Appendix: Coding of Variables
Exposure to advertising / The construction of these variables is described in detail in Appendix B of Freedman et al. (2004), the data used in this paper. The authors used Campaign Media Analysis data on which ads were aired when, and in which media market, coded by the Wisconsin Advertising Project as positive, contrast, or negative in tone, in combination with ANES survey data on how often respondents watched certain television shows (e.g., Jeopardy) and which county respondents resided (thereby knowing the ads shown in the respondent’s locale during those shows). Roughly two-thirds of ads aired during the specific shows mentioned in the survey. Estimates of how many of the additional one-third of ads respondents were exposed to were based on an individual index of overall television viewing.For the volume measures the log of exposure to negative ads was taken. To gauge the proportion of negative ads to which an individual was exposed I divided the estimate of estimate of exposure to negative advertising by the estimate of the overall volume of exposure to political advertising for each respondent.
Remember seeing an ad / (v000338) 1 = Yes, 0 = No
Perceptions of closeness of presidential race / (v000486-v000487). 0 = One candidate will win by quite a bit, 1 = Will be close
Ability to place the presidential candidates ideologically on issues / Eight issues in v000562, v000568, v000592, v000597, v000625, v000630, v000655, v000660, v000696, v000698, v000718, v000723, v000735, v000739, v000783, v000790. Coded correct where Gore placed to left of Bush. Recoded to 0 for 0-1 correct placements, 1 for 2-3, 2 for 4-5, and 3 for 6-8.
External efficacy / (v001527, v001528, v001529). Scales summed and divided by three, with 1 = lowest efficacy and 5 = highest efficacy
Perceptions of government power / (v001513, v001514) 1 = Government too powerful, 2 = Government not getting too strong/ no opinion, 3 = Government should have more power
Certainty of candidates’ ideology / (v001373, v001375). Scales reversed, coded 1, 2, 3, summed, and divided by two. 3 = Certain about both candidates, 1 = Uncertain about both candidates
Attention to the campaign / (v000301) Not much interested = 1, Somewhat interested = 2, Very much interested = 3
Bush/ Gore likes and dislikes / Likes from v000306-v000310, v000318-v000322 for scale from 0 to 10.
Dislikes fromv000312-v000316, 000324-v000328 for scale from 0 to 10.
Turnout / Voted = 1 (Coded 4 in v001241), Did not vote = 0 (Coded 1, 2, or 3 in v001241).
Local news viewing / (v000331, v000332). Days per week watch early and/ or late local news, summed, and divided by two, with 0 = do not watch either, and 7 = watch both every day
Newspaper reading / (v000335). Days per week read newspaper
Age / (v000908). From 18 to 97
African-American / (v001006a, v001006b, b001006c). African-American if self described as black to any of these questions. 1 = African-American
Female / (v001029). Female = 1, Male = 0
Income / (v000994). Coded as in ANES from 1 = none or less than $4,999 to 22 = $200,000 and over
Education / (v000913). Coded as in ANES from 1 = 8 grades or less, to 7 = advanced degree
Political information / (v001356, v001357, v001446-v001457). Knowledge of which party controls the House, the Senate, and correct identification of who Trent Lott, William Rehnquist, Tony Blair, and Janet Reno are. Coded 1 for correct, 0 for incorrect and summed, with 0 = no correct answers and 6 = all correct
Strength of partisanship / (v000523). 0 = Independent, .33 = Independent Republican or Democrat, .67 = Weak Republican or Democrat, 1 = Strong Republican or Democrat
Mobilized / (v001219, v001222). Coded 1 = party talked to respondent or sent mail, 0 = party did not talk to respondent or send mail
House race competitive / Coded 1 if respondent lives in a competitive House district (from Cook Political Report)
Senate race competitive / Coded 1 if respondent lives in a competitive Senate state (from Cook Political Report)
Presidential race competitive / Coded 1 if respondent lives in a competitive Presidential state (from Cook Political Report