DEVELOPING INTIMACY WITH YOUR DATA SUBJECT: Police Killings This exercise involves you working with an already acquired dataset to undertake the remaining three key steps of examining, transforming and exploring your data to develop a deep familiarisation with its properties and qualities. This spreadsheet provides you with two contrasting worksheets showing snapshot details of recorded deaths caused by US law enforcement agencies, from The Guardian (“The Counted“) and the Washington Post (“Fatal Force“). For each dataset: Examination: Articulate the meaning of the data (its representativeness and phenomenon) and thoroughly examine the physical properties (type, size, condition) noting down your descriptions in each case. Compare what the two datasets offer and contrast their differences. Transformation: What could you do/would you need to do to clean or modify the existing data? What other data could you imagine would be valuable to consolidate the existing data? Exploration: Use a tool of your choice (common recommendations would be Excel, Tableau, R) to visually explore the two datasets separately in order to deepen your appreciation of their physical properties and their discoverable qualities (insights) to help you cement your understanding of their respective value.

The dataset provided in this exercise is focused on recorded deaths caused by law enforcement agencies in the United States, with two contrasting worksheets from The Guardian’s “The Counted” and the Washington Post’s “Fatal Force.” To develop a deep familiarization with the data, we will undertake the remaining three key steps: examination, transformation, and exploration.

In the examination step, we need to articulate the meaning of the data and thoroughly examine its physical properties. This involves understanding the representativeness of the data and the phenomenon it represents. For example, we need to consider whether the dataset covers all instances of police killings in the United States or if it represents a sample or subset. We also need to examine the type of data, such as whether it is numerical, categorical, or text-based, and the size and condition of the dataset.

Comparing the two datasets, we need to identify the similarities and differences between them. It is essential to note any variations in the scope, methodology, or criteria used to record the deaths caused by law enforcement agencies. For instance, The Guardian’s dataset may include deaths caused by off-duty police officers, while the Washington Post’s dataset may focus solely on on-duty incidents. These differences can significantly impact the analysis and interpretation of the data.

Moving on to the transformation step, we need to consider what cleaning or modifications are required for the existing data. This may involve standardizing the format of variables, addressing missing or inconsistent values, and removing irrelevant or duplicate entries. Making the data consistent and coherent is critical for accurate analysis. Additionally, we should assess what additional data could be valuable to consolidate the existing dataset. This could include demographic information about the victims, the circumstances surrounding the incidents, or data on police training and policies.

Lastly, in the exploration step, we are tasked with using a tool of our choice (such as Excel, Tableau, or R) to visually explore the two datasets separately. This process allows us to gain insights into the physical properties of the data and discover its qualities. By creating visualizations, we can identify patterns, trends, outliers, and other descriptive statistics that help deepen our understanding of the data. This exploration stage aims to uncover valuable insights that will inform further analysis and interpretation.

By completing these three steps, we can develop a deep familiarity with the dataset on police killings. Through examination, transformation, and exploration, we gain a comprehensive understanding of the data’s properties, identify any necessary adjustments, and unearth valuable insights that enhance our understanding of the dataset’s value.