Lecture 5.5: Data: Reporting

AREC 570 / ECON 530 · Unit 5

Lauren Chenarides

Before class

  • No assigned reading. We will apply the Unit 5 material and the course data checklist.

Readings and course handouts are available on Canvas.

Make the evidence assessable

A reader should understand:

  • Where the data came from.
  • Which population and period they describe.
  • How the analysis sample was constructed.
  • What the key variables measure.
  • Which limitations remain.

Start with provenance

For each source, report:

  • Provider and dataset.
  • Collection purpose and method.
  • Release or retrieval date.
  • Geographic and time coverage.
  • Access conditions relevant to replication.

Cite the dataset as well as papers using it.

Define the unit of observation

“County data” is not enough.

Is one row a county, county-year, county-industry-year, or county-month?

Explain the key and how the observation unit relates to the population in your question.

Show sample construction

Step Observations retained
Raw records 1,000
Keep target period 850
Remove verified duplicates 830
Require observed outcome 790

Illustrative counts. Explain each exclusion and whether excluded observations differ systematically.

Missing is not zero

Distinguish a true zero from:

  • An unanswered question.
  • A suppressed value.
  • An inapplicable question.
  • A failed link between datasets.

Report your treatment of each relevant category.

Explain variable construction

Include:

  • Original field or survey question.
  • Units and reference period.
  • Transformation.
  • Interpretation.

For monetary measures, state whether values are nominal or adjusted; identify the base period if adjusted.

Report merges and aggregation

  • Which identifiers were used?
  • What matching relationship was expected?
  • How many records failed to match?
  • Did merging duplicate observations?
  • How were finer observations aggregated?

State whether weights were used and what they represent.

Descriptive statistics should answer questions

  • What does the sample look like?
  • How much relevant variation is present?
  • Are extreme values plausible?
  • Do treatment and comparison groups differ?
  • Does each row use the same sample?

Include units and sample sizes in table notes.

Workshop: the course data checklist

Complete the three parts:

  1. Scoping the question: mechanism and expected outcomes.
  2. Understanding the data: fields, coverage, unit, key, quality, and cleaning.
  3. Developing measures: concept, advantages, limitations, and validation.

The blank checklist and Champ example are on Canvas.

Critique the worked example

Using the Champ checklist example:

  • Does the willingness-to-pay measure match the question?
  • What is lost by deleting incomplete responses?
  • What justifies calling a value an outlier?
  • How could alternative decisions change the sample?

Assess the consequences of each cleaning rule.

Write one data paragraph

Include source, observation unit, sample boundaries, key variables, an important transformation, and the largest limitation.

Exchange paragraphs. Can your partner reconstruct the sample?

Before you leave

Identify one claim your data support and one they cannot support without additional information.

Course navigation