Skip to Main Content
University of Phoenix logo
University Library

Evaluate Data

Not all data is reliable. Checking its source, collection method, and accuracy helps you avoid misinformation and build solid, trustworthy conclusions, like inspecting the foundation before building your case on top of it.

Here are things you should consider when evaluating data.

Look closely.

  • Examine charts and tables carefully.
  • Understand what numbers and percentages truly represent.
  • Watch for tricky visuals - check if axes are labeled and scaled fairly.

Use the SMART test.

Make sure the data passes the SMART test before believing any conclusions.

SOURCE

Who or what is the source?

  • Who collected the data?
  • Are they credible?

MOTIVE

What's in it for them?

  • Was the data meant to inform, sell, or persuade?
  • Was the data collected for advocacy or business purposes?
  • Who is the intended audience?

AUTHORITY

Who produced the data/statistics?

  • What are the author's/data collector's qualifications?
  • Are they an expert on this subject?
  • What organizations are they affiliated with?

REVIEW

Have you carefully reviewed the collection methods and completeness?

  • What research methodology was used?
  • How was the data collected?
  • Is the data current enough for your needs?

TWO-SOURCE TEST

Have you double-checked everything possible?

  • Do other sources have similar data on this subject?
  • Do other sources confirm or contradict the data?

Even trusted sources can have hidden bias. Stay curious and skeptical!

Source: University of Washington Libraries. (n.d.). Savvy info consumers: Data & statistics.

Keep an eye out for bias.

Bias in datasets can lead to misleading conclusions. Here are common types to watch out for:

Type of Bias What It Means Example
Sampling Bias The sample doesn't reflect the whole population. Only polling people in your neighborhood about an increase in city transportation costs
Selection Bias Specific data is intentionally or unintentionally excluded. Only including happy employees in a job satisfaction study, testing a new medication only on men without investigating its effects on women
Measurement Bias Errors in data collection distort results. Thermometers consistently reading too high can throw off weather data. Water samples taken upstream of a known pollution source miss vital readings.
Cultural Bias Data reflects one group's values, ignoring others. Studying reading instruction methods from only one country may miss innovations. Surveying only young people who have smartphones may miss lower-income youth.
Label Bias Groups are overgeneralized through labels. Calling all millennials "tech-savvy" without acknowledging diversity, mislabeling people who don't attend follow-up medical appointments as non-compliant without accounting for their reasons for not coming
Social Desirability Bias Participants provide answers they think are socially acceptable. Overreporting healthy habits like exercise or vegetable intake, university students reporting fewer mental health struggles than they have because they fear being seen as less resilient

Being aware of bias helps create more reliable and fair insights. See this video and overview from Sage Research Methods for a more in-depth look at understanding and preventing data bias.