Details on Variable Coding for Multivariate Models throughout the Book
MODELS IN CHAPTERS 2 AND 6 (APPENDICES A AND F)
All data come from the American National Election Study’s cumulative data file. Not all variables correspond to each model (see Appendices A and F).
Income: VCF0114, coded from 0 (“0 to 16 percentile”) to 1 (“96 to 100 percentile”).
Education: VCF0140, coded from 0 (“8 grades or less”) to 1 (“BA degree or more”).
Unemployed: VCF0150, coded 1 if “Temporarily Laid Off” or “Unemployed”; 0 otherwise.
Working: VCF0150, coded 1 if respondent indicates that s/he is “working now” (including all combinations of response options that include “working now”); 0 otherwise.
Retired: VCF0150, coded 1 if “Retired”; 0 otherwise.
Not in Labor Force: VCF0150, coded 1 if respondent indicates not being in the labor force at all (not working now fulltime or part-time and not temporarily laid off/unemployed); 0 otherwise.
Age: VCF0101, coded between 0 and 1 for all models except the health care costs model in Chapter 2. For health care costs, the coding reflects the availability of government-provided health insurance at age 65 and over. In that case, the age variable is coded dichotomously in which 1 refers to people 65 and over and 0 for everyone 64 years old and younger.
Partisanship: VCF0301, coded from 0 (“Strong Republican”) to 1 (“Strong Democrat”).
Party Strength: VCF0301, coded from 0 (“Independent-Independent”) to 1 (“Strong Democrat” or “Strong Republican”).
Candidate Affect: VCF0409, coded as the absolute value of the sum of the things that the respondent likes about the Democratic candidate and dislikes about the Republican candidate minus the things that the respondent dislikes about the Democratic candidate and likes about the Republican candidate; response transformed to be coded from 0 to 1.
Care Who Wins: VCF0311, coded as 1 if “Care a good deal [about which party wins the presidential election],” 0 if “Don’t care very much/other.”
Party Affect: VCF0322, coded as the absolute value of the sum of the things that the respondent likes about the Democratic Party and dislikes about the Republican Party minus the things that the respondent dislikes about the Democratic Party and likes about the Republican Party; response transformed to be coded from 0 to 1.
Ideology: VCF0803, coded from 0 (“Strong Conservative”) to 1 (“Strong Liberal”).
Nonwhite: VCF0106a, coded 0 if “White”; 1 otherwise.
Black: VCF0106a, coded 1 if “Black”; 0 otherwise.
Hispanic: VCF0106a, coded 1 if “Hispanic”; 0 otherwise.
Southern: VCF0113, coded 1 if respondent lives in “political South” (i.e., one of the secession states); 0 otherwise.
Gender/Female: VCF0104, coded 1 if “Female,” 0 if “Male.”
Marital Status: VCF0147, coded 1 if “Married” or “Partners/Not Married (Volunteered)”; 0 otherwise.
Children: VCF0138, coded from 0 (“None”) to 1 (“Three” to “Eight to more”).
Media Consumption: VCF0728, coded from 0 (“No media”) to 1 (“All four media”).
Laid Off: VCF0156, coded 1 if “Yes,” 0 if “No.”
Hours Reduced: VCF0157, coded 1 if “Yes,” 0 if “No.”
Delayed Health Care: VCF0873, coded 1 if “Yes,” 0 if “No.”
Years in Community: VCF9001, Log of variable used.
Church Attendance: VCF0130, coded from 0 (“Never” or “No religious preference”) to 1 (“Every week”).
Homeownership: VCF0146, coded 1 if “Yes, own,” 0 if “No, not owned.”
Contacted by Political Group: VCF9030a and VCF9031, coded 1 if respondent reported being contacted by a major party or anyone else during the campaign to talk about supporting specific candidates during the election; 0 otherwise.
External Efficacy: VCF0609, coded from 1 (“Disagree [that public officials don’t care much what people like me think]”) to 0 (“Agree”).
Internal Efficacy: VCF0614, coded from 1 (“Disagree [that sometimes politics and government seem so complicated that a person like me can’t really understand what’s going on]”) to 0 (“Agree”).
Perceived Closeness: VCF0714, coded 1 if respondent thinks that the presidential race will be a close one, 0 if respondent thinks one candidate will win by quite a bit.
Issue Importance of Unemployment: VCF0875b, coded 1 if “Unemployment” or “Create Jobs”; 0 otherwise.
Issue Importance of Health Care Costs: VCF0875b, coded 1 if “Health Problems/Cost of Medical Care”; 0 otherwise.
Issue Importance of Retirement Costs: VCF0875b, coded 1 if “Aged/Elderly” (which mostly includes Social Security, elderly medical care concerns, Medicare benefits) or “Increased assistance to the aged” or “Social Security won’t be around in the future/paying into a system that won’t benefit me/them”; 0 otherwise.
Issue Importance of Education Costs: VCF0875b, coded 1 if “Education” (note that the first subreference in this category is related to education costs: “financial assistance for schools/colleges/students”); 0 otherwise.
Economic Insecurity Most Important, Other Economic Issue Most Important, Noneconomic Issue Most Important: All derived from VCF0875b and defined as noted in main text.
Party Handling Preference: VCF9012, coded 1 if “Democrats” or “Republicans,” 0 if “Not much difference” or “Neither.”
Donated: VCF0721 and VCF0741, coded 1 if respondent indicated making a contribution to a political party, to an individual candidate running for office, or to any other group that supported or opposed candidates [during the election year]; 0 otherwise.
Spent Time: VCF0718 and VCF0719, coded 1 if respondent indicated attending any political meetings, rallies, speeches, or dinners or doing any other work for one of the parties or candidates; 0 otherwise.
CODING FOR VARIABLES IN EDUCATION COST EXPERIMENTS (APPENDIX C)
Male: Coded 1 for male, 0 for female.
Year in School: Coded from 1 to 0 representing senior in school to freshman in school.
Financial Resources: Coded from 1 to 0, representing how much money respondent currently has in personal checking and savings accounts combined (not including investment accounts).
CODING FOR VARIABLES IN HEALTH CARE COST EXPERIMENTS (APPENDIX C AND APPENDIX D)
Financial Resources: Coded as a combination of family income and whether or not respondent indicates having health insurance; see Chapter 5 for more details on coding as well as robustness checks using alternative measures.
Education: Coded from 0 (“less than high school”) to 1 (“Bachelor’s degree or higher”).
Age: Coded from 0 to 1.
Male: Coded 1 for male, 0 for female.
Cancer Personal Experience: Coded 1 if respondent indicates that s/he, family member living in household, family member not living in household, close friend living in household, or close friend not living in household has been diagnosed with and treated for cancer; 0 otherwise.
Number of Children: Coded from 0 (none) to 1 (three or more).