals in the sample who were likely to have successfully completed all three stages in the innovation process. We therefore deliberately sought out individuals who either had posted a description of an innovation they had developed on the Internet or had been featured on a German TV program devoted to individual inventors.
The net result was a sample containing both free innovators and entrepreneurial household sector innovators hoping to commercialize their innovations. In total, the sample we recruited for the study consisted of 546 individuals in the German household sector, 443 of them identified via the first method and 103 via the second. The two subsamples of respondents were similar in their demographic characteristics and were combined for analysis. Data were collected from respondents by means of an online questionnaire.