Negative campaigning has traditionally been considered ‘bad’ for democracy. In addition to the research associated with the demobilization hypothesis, there is the argument that over-negativity in political campaigning decreases voter turnout and, more generally, suppresses enthusiasm for the political process (Ansolabehere and Iyengar 1995; Ansolabehere et al. 1994; Ansolabehere et al. 1999; Buchanan 2000; Patterson 2002). Pundits, practitioners and academics alike have denounced attack advertisements for their detrimental effects on the information environment (Jamieson 1992; Kamber 1997; Mark 2007). As the argument goes, elections are one of the primary means of communication between potential representatives and the constituency. Campaign advertisements are supposed to function as proposals from candidates on which the citizenry base their decisions when choosing their representative (Riker 1996). A common claim is that attack advertisements are designed to mislead voters or, at least, are more likely to contain misinformation that misleads voters. Another claim can be made on the basis of limited resource allocation. If candidates are allocating funding to criticize their opponents, then they are not spending as much on communicating from their own platforms. Likewise, a voter’s information processing ability is finite, and attention paid to negative advertisements is attention diverted away from positive, supposedly more informative sources. If the above points are valid, and assuming that information and dialogue between representatives and constituents form the basis of a healthy democracy, then the reasonable prescription would be to curtail negative campaigning for the sake of democracy. Recent research from US politics, however, suggests that such concerns may be exaggerated or even misplaced, and that negative campaign advertisements in fact contain a higher degree of information than positive advertisements because of the perceived need to support criticism with evidence (Geer 2006). Furthermore, contrary to the initial studies supporting the demobilization hypothesis, recent studies tend to suggest that although negativity affects the voters’ perception of the fairness and efficacy of the overall political process in complex ways (Brooks and Geer 2007; Jackson et al. 2009; Lau et al. 2007; Sides et al. 2010) and may affect the voters’ perception of the targeted candidate, depending on

each voter’s characteristics (Fridkin and Kenney 2011), it does not decrease voter turnout (Freedman et al. 2004; Freedman and Goldstein 1999; Garramone et al. 1990) and, in some cases, may even result in a slight increase (Brooks and Geer 2007). Similarly, the findings from US politics have been upheld by the quantitative research on negative campaigning in Taiwan (Sullivan 2008, 2009, 2010), contrary to what might be suggested by the qualitative observations of Taiwanese elections made by both the popular media and academic sources (Rawnsley 2000; Schafferer 2006). Using data obtained from the four presidential elections between 1996-2008, Sullivan found that, in general, negative claims, defined as a candidate’s statements about an opponent, are much more likely to be supported by evidence than are positive claims, defined as a candidate’s statement about herself/himself. In the four presidential elections, while the evidence rate of negative claims is admittedly low at 9.3 per cent for newspaper advertisements and 6.8 per cent for television advertisements, it is considerably higher than the 0.9 per cent and 0.8 per cent evidence rate of positive claims. Our defence of negativity is not to do with the argument that negative campaigning is necessarily good or bad; our intention is to consider its benefit or detriment to the information environment relative to positive campaigning. In line with Geer’s analysis of US politics, Sullivan’s findings clearly support the claim that, on the whole, negative claims are much more likely to be supported with evidence. How can we explain the discrepancy between qualitative and quantitative observations? Is there something else at play other than our senses being inherently unsuited to qualitatively assessing campaign advertisements, something specifically designed to manipulate our perception? Drawing upon Carmines and Stimson’s distinction between ‘easy’ and ‘hard’ issues (1980), we contend that the discrepancy is due to the high salience of a subset of negative advertisements specific to Taiwan: that is, attack advertisements based on the prevalent antiChina sentiment in Taiwan. Because the cross-Strait relations issue is a very ‘easy’ one that evokes immediate ‘gut responses’ from the Taiwanese population, candidates and their campaign managers tend to disregard the traditionally perceived need to buttress criticism with evidence. In order to better understand the impact of ‘China’ on negative campaigning in Taiwan, we tested our hypothesis in two stages. First, using the same sample of campaign advertisement as Sullivan (2010), we examined whether introducing the China factor negatively impacts on the evidence rate of campaign advertisements, especially negative ones. We found that there is a significant difference in the evidence rate between China and non-China attack advertisements. In order to assess whether the easiness of ‘China’ or ‘cross-Strait relations’ is the factor contributing to this phenomenon, we reviewed three significant instances of politicians inciting anti-China sentiments for political purposes. While it is difficult to conclusively test this mechanism, we observed that the discourse in these instances supports our contention that the easiness of the China issue is encouraging candidates not to provide evidence for their claims.