One of the most interesting examples of blogs having political consequences was the resignation last year of Alberto Gonzales as a result of a political scandal that Josh Marshall and other left-leaning bloggers helped precipitate, through uncovering evidence of patterns in hiring and firing of US Attorneys by the Bush administration. This is a politically consequential question in large part because of suggestions that prosecutors may have partisan bias – and that these biases may be reinforced if prosecutors fear that they will be fired for not being partisan enough in their prosecution patterns. But good evidence on the consequences of partisan bias is hard to get, because we don’t have good data on the cases that prosecutors _choose among_ when they decide whether or not to start a prosecution; we only have data on the cases that prosecutors _have chosen._ In other words, we don’t have good information on the cases that prosecutors have decided _not_ to take up.
Sanford Gordon has a “really interesting paper”:http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1166343 at SSRN which attempts to figure out whether US Attorney prosecutions of public corruption cases are indeed biased given available information. His arguments may seem at first counterintuitive – but they seem sound to me. He claims that if Republican US Attorneys are biased so that they are more likely to prosecute Democrats, and Democratic US Attorneys are more likely to prosecute Republicans, we should expect that Democrats will on average receive _lower_ sentences when prosecuted by Republicans, and Republicans will receive _lower_ sentences when prosecuted by Democrats. The reason is straightforward. As Gordon argues, if a prosecutor is enthusiastic to prosecute political opponents, the prosecutor will pursue weaker cases against these opponents than against her political allies. Weaker cases mean, on average, lower sentences.
To test these arguments, Gordon looks at prosecution patterns for a two year period under both the Clinton and Bush II administrations. He finds that, as expected, there is evidence of partisan bias both by Republican and Democratic US attorneys. Both seem inclined to prosecute weak cases against their opponents that they might not pursue if they were cases against political allies. Interestingly, Gordon also finds some reason to believe that his measure under-estimates the degree of bias under the Bush administration than the Clinton administration. In his words:
bq. I also ?nd evidence that defendants with identi?able partisan affiliations are disproportionately Democratic under both administrations. This persistent disproportionality suggests that the ?ndings concerning disparities in sentencing might understate the actual level of partisan bias under the Bush administration (i.e., the true level of bias under Bush is higher than measured), while overstating partisan bias under the Clinton administration (i.e., the true level of bias under Clinton is lower than measured).