Garrett 4
of illegality. But, as Roberds (1997, 87) notes, “not all scandals involve illegal behavior.”
Regardless of definition, it is clear that even narrowly defined campaign crises such as
scandals are common and have serious impacts on campaigns and elections. Research
consistently demonstrates that scandals of all kinds have a negative impact on electoral
outcomes, either in vote share or in retirement decisions (Peters and Welch 1980; Roberds 1997;
Lough 1998; Jacobson and Dimock 1994; Dimock and Jacobson 1995; Bridgmon 2002).
3
In one of the more methodologically advanced examinations of scandal effects, Goidel and Gross
(1994) find that scandal has the most predictive power of any independent variable on challenger
quality, challenger and incumbent expenditures and incumbent vote share.
4
Overall, scandals
cost House incumbents about 7 percent of the total vote share and yield an additional $100,000
in challenger fundraising (Goidel and Gross 1994, 140-142). Scandal is so electorally and
professionally dangerous because blame is easily affixed to individuals. For example, even
though the House bank scandal broke no laws, it “ended many more congressional careers than
policy disasters such as the savings-and-loan debacle, which left taxpayers holding the bag for
hundreds of billions of dollars,” (Jacobson 2001, 176-177).
To summarize, despite some contradictions, definitions of political scandal emphasize
three characteristics: personal behavior, ethics and publicity. Existing works measure scandal
impact externally, usually via impact on votes and fundraising. I will return to this point in the
“findings” section of this paper in exploring how political professionals conceptualize crises
3
Most works do not differentiate between scandal types. However, there are notable exceptions. In their
examination of House corruption charges between 1968 and 1978, Peters and Welch (1980, 705) find that morals
and bribery charges have particularly damaging impacts on incumbents, leading to vote losses between 12 percent
and 22 percent for members of both parties. Relying on Peters and Welch’s (1978; 1980) typology, Lough (1998)
examines differential scandal impact on retirement decisions, vote share and challenger emergence between 1976
and 1994.
4
The Goidel and Gross article is primarily devoted to campaign finance, but scandal serves as a dummy variable in
their models. Their method is particularly advanced because it uses a three-stage least squares (3SLS) regression
approach, which marks a departure from the two-stage (2SLS) approach traditionally employed. 3SLS allows for
estimating an entire set of equations simultaneously.
Convention