Blocking and versions in DCE experimental designs
A full D-efficient design may require 20 or more choice sets - too many for any single respondent. Blocking divides the design across respondents without losing efficiency.
How blocking works in DCE experimental design, when it is necessary, and how to implement it in SurveyEngine.
Knowledge Base -> Experiment Design -> Methods & Academic
Ben White, 07.07.2026
The problem of respondent burden
The number of choice sets required for any experiment design increases with the number of attributes and levels. For example, a study with 6 attributes at 3 levels may require 24 choice sets.
You now have a difficult choice - give one set to each person - for which you would need at least (480 people). Or give everyone all 24 sets with only a minimum 20 people, but risk major respondent fatigue.
How blocking maintains statistical efficiency
A blocked design does not reduce total information - it distributes that information across respondents rather than requiring each to provide all of it. With sufficient respondents per block, parameter estimates are as precise as if all respondents had completed all choice sets.
As a rule of thumb, each block should be completed by at least by a student sample of 20 respondents. Ideally this should be 50 for model estimation. With 3 blocks this implies a minimum sample of 60 but ideally 150.
TLDR Quick links
Setting up blocks in SurveyEngine
Step 1: Generate the full design. either an orthogonal or complete D-efficient design for all required choice sets before blocking.
Step 2: Set the number of blocks. Divide total choice sets by the target tasks per respondent. A 20-task design divided into blocks of 10 requires 2 blocks.
Step 3: Assign choice sets to blocks. SurveyEngine's design generator assigns choice sets to blocks to maximise within-block efficiency.
Step 4: Configure random block assignment. SurveyEngine handles this automatically when blocking is configured in the design generator.
Step 5: Check within-block efficiency. Verify each block has acceptable D-efficiency independently.
Worked example - 4-block design for a patient survey
A health preference study with 6 attributes requires 24 choice sets for a D-efficient design. The team judges 12 tasks is the maximum acceptable burden for the target patient population.
The design is divided into 2 blocks of 12, with respondents randomly assigned. With 150 respondents per block (300 total), parameter estimates achieve the required precision. Respondents see a complete 12-task experiment with no indication that others are seeing different tasks.
References
Street, D.J. and Burgess, L. (2007). The Construction of Optimal Stated Choice Experiments. Wiley.
Rose, J.M. and Bliemer, M.C.J. (2009). Constructing efficient stated choice experimental designs. Transport Reviews, 29(5), 587–617.
Ready to set up a blocked design for your DCE? Log in to SurveyEngine and configure your design blocks.
Or Contact us at support@surveyengine.com — we're glad to help.