ABSTRACT
Having seen that public opinion has been a strong driver of defense spending and evaluated several models for defense spending preferences, we have constructed our model to include main variables of interest and, following the discussions in Bartels (1994), a group of demographic variables to control for the impact of various other demographic and socio-economic factors such as age, race, education, marital status, gender, and employment status. The model also includes a group of dummy variables to control for the individual year of the study as well as two interaction variables of interest: Defense Spending Preferences = β0 + β1 Isolationist + β2 Scale Political Party Identifica-
tion + β3 Economic Stake + β4 Married + β5 Female + β6 Age + β7 Education + β8 Unemployed + β9 Retired + β10 Not Working
8 + β11 Black + β12 Asian + β13 Native American + β14 Hispanic + β15 Other Race
9 + β16 Dummy Variables to Control for Year (1980-2004 with 2008 as the baseline) + β17 (Married × Female = Security moms) + β18 Retired Women. Based on the literature to date, we expect the Isolationist dummy variable to have a strong
negative correlation with defense spending preferences. In addition, we expect a strong positive correlation between defense spending preferences and scale political party identification – a high score on the political party identification is ‘very Republican.’ We also expect the economic stake variable to have a positive correlation. Finally, previous literature (e.g. Bartels, 1994) has shown that an individual’s education level will have a strong negative correlation with defense spending preferences and that being female has a negative correlation with the dependent variable. Based on public discussions on ‘security moms,’ we also expect security moms to have a positive effect on defense spending preferences.10
