Uniform Crime Reporting Program Data: Law Enforcement Officers Killed and Assaulted (LEOKA) 1975-2015

Version 3 release notes: Fix bug where Philadelphia Police Department had incorrect FIPS county code. The LEOKA data sets contain highly detailed data about the number of officers/civilians employed by an agency and how many officers were killed or assaulted. Each data set contains over 2,200 columns and has a wealth of information about the circumstances of assaults on officers. All the data was downloaded from NACJD as ASCII+SPSS Setup files and read into R using the package asciiSetupReader. It was then cleaned in R. The "cleaning" just means that column names were standardized (different years have slightly different spellings for many columns). Standardization of column names is necessary to stack multiple years together. Categorical variables (e.g. state) were also standardized (i.e. fix spelling errors). About 7% of all agencies in the data report more officers or civilians than population. As such, I removed the officers/civilians per 1,000 population variables. You should exercise caution if deciding to generate and use these variables yourself. I did not make any changes to the numeric columns except for the following. A few years of data had the values "blank" or "missing" as indicators of missing values. Rows in otherwise numeric columns (e.g. jan_asslt_no_injury_knife) with these values were replaced with NA. There were three obvious data entry errors in officers killed by felony/accident that I changed to NA. In 1978 the agency "pittsburgh" (ORI = PAPPD00) reported 576 officers killed by accident during March. In 1979 the agency "metuchen" (ORI = NJ01210) reported 991 officers killed by felony during August. In 1990 the agency "penobscot state police" (ORI = ME010SP) reported 860 officers killed by accident during July. No other changes to numeric columns were made. Each zip file contains all years as individual monthly files of the specified data type It also includes a file with all years aggregated yearly and stacked into a single data set. Please note that each monthly file is quite large (2,200+ columns) so it may take time to download the zip file and open each data file. For the R code used to clean this data, see here. https://github.com/jacobkap/crime_data. The UCR Handbook (https://ucr.fbi.gov/additional-ucr-publications/ucr_handbook.pdf/view) describes the LEOKA data as follows: "The UCR Program collects data from all contributing agencies ... on officer line-of-duty deaths and assaults. Reporting agencies must submit data on ... their own duly sworn officers feloniously or accidentally killed or assaulted in the line of duty. The purpose of this data collection is to identify situations in which officers are killed or assaulted, describe the incidents statistically, and publish the data to aid agencies in developing policies to improve officer safety. "... agencies must record assaults on sworn officers. Reporting agencies must count all assaults that resulted in serious injury or assaults in which a weapon was used that could have caused serious injury or death. They must include other assaults not causing injury if the assault involved more than mere verbal abuse or minor resistance to an arrest. In other words, agencies must include in this section all assaults on officers, whether or not the officers sustained injuries." If you have any questions, comments, or suggestions please contact me at [e-mail]

版本3更新说明：修复了费城警察局（Philadelphia Police Department）FIPS县代码错误的问题。
执法人员遇袭与殉职（LEOKA）数据集包含各执法机构雇员（警员与文职人员）规模、警员遇袭及殉职情况的详尽数据。各数据集均包含超过2200个字段，涵盖警员遇袭事件的丰富背景信息。所有数据均从全国犯罪司法数据档案馆（National Archive of Criminal Justice Data, NACJD）下载，格式为ASCII+SPSS安装文件，通过`asciiSetupReader`包读入R语言环境，随后在R中完成数据清洗。
本次清洗工作主要包括两部分：一是标准化字段名称——由于不同年份的字段拼写存在细微差异，标准化字段名称是实现多年份数据堆叠的必要前提；二是对分类变量（如州名）进行标准化修正，包括修正拼写错误。
数据集中约7%的机构上报的警员或文职人员规模超出辖区总人口规模，因此我已移除「每千人警员/文职人员占比」相关变量。若您计划自行生成并使用此类变量，请务必谨慎操作。
除以下情况外，我未对数值型字段做任何修改：部分年份的数据将「空白」或「缺失」作为缺失值标识，针对此类出现在数值型字段（如`jan_asslt_no_injury_knife`，即1月未造成伤害的刀具袭击相关字段）中的此类值，已将其替换为NA值。此外，我修正了3处明显的警员殉职（过失致死/因公殉职）数据录入错误，将其替换为NA值：1978年，机构「匹兹堡」（ORI编号：PAPPD00）上报3月有576名警员因意外致死；1979年，机构「梅特肯」（ORI编号：NJ01210）上报8月有991名警员因重罪袭击致死；1990年，机构「佩诺布斯科特州警察局」（ORI编号：ME010SP）上报7月有860名警员因意外致死。未对数值型字段做其他修改。
每个压缩包包含所有年份的单月数据文件，格式与指定数据类型一致；同时包含一个将所有年份数据按年度聚合并堆叠为单数据集的文件。请注意，每个单月数据文件字段量均超过2200个，体积较大，下载压缩包及打开单个数据文件可能需要较长时间。
若需查看本次数据清洗所用的R代码，请访问：https://github.com/jacobkap/crime_data。
《统一犯罪报告手册（UCR Handbook）》（https://ucr.fbi.gov/additional-ucr-publications/ucr_handbook.pdf/view）对执法人员遇袭与殉职（LEOKA）数据集的描述如下：「美国统一犯罪报告（UCR）项目从所有参与机构收集警员因公殉职及遇袭数据。上报机构必须提交本机构正式宣誓警员在执行公务时遭重罪袭击或意外致死、致伤的相关数据。本次数据收集的目的在于梳理警员遇袭及殉职的场景，对事件进行统计描述，并发布数据以协助各机构制定提升警员安全的政策。」「……机构必须记录针对宣誓警员的袭击事件。上报机构需统计所有造成严重伤害的袭击，或使用了可能造成严重伤害或死亡的武器的袭击事件。若袭击不仅涉及言语辱骂或轻微拒捕，即便未造成警员受伤，也需纳入统计。换言之，机构需将所有针对警员的袭击纳入本节统计，无论警员是否受伤。」
若您有任何疑问、意见或建议，请通过邮箱[e-mail]与我联系。