Find a section
Find an answer
Start with the question you have about a file. Each link goes to a specific section of the book.
Browse all chaptersDefine the quantity
- Why did my turnout denominator change?Separate the Census convention from a respondents-only rate.Chapter 1 · What Are We Trying to Estimate?Also: What missing answers allow
- Which weight goes with the unit I am counting?Choose whether a rate describes households, people, children or days.Chapter 1 · What Are We Trying to Estimate?
Declare the survey
- Where is the guide for my survey?Compare the designs before opening one of six release-specific guides.Chapter 8 · Field Guide: Six Surveys, Declared
- Which weights declare the ACS design?Find the ACS person or household weight and its replicate weights.Chapter 8 · Field Guide: Six Surveys, DeclaredAlso: What replicate weights change
- Which weight for a NHANES examination?Start with the NHANES release guide and the measurement-specific weight.Chapter 8 · Field Guide: Six Surveys, Declared
- How should I declare BRFSS strata and PSUs?Check the survey design and the rule for single-PSU strata.Chapter 8 · Field Guide: Six Surveys, Declared
Estimate and report
- Can I trust the standard error for a small subgroup?Estimate a domain without deleting other sampled records first.Chapter 3 · Point Estimates and Design-Based Uncertainty
- How do I estimate a difference between groups?Account for both groups being measured within the same survey.Chapter 3 · Point Estimates and Design-Based Uncertainty
- Why do weighted and unweighted coefficients differ?Identify the population projection and the observations entering it.Chapter 4 · What Are We Weighting For?Also: Implicit regression weights
- Should I cluster by state or by county?Trace the sampling and assignment units before choosing a variance rule.Chapter 6 · Sampling Design, Assignment Design, and Clustering
- Are there enough treated clusters?Inspect the limits of the policy comparison and its uncertainty checks.Chapter 6 · Sampling Design, Assignment Design, and Clustering
Check identification
- What can unanswered turnout questions establish?See the range the file supports before imposing a response assumption.Chapter 9 · Can These Data Identify the Quantity?
- What if my comparison has no overlap?Understand why an empty cell cannot supply an observed contrast.Chapter 9 · Can These Data Identify the Quantity?
New to the subject? Start with Chapter 0.