One of the greatest challenges facing researchers interested in expert performance is the limited access to experts. Experts are rare, particularly if one adopts a definition of expert performance in which the expert is defined as statistical outlier. Nevertheless, in research on expert performance, there are alternatives to laboratory studies. Ericsson describes three complementary alternative approaches.
For some domains, such as baseball and golf, detailed records on players’ performance are kept. These data provide a potentially rich source of information about expert performance. For example, Simonton has used archival information to answer questions about age and peak productivity. His work shows that productivity peaks during the 30s or 40s, with earlier peaks in domains such as mathematics and later peaks in domains such as history. Simonton has also shown that the ratio of outstanding achievements to total output is constant across age, but that total output decreases.
Research based on archival data also show that peak performance is usually attained 5-10 years after physical maturation. As Ericsson notes, "This discrepancy in the age of full biological maturation and that of peak performance implicates an important role of preparation" (p. 9). This points to a distal mechanism of expert performance. Historical data also show that absolute levels of domain-specific performance have increased markedly over time. Complexity has increased. This finding indicates that the proximal mechanisms governing expert performance are acquired.
In summary, historical data provide inferential evidence for questions about the mechanisms that govern expert performance, and about how these mechanisms develop within the individual.
The goal of a second approach is to develop laboratory tasks sensitive to actual skill differences, and to identify proximal mechanisms governing expert performance. As Ericsson describes, this approach "focuses on the mechanisms responsible for the consistently superior performance of experts on representative tasks in their domain of expertise" (p. 13). Methods for performing a representative sample of tasks are identified by using "process-tracing" and experimental methodologies. Chase and Simon's work on chess demonstrates this approach. They concluded that pattern recognition is responsible for expert performance. Ericsson and others explored this, and argued that superior recall for domain-specific material is a long-term memory phenomenon. Recently, in their theory of long-term working memory, Ericsson and Kintsch have proposed a systematic account of retrieval structures.
Study of Acquisition
Ericsson identifies study of acquisition of expert performance as a third major approach in research on expert performance. What factors lead to the acquisition of expert performance? This approach differs from the previous approach in that the emphasis is on distal, rather than proximal, mechanisms. That is, what events in the performers life (prior to the testing situation) can explain expert performance? Based on questionnaire and interview research with experts, Ericsson claims that deliberate practice is the primary mechanism responsible for expert performance. He distinguishes this activity from other forms of experience, including "work" and "play."
Integration of Approaches
The three approaches described above are complementary and, together, provide a more complete picture of expert performance than any one approach. Inferences about processes and mechanisms involved in expert performance can be made from historical data. The historical increase in domain complexity suggests that the mechanisms responsible for superior performance are acquired. Laboratory research provides a detailed account of what these mechanisms are. The third approach suggests an account of how these mechanisms are acquired.