3. Methodology Sample The following

3. Methodology Sample The following procedure was used to select firms for the sample. First, all publicly traded restaurant firms with relevant data available on Compustat Industrial Annual from 1965 to 2004 were selected. The process of internationalization is a long-term strategy employed by restaurant firms and because annual data is aggregated it can better depict general trends. Next, international restaurant firms were identified and kept, while all domestic firms were removed from the sample.Purely domestic firmswereremovedbecauseonlyasmallproportionoffirmsoperateinternationallyand thus, the result of this process was a homogeneous sample of firms that are doing business abroad. There were some firms that had data for the entire duration of our sampling time frame; therefore, including all data for each year would have resulted in survivor ship bias and auto correlation problem. To eliminate these two problems a screening process was utilized to determine whether a firm-year may enter the sample. First, a reference year was established, defined as the year in which any firm first reported foreign earnings. Then, three years prior to the reference year were selected as well as also two more years of data after the reference year. This procedure had many advantages. First, because the sample consisted exclusively of firms reporting foreign earnings, a perfect match existed with each firm providing data before and after the process of internationalization had commenced. Second, because each firm only contributed six years of data, the problem of auto correlation was reduced. Finally, the problem of survivor ship bias was avoided, because all firms were included and not just those that survived during our sample selection period. Compustat and LexisNexis Academic Universe were consulted to identify the reference year. Finally, at least five firm-year observations were required for each firm to ensure the validity of the regression model estimation. The restaurant setting and the procedure employed ensured two desirable properties for follow up ordinary least squares analyses: normality and homogeneity. The final sample consisted of 147 firm-year observations, with 27 international firms. One major problem when using any procedures to screen firms is that the final sample might not be representative of the total universe of publicly-traded international restaurant firms. Therefore, we conducted two tests to make sure that our firms are a reasonable representation of the population. First, the total sales of the sample firms were $52,527.15 million, amounting to 93.35 percent of total sales of all international restaurant firms.Second,the total market capitalization of sample firms was $68,052.81 million, which is 99.31 percent of the total market capitalization of all international restaurant firms. The sample firms own and operate, and/or franchise restaurants in all three major segments: quick service; mid-scale; and casual and fine dining. Some firms specialize within a single segment and across menus, while others operate multiple brands across segments and menus.