Endeavor

a final deductive process, empirically oriented and ana- lytically consistent. It is important to note that as a result of the transforma- tion of variables and standards, the final values of the de- terminants — and therefore of the rankings and the final index — are relative. For example, the fact that Curitiba is positioned first in the "infrastructure" determinant and received a value of 7.63 does not mean that Curitiba is only 2.37 (10.00 — 7.63) points away from having perfect infra- structure to encourage entrepreneurial activity. Likewise, Belem isn't 4.79 from having a completely inadequate infrastructure for entrepreneurs. The value given to the cities in the determinants only indicate the relative posi- tions and how far each city is from the overall average of the 14 capitals. ConstruCtIon oF tHE EntrEprEnEurIal pErFormanCE mEasurEmEnt Entrepreneurial performance is a comprehensive con- cept and can be defined in various ways, for example, as wealth creation by entrepreneurs or as simply the increase of entrepreneurs and self-employment. The challenge of conceptualizing and measuring entrepreneurial perfor- mance is therefore similar to building the determinants for performance. Following the comparative frameworks for countries, per- formance indicators were collected that could adequately represent three fundamental aspects: [1] the intensity of entrepreneurial activity in Brazil, as measured both by the number of entrepreneurs in the field and the creation and survival of new businesses; [2] economic performance of entrepreneurs, in particular, high-impact entrepreneur- ship; and [3] the wealth capacity and/or jobs generated by business activity. Various performance indicators were collected and, after eliminating redundancies or measure- ment problems, eight variables remained, with the descrip- tions and sources are found below (page 95). However, the indicators represent dimensions sometimes quite distinct from the entrepreneurial performance and combining them requires care. In particular, many of the collected performance variables highly correlate with one another. For example, the growing number of companies, the survival rate of pre-existing businesses and the cre - ation of jobs tend to be the result of these economic processes, and therefore correlate, even if the measure- ments are well defined and different from one another. The Principal Component Analysis, already used for the exam- ination of determinants, was applied to the performance indicators in order to deal with this problem. The Principal Components Analysis' main objective is to represent a set of many correlated variables from a small- er set of components that, by construction, do not cor- relate with each other, without losing relevant information presented in the data (Bartholomew et al, 2008). In other words, with this technique we can build dimensions — rep- resented by the components — common for indicators and that represent them more thoroughly. If one of the indica- tors is strongly explained by the same social or economic process — expansion of the economy and consumption, for example — it's likely to find a component to represent it. From a technical point of view, the Principal Component Analysis results in a set of components that explain, in de- scending order of importance, exactly the same data vari- ation explained by the original indicators. The first compo- nent has higher explanation power of the variation and the last, the least. The chart below, from a hypothetical exam- ple illustrates how the variables are arranged around the first two components of a Principal Component Analysis. Example of the Principal Component Analysis Method 3 11 10 6 14 5 7 8 12 1 13 4 2 9 -0,6 -0,4 V3 V7 V8 V6 V4 V2 V1 -0,2 0,0 0,2 0,4 -0,6 -0,4 -0,2 0,0 0,2 0,4 -4 -2 0 2 -4 -2 0 2 PC2 PC1 92

• Cover