THE PREVIOUS CHAPTER SHOWED that whether monitors denounce an election depends partly on the types of irregularities committed. This raises an important issue: If politicians realize that they are less likely to be denounced for certain types of irregularities, do they or their accomplices stop cheating overtly and instead shrewdly shift their cheating to these less risky forms to evade criticism?1 If they do, monitors may not only misjudge elections at times, but also distort how they are actually conducted. This is a serious concern.
But is it justified?
Politicians certainly do cheat quite creatively; some incumbents even try to manipulate the monitors as in the Kazakh example from Chapter 3. Despite these efforts, monitors were not fooled. Harassment and intimidation, restrictions on campaigning, media bias (even confiscation of entire newspapers), biased election administration, ballot box stuffing and failure to seal the ballot boxes, problems with the tabulation of the votes, and so on, were rampant. The Organization for Security and Co-operation in Europe (OSCE) report discussed an extensive mix of both overt and subtle forms of cheating and concluded that “the election did not meet a number of OSCE commitments and other international standards for democratic elections.”2 The use of subtler forms of fraud was part of a mix of forms of manipulation and did not appear chosen only to deceive monitors, nor was it successful at doing so.
In other cases, however, countries may be more successful. As Chapter 3 showed, some incumbents strategically invite less critical organizations. Therefore, the question is: Do politicians also shift into certain behaviors to evade criticism? If such shifts occur, how widespread is the problem and how much is it really driven by the presence of international monitors?
This question is not whether politicians shift to clandestine fraud. In theory, monitored countries could indeed deftly shift to clandestine forms of cheating, and if they do, then we will by definition never know. However, in practice, clandestine cheating is probably not that easy. Controlling the entire electoral apparatus is nearly impossible, and irregularities can occur at multiple levels. In heavy-handed regimes, fear may lead various decentralized actors to seek to please the regime by delivering the expected electoral outcomes through a variety of unsanctioned methods. And even if politicians skillfully conceal cheating at all levels, their repeated victories will raise suspicions and the world would witness a rise in countries that hold seemingly impeccable elections but never oust their governments despite various domestic problems. That said, some adept use of clandestine cheating may well occur on the margins, and no analysis can definitively rule out that politicians are using clandestine cheating to evade monitors.
However, it is more realistic to consider that fraud does not simply come in two forms: detectable and undetectable. Rather, all types of fraud are associated with a certain probability of detection, and, as explored in Chapter 4, a certain probability of criticism. And although politicians in monitored countries may have an even greater incentive to try to hide their cheating, with monitors present to scrutinize what is going on hiding malfeasance becomes harder.
Monitors are cognizant of the many possible weaknesses in the electoral process. They engage in a broad range of activities to uncover problems. In addition to direct observation, international monitors consult a wide array of sources, including incumbent parties, government officials, opposition parties, nongovernmental organizations (NGOs), and civil societies. Their often-extensive reports examine a vast array of behaviors, including manipulation of the media, voter rolls, election commissions, and so on. Thus, politicians may be creative in the ways they cheat,3 but all cheating risks detection and monitors do report, though do not always denounce, a wide variety of irregularities. In the 1990s, for example, Kenya tried to institutionalize manipulation by using malapportionment and “nationality clauses” to discourage true political competition. And in 2000 both Russia and Peru used “borderline categories of targeted ‘low-intensity’ repression.”4 However, none of these behaviors went unnoticed by international monitors even if, as in the case of Russia in 2000, monitors still hesitantly accepted the election.
Thus all cheating risks detection and inviting international monitors does not make it easier to hide. Therefore, the question this chapter explores is not whether monitors cause politicians to shift into hidden forms of fraud. Rather it is: Knowing that monitors may tolerate some irregularities more than others, do politicians successfully change the way they cheat to evade outright denouncement?
If riskier irregularities are those that monitors are more likely to denounce, while safer forms of irregularities are those monitors are more hesitant to denounce, then if monitors cause politicians to successfully shift to safer forms of cheating, two patterns should hold.
The first pattern to expect pertains to the profile of any one given election. If monitors cause politicians to substitute successfully into safer forms of cheating to evade criticism, then in any one given election, when safer forms of cheating occur, riskier forms of cheating should be less common. However, this pattern should not hold in nonmonitored elections, where, if present, risky and safer forms of cheating should be complements.
The existence of this pattern does not in and of itself mean that monitors caused it. All dishonest politicians have incentives to vary how they cheat to evade attention and maximize their legitimacy.5 As Andreas Schedler has argued, “The limits of the authoritarian imagination are not logical, but empirical. Rulers may choose a number of tactics to help them carve the democratic heart out of electoral contests.”6 Indeed, governments have long engaged in a variety of subtle methods of manipulation throughout the election period. Attempts at weakening the opposition were present during ancient Rome’s strategy of divide et impera. Governments have long sought to prevent voters from acquiring true information about their choices. Singapore, Malaysia, and Thailand have all practiced such pseudo democracy since the mid-1960s, since the early 1970s, and during most of the 1980s, respectively.7 Some of the redistricting battles in the United States attest to the continued forms of subtle manipulation even in established democracies, as did the longstanding attempts to bar black voters from registering to vote. Thus, the mere presence of safer forms of fraud is not unusual and does not prove that monitors caused the behaviors. However, if safer irregularities nearly always coexist with riskier irregularities, this undermines the argument that monitors cause politicians to switch to safer irregularities to avoid criticism.
The second pattern to expect pertains to changes in the mix of election irregularities over time. If monitors cause politicians to successfully switch to safer forms of cheating, then from one election to the next, increases in safer forms of cheating should correlate with decreases in riskier forms. However, this pattern should be less prevalent in nonmonitored countries.
Once again, however, even if monitors are associated with such a shift, this does not necessarily mean that they caused it. The conduct of elections may fluctuate for various reasons. For example, newly democratic countries often backslide.8 Fraudulent elections may occur as a deviation from a better history, as in the 2000 election in Peru.9 Indeed, a host of forces bring oscillation between democratic progress and setbacks,10 producing a mixed quality of elections over time. In democratizing countries, some irregularities may take longer to address, because they require the passage of legislation and the creation of implementation mechanisms.11 Overt fraud may be easier to eradicate, while subtler types may persist due to underdeveloped capacity and inexperience. Thus, if international election observers are present, they may then be associated with a decrease in riskier fraud but persistence in subtler, safer types of irregularities. Because safer types of irregularities tend to be administrative, they may even increase during a period of democratization as more people are learning how to register to vote and as more parties compete. An apparent rise in safer irregularities could also be because some monitoring organizations have improved reporting practices over time. For example, the early OSCE reports were short and almost exclusively focused on election day. Since 1996, the reports have considered a variety of problems, but this does not necessarily mean that these problems have grown. Thus, a rise in safer irregularities does not prove that monitors brought this about. However, if countries where safer irregularities appear to increase over time do not also show a concomitant decline in riskier behaviors, then it makes little sense to accuse monitors of causing strategic substitution of safer behaviors.
Thus, neither the presence of a particular pattern of irregularities nor the presence of shift between types of irregularities necessarily means monitors are to blame. However, if safer forms of cheating are accompanied by continued or increased risky forms of cheating, then the use of the safer forms of cheating can hardly be characterized as strategic efforts to hide from monitors, and thus monitors can hardly be blamed for them. It would also be hard to blame monitors for a shift if the patterns in monitored elections really do not differ from those in nonmonitored elections. Thus, for it to be at least plausible that monitors cause politicians to change how they cheat, then safer and riskier forms of cheating should be substitutes, not complements. However, this pattern should not hold in nonmonitored elections, where, if present, risky and safer forms of cheating should be complements.
Even if we do see increases in safer forms of cheating accompanied by decreases in riskier forms, then it is necessary to investigate further whether the observed shift in a given country is possibly because of monitors, or whether it has alternative explanations.
Safer forms of cheating are those that monitors are less likely to denounce. Thus, politicians may seek to shift the timing of their manipulation. As noted in the previous chapter, it is a standard criticism that monitors pay most attention to the events on election day and ignore events before the election.12 Certainly international monitoring organizations have the largest delegation present on the day of the election and are more active on that day than in the pre-election period. Therefore, if politicians understand the biases of monitors and seek to exploit them, then the strategic response would be to shift manipulation from election day to the pre-election period. Instead of gross ballot box fraud, politicians would control the media, abuse incumbent resources, or manipulate the voting lists. Instead of buying votes and violating the secrecy of the vote, politicians would stack the electoral commission in their favor, make candidate registration difficult, and so on.
However, the previous chapter found that monitors’ overall assessments depend less on the timing of the irregularities than on their nature. Monitors were more likely to criticize blatant forms of cheating both on election day and in the pre-election period. Thus, tabulation problems, voting fraud, intimidation, misuse of media or government resources, or restriction on freedom to campaign, and so on—those behaviors that are clearly intended to manipulate the outcome—were more likely to trigger negative assessments from the international monitors. Similarly, legal frameworks that enable biased electoral commissions, or restrict the scope of the office, or restrict who can vote or stand in elections and so forth, were also likely to draw criticisms. In contrast, monitors appeared willing to give countries the benefit of the doubt when the problems might stem from inexperience or lack of capacity. Administrative problems such as errors in voter lists, complaints about the conduct of the electoral commission, problems with voter information, logistical issues during voting, and so forth were not likely to trigger criticism by monitors. To exploit these tendencies, politicians would therefore shift away from the more overt irregularities to those that can be dismissed as administrative shortcomings.
Finally, violence and repression are common tactics for leaders to use to hold on to power.13 Pre-election violence has been particularly prominent in countries such as Colombia, India, Iran, South Africa, and the Philippines. Violence during the pre-election period is relatively easy to observe and such violence is also generally associated with poor overall election quality. Monitoring organizations are therefore hard pressed to comment on it and they frequently do. Nevertheless, Chapter 4 found that monitors were less likely to denounce elections outright when pre-election violence was high. This hesitation to criticize could well be because monitoring organizations fear galvanizing further conflict after the election. If politicians realize this, then they may shift away from other types of fraud and instead increase their use of pre-election violence.
In sum, if politicians are seeking to reduce their risk of denouncement by international monitors, they might rely particularly on safer tactics such as (1) pre-election manipulation, (2) manipulation that could be construed as administrative, or (3) pre-election violence.
To examine the patterns of cheating, the analysis uses the data already discussed in Chapter 4, Table 4.2. For review, these larger categories are listed in Table 5.1. They are all ordered variables that rank problems as none, minor, moderate, or major.14 Two datasets exist of these variables. One is based on election monitoring reports and therefore is only available for monitored elections. The other is based on U.S. Department of State Reports on Human Rights Practices, and is available for both monitored and nonmonitored elections. This chapter uses both sets of data. For more on these, see Appendix A.
The difference between administrative problems and outright cheating is not their probability of detection. If done cleverly, gross forms of cheating such as ballot stuffing or vote buying may be no more visible than misconduct by the electoral commission in the pre-election period. Indeed, problems with voter lists, for example, may be more visible than misuse of public funds. Rather, the difference between administrative problems and those categorized as cheating is the degree to which they can be assumed to be intentional and therefore be likely to draw strong criticism.
The behaviors classified as cheating are those where intent is reasonably certain. Politicians do not intimidate voters by accident. Vote buying is not a product of undeveloped logistical capacity. Reporting more votes for the incumbent than turnout justifies is not because the election officials cannot count. Similarly, incumbents do not restrict or dominate state media because no contrary voices exist, they do not ban campaign activities of opponents because there is nowhere to hold rallies, nor do they use government coffers to fund their campaigns because they forgot to open a separate bank account. Instead, the most likely intention of these behaviors is to manipulate the outcome of the election.
In contrast, for behaviors classified as administrative problems the intent is more ambiguous. Failure to inform voters of their polling station locations could be intentional, but it also could be because the information is difficult to get out. Voter list problems could be due to systematic manipulation to favor some parties and politicians, but they also could be caused by problems with processing paperwork. Lack of professionalism or neutrality on the electoral commission could be planned, but it could also be the result of a lack of proper training and experience. Not ensuring that voters have full secrecy or that campaign materials are kept at a proper distance from the polling station could be abusive, but such problems could also be oversights or lack of training for poll workers. So, in general, administrative problems could be intentional, or they could be blamed on lack of capacity or experience.
TABLE 5.1
Main categories of irregularities
|
Legal problems Pre-election administration Election day administration Pre-election cheating Election day cheating Pre-election violence Election day violence |
Legal problems are in a separate category. First, they do not occur at a specific time, so one cannot categorize them as pre-election or election day problems. Second, the degree to which their intent is discernable varies. Limiting the scope of an elected office is clearly part of a design to monopolize power, as is disenfranchising parts of the population, but other legal problems stem from inexperience with elections. Newly established democracies may take some time to develop proper campaign finance laws, to institute media regulations, to institutionalize the appointment of electoral officials, and so forth. In this analysis, then, only elections with overt restrictions on the scope of the office, voting rights, and so forth were classified as having a high level of legal problems that could be classified as overt cheating that therefore has a higher risk of drawing monitor denouncement.
As discussed earlier, riskier irregularities are those that monitors are more likely to denounce, while safer forms are those that monitors are more hesitant to denounce. Based on the discussion above, if the argument is that politicians change the timing of the manipulation, then safer tactics are all those that fall under pre-election administration and pre-election cheating, while riskier forms are election day administration and cheating. If the argument is that politicians change the type of their manipulation, the safer tactics are irregularities that pertain to the pre-election and election day administration, while riskier forms are pre-election and election day cheating as well as overt legal irregularities. In addition, pre-election violence may be considered a safer form of irregularity.
It is now possible to examine whether the data on irregularities produce the patterns discussed earlier that would be expected if monitors were causing evasive behaviors. The first part of the investigation looks only at monitored elections. This makes it possible to take advantage of the higher density of the data in the election monitoring reports. Furthermore, multiple organizations are often present, which improves the reliability of the data. The second part of the investigation compares patterns in monitored and nonmonitored elections based on the U.S. State Department reports.
For overview, Figure 5.1 shows the distribution of the various types of irregularities reported by monitors. The most common type of irregularity is administrative problems, but overt cheating and legal problems are nearly as widespread. Aside from violence, which is least common, other irregularities appear with remarkably similar frequency.
Indeed, if one considers type of irregularities, then nearly one-third of monitored elections undertake the highest levels of at least one of the riskier manipulations, pre-election cheating, election day cheating, or legal problems.15 If one instead considers timing, then nearly one-quarter of all elections had the highest levels of either election day cheating, or election day administrative problems, or both. These figures alone do not prove that some politicians are not changing tactics to evade criticisms, but they do show that, if politicians are trying to shift away from riskier types of irregularities or election day irregularities, then many are doing a poor job. Changing tactics may indeed be difficult.
The first argument was that if monitors cause politicians to switch to safer forms of cheating, then when safer forms of cheating occur, riskier forms of cheating should be less common. In the data from the monitoring reports, however, safer and riskier forms of cheating correlate positively. Most elections with many pre-election problems tend to also have many election day problems and those with few pre-election problems tend to hold better election days. The same relationship exists between administrative problems and more overt cheating. Elections with a lot of overt cheating and legal problems also have a lot of administrative problems and those with little overt cheating also have fewer administrative problems.
Figure 5.1: Incidence of various election irregularities in monitored elections as reported by monitoring reports
Number of observations for which data were available for these variables = 287 to 292 elections. Reports were commonly missing from the OAU/AU, the ECOWAS, and the UN. La Francophonie reports were also not coded on these detailed variables. Because these organizations operate in less democratic elections, Figure 5.1 may underestimate the incidences of different types of irregularities.
One way of looking at this more in depth is to consider only the elections where safer irregularities were high, because those are the elections where politicians may be using safer irregularities to evade criticism. Were riskier irregularities low in these elections, as a pattern of evasion would suggest? If one first considers types of irregularities, then sixty of the monitored elections, or about 21 percent, had the highest level of at least one of the two safer administrative forms of irregularities. That is, these elections can be labeled cases where safer forms of irregularities were particularly high. However, of these sixty elections, a full two-thirds also had the highest level of at least one type of risky irregularity. This is unexpected if the countries were strategically shifting from riskier to safer forms of fraud. Another 25 percent had a moderate level. Indeed, only four of the sixty elections, or less than 10 percent, had a low level of all the riskier forms of irregularities.
The patterns do not change much if instead one examines the timing of irregularities. Here, if one leaves out legal problems and violence and looks only at administrative and cheating problems that occur on the day of the election or in the pre-election period, it turns out that of the seventy-eight elections that have the highest level of at least one form of pre-election irregularity, only 14 percent have a low level of both of the election day irregularities. The concurrence of high levels of legal problems and high levels of obvious cheating is equally striking.
Finally, a similar picture emerges when one considers pre-election violence. Unfortunately election-related violence occurs regularly, even in societies that are not normally embroiled in conflict. Indeed, pre-election violence is much more common than election day violence or post-election conflict. About one-fifth of all the elections included in this study had high pre-election violence, which was often deadly. In terms of types of irregularities, of the forty-five monitored elections where pre-election violence was moderate or high, at least one of the forms of obvious cheating was also high in thirty-one elections, or nearly 70 percent of the time. In terms of timing, at least one form of pre-election irregularity was also high in half the elections. In just ten of the elections—or less than one-quarter—was high pre-election violence accompanied by only minor forms of other irregularities.
So, the available evidence suggests that monitored countries that use safer forms of manipulations tend to also use risky forms of manipulation. Only a few countries use high levels of safe manipulation combined with low levels of risky irregularities. Nevertheless, this is the combination of irregularities that would be most expected if indeed the presence of monitors were prompting politicians to switch from risky to safer forms of fraud. Rather, in monitored elections, the reality seems to be that bad things tend to go together. That is, safer and riskier forms of irregularities are complements, not substitutes, as would be expected if politicians were shifting into safer forms of manipulation to avoid criticism.
The second argument to consider was that if monitors cause politicians to switch into safer forms of cheating, then increases in safer forms of cheating should correlate with decreases in riskier forms of cheating. As opposed to the static comparison of levels of different irregularities above, this claim is about changes in levels from one election to another. The election data based on the U.S. State Department reports facilitate comparison of the correlations for both monitored and nonmonitored elections.
To focus on changes, my analysis ignored all the cases where nothing changed from one election to another, and instead focused on the elections where the patterns of irregularities varied. Although this substantially reduced the sample of elections, it illuminated whether at least a subsample of politicians shifted away from risky behaviors to safer ones, and if so, whether this was specific to monitored elections or occurred more generally.
Figures 5.2 and 5.3 show how shifts in one type of behavior correlated with shifts in another type for both monitored and nonmonitored elections.16
In terms of types of cheating, the likely scenario for a strategic shift would involve countries where overt types of cheating decrease on or before election day, or both.17 Figure 5.2 examines this subset of cases. The darker of the central set of columns shows that when monitored countries decreased overt cheating, they most commonly did not increase safer administrative irregularities. On the contrary, nearly 50 percent committed the same level of administrative irregularities and nearly 40 percent actually reduced their administrative irregularities as well. Indeed, countries that decreased overt cheating were more likely to also decrease administrative irregularities than increase them, although in about 10 percent of elections increases did occur. Moreover, monitored countries were not more likely than nonmonitored countries to switch into administrative irregularities. If anything, countries that cut back their overt cheating also reduced their administrative problems more often when international monitors were present.
Because one might consider legal problems another type of overt cheating, Figure 5.3 inspects what happens when legal problems decrease. Figure 5.3 shows that when countries decreased their legal deficiencies, they did not typically shift to safer administrative problems.
Figure 5.4 examines timing: Did governments substitute election day cheating for pre-election cheating? Again, this was not the most common pattern. It did occur sometimes, but notably, politicians who decreased election day cheating increased their pre-election cheating more often when monitors were not present, suggesting that even if this pattern occurred sometimes, monitors were not to blame for the shift.
Finally, Figure 5.5 examines what happens to pre-election violence when overt cheating on election day decreases. Here the monitored and nonmonitored elections show nearly identical patterns, and pre-election violence is as likely to decrease as to increase when countries use less overt election day cheating.
One could continue to examine other shifts, but they tend to show similar patterns: Different types of problems tend to go together. Safer and riskier forms of irregularities are complements, not substitutes. Monitored countries do not behave markedly different from nonmonitored.
Figure 5.2: Changes in administrative irregularities when overt cheating decreases
Number of observations = 100.
Reproduced with permission from Kelley 2012.
To further determine whether there is great cause for concern that monitors prompt a switch toward safer forms of fraud, it is informative to look at a few cases. The short list of monitored elections that fit the profile of having the highest level of safer, administrative types of irregularities and a low level of risky overt cheating comprises Lesotho in 1993, El Salvador in 1994, Guyana in 1997, and Suriname in 2000.18 Are these elections possible cases of monitor-induced shifts in patterns of cheating?
Lesotho’s 1993 election was a first multiparty election in a country unaccustomed to running elections. Thus, the low level of risky fraud was likely due to goodwill to run a clean election, whereas the high level of safer irregularities very well could have been due to administrative inexperience.
In El Salvador the monitoring reports covered election problems extensively. However, the pattern was unsupportive of the notion that monitors induced a shift in type of irregularities. Appendix E discusses the case more extensively, but basically monitors reported low levels of risky cheating with moderate or high levels of safer irregularities throughout the 1990s. Improvements began with the gradual erosion of support for the Alianza Republicana Nacionalista (ARENA), domestically as well as by traditional supporters such as the United States. Thus, it does not seem that the 1994 election with its high level of safer irregularities was really a shift away from risky forms of cheating.
Figure 5.3: Changes in administrative irregularities if legal problems decrease
Number of observations = 74.
Reproduced with permission from Kelley 2012.
Guyana held two elections in the 1980s, one monitored (1980) and one not monitored (1985). A similar pattern of risky fraud and low levels of safer forms of fraud characterized both elections. The presence of monitors in one and not in the other thus did not appear to change the pattern of irregularities. It is possible that reporting at the time was not extensive enough to pick up on other types of problems or the problems were considered irrelevant given the high levels of obvious cheating. The elections in 1992 and 1997 were both monitored and were characterized by lower levels of risky problems while higher levels of safer problems persisted, so these elections could perhaps fit the profile. But then, in the monitored election in 2001, the risky problems in the pre-election period increased again. All these elections are characterized by violence, particularly on election day. The lack of a shift between the monitored and nonmonitored elections in the 1980s, combined with an increase in risky cheating in the monitored election in 2001, contradicts the idea that the patterns observed in the 1997 election resulted from a strategic shift due to the presence of monitors.
Suriname comes closer to a pattern of a strategic use of safer irregularities, because in the 1987, 1991, 1996, and 2000 elections the level of safer types of fraud grows steadily in the presence of monitors. However, even here, this movement does not coincide with a clear shift away from more risky types of cheating, because, perhaps with the exception of the 1992 election, overt and risky cheating was never reported as being particularly prominent.
Figure 5.4: Changes in pre-election cheating if election day cheating decreases
Number of observations = 117.
A similar examination can be done of the elections that displayed a pattern of low election day problems but high pre-election problems. The list of elections that fit this pattern includes Belarus in 1995, Colombia in 2002, Croatia in 1997, Guatemala in 1985 and 1990, Liberia in 1997, Paraguay in 1988, Russia in 1999, Uganda in 1996, and Zimbabwe in 1985. Do these elections suggest that these countries were reacting to the presence of monitors by shifting their manipulation to the period to the pre-election period when fewer monitors were present?
Not really. For most of these countries the combination of low election day irregularities and high pre-election irregularities look like an aberration from a trend where these problems either appear together or not at all. Gross levels of both election day and pre-election day problems, for example, have characterized the last three elections in Belarus, although they have all been monitored. Similarly in Guatemala: After elections in the 1980s, which were characterized by high pre-election problems, in the 1990s Guatemalan elections improved—even the 1994 election, which was not monitored, was decent. The monitored election in 1995 worsened a bit, and then the monitored election in 1999 improved again. But then in 2003, election day problems were again quite high—despite the presence of monitors. Thus, there is no clear pattern that suggests the fraud was being shifted from one period to another in reaction to the presence of monitors. And although the problems in Zimbabwe’s 1985 election were mostly confined to the pre-election period, this too appears to be an aberration. The widespread problems in 2000 and 2002 suggest that Zimbabwe has not learned to manipulate the observers by cheating only in the pre-election period.
Figure 5.5: Changes in pre-election violence if election day cheating decreases
Number of observations = 97.
Colombia held two elections in 2002, a legislative election in March and a presidential election in May. Both elections had high pre-election levels of intimidation. Both elections were monitored. The presidential election likely had a lower level of election day problems because of the outpouring of criticisms of the legislative election, rather than a strategic shift into pre-election irregularities. The Organization of American States (OAS) report talks about the attention paid to the parliamentary election by the media, which criticized officials for ignorance of the law, the lack of transparency in the selection of election judges, the lack of regulation of pre-count procedures, and so on. The OAS criticism of the legislative election may also have contributed to a better-run presidential election day, but that hardly means that the monitors caused a strategic shift from high election day problems into high pre-election irregularities, given that pre-election irregularities were high in both elections.
The 1999 election in Russia is the classic example of a smooth election day but heavily manipulated pre-election environment. Indeed, the Kremlin may have intentionally used and continued to rely on such hefty manipulations of the pre-election environment to minimize monitor criticisms. Russia has held seven monitored elections, and most of them have had problems in the pre-election period or on election day. However, it seems that the pre-election irregularities are simply the style of Russian “managed” elections that has earned Russia’s the title of a “managed democracy,” and that it serves primarily to manipulate the Russian voters rather than deceiving Western observers. Indeed, Russia’s recent efforts to discourage OSCE participation altogether suggest that by now the Kremlin is not very interested in Western endorsements.19
This chapter began by asking whether the expected presence of international monitors causes politicians to switch to forms of manipulation that monitors may be less likely to denounce. It argued that for this to at least be plausible risky and safer irregularities should be substitutes, not complements. However, the available data do not support this. In contrast, the exploration of the data suggests that risky forms of cheating are common in the presence of monitors; that safer and riskier irregularities are positively correlated; and that shifts away from risky irregularities do not coincide with shifts into safer forms. The examination of the individual countries where an election had displayed few risky but many safe forms of irregularities also suggested that these election patterns most likely were not the result of a shift in response to the presence of international monitors.
It is a concern that politicians may use violence in the pre-election period to unnerve the international monitors and the international community such that they are more likely to accept the election outcome. However, because a high level of other problems usually accompanies pre-election violence, politicians do not appear to be using pre-election violence instead of other forms of cheating. Pre-election violence is more common in monitored elections than in nonmonitored elections, but it is unclear whether monitors actually increase the use of pre-election violence. In a simple regression analysis, a history of pre-election violence is the main predictor of pre-election violence and the presence of monitors is not statistically significant. This suggests pre-election violence may simply be part of the political culture or of a larger historical pattern of gross irregularities intended to steal the election.
Overall, the data suggest that—whether elections are monitored or not—politicians who cheat are by and large neither subtle nor safe. The data in this analysis is naturally limited, and many irregularities likely go unobserved. However, it is unlikely that there is a form of clandestine cheating that is unique to monitored elections, because if such clandestine cheating were easy to do, then all politicians would have an incentive to use it to win elections and overt cheating should decrease. In reality, all forms of fraud carry risks of detection, and the high level of reported fraud suggests that the risks are quite large. In both monitored and nonmonitored elections a great deal of irregularities and problems continues to occur and be reported. Indeed, rampant cheating persists—a fact also bemoaned by the 2011 Freedom House reports, which even argues that it is getting worse: “To be sure, rigged elections, persecution of dissidents, and rule by executive fiat are not novel developments in these countries. But the violations were carried out with a striking degree of aggressiveness, self-assurance, and disregard for outside opinion.”20
Thus, although Chapter 4 showed that monitors do sometimes falsely endorse problematic elections and that they are more likely to tolerate certain types of irregularities, politicians do not appear particularly adept at exploiting this weakness. Cheating is a difficult business and most cheaters need to avail themselves of a mix of strategies to manipulate the voters as well as the vote. Instead of shifting away from riskier forms of cheating, politicians appear to rely instead on the tactic in Chapter 3, which showed that some countries strategically invite friendly monitors. Why change the way you cheat to try to evade monitor criticism if it is possible to simply invite friendlier monitors? This does not mean that international monitors should not remain alert to efforts to cheat in new and creative ways. Indeed, their vigilance is necessary to prevent such strategies from becoming successful.
