SUBJECT: SURVEY RESPONSE RATE PARAMETERS
NCES STANDARD: 2-2
PURPOSE: To specify design parameters for survey response rates. High
survey response rates help to ensure that survey results are representative
of the target population. Surveys conducted by or for NCES must be designed
and executed to meet the highest practical rates of response. To ensure that
nonresponse bias analyses are conducted when response rates suggest the potential
for bias to occur.
KEY TERMS: assessment,
cross-sectional, key variables,
longitudinal, nonresponse bias,
stage of data collection,
target population, and universe.
STANDARD 2-2-1: Universe data collections must be designed to meet a target
unit response rate of at least 95 percent.
GUIDELINE 2-2-1A: A unit level nonresponse
bias analysis is recommended in the case where the universe
survey unit response rate is less than 90 percent. (See Standard 4-4
for a discussion of nonresponse bias analysis.)
STANDARD 2-2-2: Sample survey unit response rates must be calculated without
substitutions (see Standard 1-3). NCES sample survey data collections must be
designed to meet unit level response rate parameters that are at least consistent
with historical response rates from surveys conducted with best practices.
- For longitudinal sample surveys, the target school level unit response rate
should be at least 70 percent. In the base year and each follow-up, the target
unit response rates at each additional stage should be at least 90 percent.
- For cross-sectional samples, the target unit response rate should be at least
85 percent at each stage of data collection.
- For random-digit dial sample surveys, the target unit response rate should
be at least 70 percent for the screener and at least 90 percent for each survey
- For household sample surveys, the target response rates should be at least
90 percent for the screener, and at least 85 percent for the respondents.
- For assessments, the target response rate should be at least 80 percent for
schools and at least 85 percent for students.
Stage-Specific Design Response Rates by Type of Survey
Design Response Rates
STANDARD 2-2-3: NCES sample survey data collections must be designed
to meet a target item response rate of at least 90 percent for each key item.
STANDARD 2-2-4: A nonresponse
bias analysis is required at any stage of a data collection
with a unit response rate less than 85 percent. If the item response
rate is below 85 percent for any items used in a report, a nonresponse
bias analysis is also required for each of those items (this
does not include individual test items). The extent of the analysis
must reflect the magnitude of the nonresponse (see Standard 4-4).
In longitudinal sample surveys, item nonresponse bias analyses need only be
done once for any individual item, unless there is a substantial deterioration
in the item response rate.
STANDARD 2-2-5: In cases where prior experience suggests the potential
for an overall unit response rate of less than 50 percent, the decision to proceed
with data collection must be made in consultation with the Associate Commissioner,
Chief Statistician, and Commissioner.