Blocking support needed

Support
1

Hi, I have an inquiry regarding blocking. I have an optout experiment and a forced experiment for forced choice tasks if respondents optout. When I generated a design using Ngene, I noticed that there are 24 rows. I only have 8 scenarios. With the forced choice task, I had to assign design rows so that the choice task that shows in the optout is similar to what shows up in the forced task. How do I implement blocking so that all 24 scenarios can be presented to the respondents and not just 8? Can I do blocking for this? How will it affect the modeling and analysis?

If yes, my idea is that there will be 3 blocks. I want that the first respondent gets assigned 1 row from block 1, the next respondent gets assigned a row from block 2, and the next gets assigned a row from block 3. How do I go about this?

Design row randomization
2

Hazel,
good question.

Here’s what normally happens with blocking. Respondents get assigned a block (in your ideal case this looks like 3 blocks of 8). They then get each scenario in that block assigned to them.
There are several historical and methoological reasons researcher do this: - one is to control variation and dominance within a block and Ngene assists with this. Another is historical and pre-internet when ‘blocks’ of DCe questions were printed and allocated separately. Some commercial researchers don’t bother with blocking and just randomly allocate.

There is a tutorial called ‘Blocking’ that (preview here ) many other user had success with here Software Tutorials – SurveyEngine GmbH . I suggest you go through it in detail:

Nevertheless here is the workflow.

  1. Create your experiment and determine the blocking you want - in your case 3 blocks of 8 each.

  2. Then use the ‘deck’ element of size 3 and call it ‘block’ - this will do a random-without-replacement selection from the set (1,2,3) and will greatly help with even allocation - as non-completes get their deck allocation recycled.

  3. Then using a derived value called ‘offset’ and calculate the design offset - this is the first row for the block. In your case it should be:
    ($block-1)*8
    This will give you either (0,8 or 16).

  4. Then for each scenario (1 to 8) explicitly allocate the design row as
    $offset +1
    $offset+2
    etc.

As always publish and run simulations to verify this works.
You can also randomise the pages once you’ve got this working.

You can also use this to synchronise your forced and unforced to use the same rows.

4

Hi, thank you for your response. I published my survey and did simulations w 100 bots to see if it works. I noticed that when respondents are assigned to block 2 and 3, there aren’t any responses in the scenarios. What might be the error?

Screenshot 2026-02-04 at 4.51.38 PM
Screenshot 2026-02-04 at 4.51.38 PM2940×1912 371 KB

5

This is actually good - it shows what would have happened if you ‘willy-nilly’ went ahead without testing. So you dodged a bullet.
There is something trivially wrong in your scenario response code (not the experiment or allcation).
You have

scenario1_response =
SE_experiment_data_by_design_row(MUGI DCE duplicate, 1, pref1)

This say’s “Give Lookup response for “pref1” from Experiment MUGI DCE …, absolutedesign row 1.

But as you are blocking - Scenario 1 may not always be row 1 - it could be 9 or 17. You need $offset +1 not just 1

But now the lookup is getting complicated so instead use the (currently undocumented) functional version which is more suited to inline calculations

SE_data_by_design_row($design_offset+1, “pref1”, “MUGI DCE duplicate”);