assault to the department of justice because the methodology for counting aggravated assaults in the ucr reports does not include domestic violence cases. commissioner: i am sorry. can i just ask you a question before you go on? >> yes. thank you. commissioner: i understand that we would hope and expect that the lines from what we report are the same, but the department of justice defines of violent crime differently. is it not by definition going to have to be different? >> we looked into this and compared the data, and the definitions are, indeed, the same. there are characteristics that make these numbers slightly different, but they should not account for the disparity that you see in this graph. commissioner: ok, i thought you just said they do not recognize domestic violence as a part 1 crime while we do. did i miss here you? >> no, they do. aggravated assaults, that should be accounted for in the aggravated assault number that is reported to the department of justice through the ucr report. commissioner: and we are not reporting it because -- >> the methodology does not include do

that the blue line typically produced higher totals of part one violent crime than the green line, ucrditionally, the trends do not always the mine, which is called out in the orange call out. the department so intention is for both of the numbers to be relatively comparable. as a result of the comptroller's office analysis into why these lines look different, we identified three key findings that compromise both the accuracy and consistency in crime reporting by the department. first, the department does not have a single system of record for reporting crime statistics. rather, there are multiple data systems that may cause fluctuations that are not the result of underlying crime trends. an example of a data source weakness is that accounts for homicides, rapes, and shootings are maintained manually because the other systems cannot maintain their accounting. secondly, the actual development of the profile itself includes complex manual processes by the unit. as a result, the profile is prone to human error. lastly, as shown in the grass, gis reports of part one crime numbers are telli