cases from Bulawayo or the two Matabeleland Provinces in either data set. The strongest finding was that the correlation between the two data sets was strong and positive. 57 So both data sets indicate that violations were widespread across Zimbabwe in the period 2000 to 2006. Figure 2 Comparison of press and case reports by Province, 2000 to 2006 70 Percentageof total 60 50 40 press 30 cases 20 10 M B at a b ula w a M elel at an yo ab d el N or el an th d S ou t H h ar M ar M as e ho an na i ca M l an land d as C h en M ona tra as la nd l ho na E a la nd st W es M t as vi ng o M id la nd s 0 Province The same effect was not found for the frequency of reports by year. It can be seen in Figure 3 [over] that the trends from 2001 through 2003 are largely similar, but they diverge strongly thereafter, and, in fact, there is a negative relationship between the two data sets.58 This requires further comment [see below]. 57 Rank order correlation between press and cases reports, frequencies per Province [0.98; p=0.01]. 58 Rank order correlation between press and case reports, frequencies over time [-0.41; non-significant]. Who guards the guards? Violations by the Zimbabwe Republic Police, 2000 to 2006