1 A patent count of less than 1.0, but more than 0.0, can occur when there are two or more different county addresses on the patent. The patent count is divided across the relevant counties.

2 A patent may be included in more than one category within a dimension and also be classified along more than one dimension. For example, a patent describing a machine learning-based image processing tool for an autonomous vehicle will be counted in all three dimensions (i.e., AI technique: machine learning; functional application: computer vision; application field: transportation).

3 These categories are not mutually exclusive.

4 These analyses are based on algorithms that match names to the usual gender associated with that name, an inexact but still useful process.

5 These affiliations are identified in the documents through a two-step process that starts with automatic assignment based on unambiguous words, such as “company,” “university,” or “Canada,” followed by a second step that manually identifies ambiguous entries.

6 The tabulated data for peer-reviewed publications are from the Scopus database used in the State Indicators report (NSB 2021b). The same taxonomy of disciplines for fields of S&E and the classification of sectors that produce publications from that report are used here.

7 An alternative approach that surveys for business behavior that is consistent with continuous product and process improvement systems rather than self-reporting of innovation is described in a recent research paper by Tian, Wojan, and Goetz (2022).