Statistics and Data Science Subject Guide

When deciding whether to use a specific dataset, begin by considering some key questions. This will help you ensure that the data you use are credible and appropriate for your project.

Source

• Who collected these data? An individual? Organization? Agency? 
  • The data source and the reporter or citer are not always the same. For example, advocacy organizations often publish data that were produced by some other organization. When feasible, it is best to go to the original source (or at least know and evaluate the source).
• If the data are repackaged, is there proper documentation to lead you to the primary source? Would it be useful to get more information from the primary source? Could there be anything missing from the secondary version?

Authority

• How widely known or cited is the producer? Who else uses these data?
• Is the measure or producer contested?
• What are the credentials of the data producer?
• If an individual, are they an expert on the subject?
• If an individual, what organizations are they associated with? Could that association affect the work?

Objectivity & Purpose

• Who sponsored the production of these data?
• What was the purpose of the collection/study?
• Who was the intended audience for or users of the data?
• Was it collected as part of the mission of an organization? Or for advocacy? Or for business purposes?

Currency

• When were the data collected?
  • This is not always close to when they were released or published -- there is often a time lag between collection and reporting because of the time required to analyze the data.
• (If relevant to your project): Are these the most recent figures?
  • Sometimes the most recent available figures are a few years old. That is generally okay, as long as you can verify that there isn't something newer.

Collection Methods & Completeness

• How are the data collected?
  • Even a reputable source and collection method can introduce bias. Crime data come from many sources, from victim reports to arrest records.
• What populations included? Excluded?
• (If relevant to your project): What was the sample size? How does that compare to the size of the population it is supposed to represent?
• (If relevant to your project): What methods used to select the population included? How was the total population sampled?
• (If a survey): What was the response rate?

Consistency / Verification

• Do other sources provide similar numbers?
• Can the numbers be verified?

"Evaluating Data: Questions to Consider" is adapted from "Factors to Consider When Evaluating Statistics" created by Kristin Partlo, used under CC-BY-NC 4.0.  "Evaluating Data: Questions to Consider" is licensed under CC-BY-SA 4.0 by Audrey Gunn.