The Enubiaka XLSForm Repeat Group Standards

A reference guide for selecting the correct repeat group format for your survey design.

📋 Quick Reference Overview

Format	Structure	Default?	Best For
1. Sequential Positional Loop	Unrolled groups by slot position	No	Fixed-count sequential recording
2. Brand-Fixed Conditional Group Loop	One fixed group per brand/item	No	Brand-dedicated column analysis
3. Automated Selection-Driven Repeat Loop	True begin_repeat / end_repeat	No	Variable-count, large lists, long format analysis
4. Smart Auto-Select Default	Engine picks per repeat group	✅ Yes	Not sure which format fits — let the engine decide
5. Direct Scripting Mode	No system scaffolding at all	No	Fully custom, hand-authored repeat logic
6. Enumerator-Controlled Single-Pass Loop	Uncounted repeat, name + details together	No	Open-ended lists with light per-item follow-up
7. Enumerator-Controlled Roster Loop	Two-pass: names first, then details	No	DHS-style rosters with substantial per-item detail

Sequential Positional Loop

A fixed number of unrolled groups (one per slot position) where the enumerator manually selects a brand or item at the start of each group via a select_one question. Already-selected items are excluded from subsequent groups via choice_filter, preventing duplicates. The total number of groups is fixed at design time.

✅ When To Use

When enumerator control over item order is required
When total number of items is fixed in advance
When a guided one-item-at-a-time CAPI experience is needed
When the list is small (ideally 10 items or fewer)

⚠ When NOT To Use

When the number of items varies per respondent
When brand-fixed columns in raw data are required
When the item list is large
When enumerator control over order is not needed

📊 Best Suited Study Types

Product testing studies — handing out multiple product samples sequentially
Sensory evaluation studies — taste tests, fragrance tests, texture evaluations
Mystery shopping audits — visiting a fixed number of outlets one at a time
Ranked preference studies — deliberate item presentation order
Experimental studies with controlled stimulus order

📁 Raw Data Appearance

Exports as a single flat sheet. Columns are organised by slot position, not by item. The same brand can appear in different slot columns across interviews. Cross-respondent analysis by brand requires filtering by slotN_pick value.

Brand-Fixed Conditional Group Loop

Each brand or item has its own permanently fixed group with hardcoded questions and a unique field name prefix. The enumerator selects all applicable brands upfront via a select_multiple question, and only the selected brands' groups appear — always in the same fixed order. No reshaping is needed for brand-level analysis.

✅ When To Use

When each brand must always occupy its own dedicated columns
When cross-respondent brand comparison is a priority
When the brand list is fully known, stable, and small (under 20 items)
When all brands can be selected upfront before questions begin

⚠ When NOT To Use

When the brand list exceeds 20 items
When a guided one-at-a-time sequential flow is needed
When brands are dynamic or change between survey rounds
When respondents have vastly different numbers of applicable brands

📊 Best Suited Study Types

Brand equity and brand health tracking studies
Retail shelf audits — availability, pricing, and placement for a fixed set of SKUs
Competitive benchmarking studies
Media monitoring surveys — recall, sentiment and exposure for defined channels
Policy compliance audits and facility assessment surveys

📁 Raw Data Appearance

Exports as a single flat sheet with brand-dedicated columns. Each brand always appears under the same columns regardless of interview or selection order. Brands not selected are simply blank. No filtering or reshaping needed for brand-level analysis.

Automated Selection-Driven Repeat Loop

A true begin_repeat / end_repeat group driven by a prior select_multiple question. The repeat runs automatically — once per selected item — iterating through each selection in order using selected-at(). The number of iterations adapts to each respondent's selections. This is the recommended default for most survey designs.

✅ When To Use

When the number of iterations varies per respondent
When the item list is large or may change between rounds
When enumerator control over item order is not required
When analysis will be done in long format or reshaped programmatically
When form performance and maintainability are priorities

⚠ When NOT To Use

When brand-fixed columns are strictly required without reshaping
When enumerator control over which item appears next is required
When complex cross-iteration referencing or deduplication logic is needed

📊 Best Suited Study Types

Household surveys — details per household member, crop, asset, or income source
Health and nutrition studies — per child, illness episode, or food item
Agricultural surveys — per plot, crop type, or livestock category
Market surveys with large or variable brand lists
Panel and longitudinal studies with long-to-wide reshaping in Stata/R
Any study where item count is unknown at design time or varies widely

📁 Raw Data Appearance

Exports as two linked sheets — a main sheet (one row per submission) and a repeat sheet (one row per iteration per submission). All iterations share the same column names and are distinguished by the current_item value. Requires filtering by current_item or reshaping in Stata/R to analyse by specific item.

Smart Auto-Select ✅ Default

The engine evaluates each repeat group individually — considering the trigger question's type, the number of options, and (for enumerator-controlled repeats) the number of follow-up questions — and automatically picks the best-fitting format (1, 2, 3, 6, or 7) without requiring the author to choose manually. Different repeat groups in the same questionnaire may resolve to different formats.

✅ When To Use

When unsure which format best fits a given repeat group
When the questionnaire has several repeat groups with different ideal formats
When a solid default is wanted without deep XLSForm/EQDS format knowledge
As a starting point before deciding whether to force a specific format

⚠ When NOT To Use

When a specific, guaranteed format is required for a particular repeat group regardless of its structure
When the raw data shape must be fully predictable and controlled upfront
When the engine's heuristics don't match a study's specific needs (force the format explicitly instead)

📊 Best Suited Study Types

General-purpose default for most questionnaires
Multi-module surveys mixing dedicated brand tracking with variable long-format modules
Teams new to EQDS's repeat format conventions
Any questionnaire where different repeat groups genuinely call for different formats

📁 Raw Data Appearance

Varies per repeat group depending on what the engine resolves it to — some groups may export as flat sheets with dedicated columns (Format 1/2 outcomes), others as separate repeat sheets (Format 3/7 outcomes). Check each repeat group's resolved format individually when planning analysis.

Direct Scripting Mode

No system-generated repeat scaffolding is applied at all. Questions inside a "Repeat group for..." boundary are scripted exactly as written in the questionnaire — group by group, question by question — with no begin_repeat/end_repeat, no slot unrolling, and no brand-fixed groups. The author takes full manual control over any repeat logic.

✅ When To Use

When a highly custom repeat structure is needed that no standard format can express
When the author is comfortable hand-authoring XLSForm logic directly
When nesting repeats in an unusual way the engine doesn't support
When bypassing the engine's assumptions entirely is required

⚠ When NOT To Use

When the engine should handle repeat mechanics automatically
When comfort with raw XLSForm authoring is limited
When one of the standard formats (1, 2, 3, 6, 7) already covers the need

📊 Best Suited Study Types

Highly specialised or non-standard survey logic
Advanced XLSForm authors with specific technical requirements
Edge cases none of the standard repeat formats can express

📁 Raw Data Appearance

Entirely dependent on what the author manually scripts — the engine imposes no predictable structure of its own in this mode.

Enumerator-Controlled Single-Pass Loop

No fixed choice list at all — the enumerator types each item's name on the spot via a text question, then immediately answers the detail questions for that same item, one item at a time. Uses ODK's native uncounted-repeat "Add Group?" prompt, so the enumerator decides in the field whether to continue after each item.

✅ When To Use

When the set of items can't be known in advance (genuinely open-ended lists)
When items are typically few and per-item detail-gathering is simple
When field-level flexibility for the enumerator matters more than raw-data predictability

⚠ When NOT To Use

When the full list of items must exist before any follow-up question (use Format 7 instead)
When the set of items is actually known in advance from a fixed list (use Format 1, 2, or 3 instead)
When detail-gathering per item is extensive — a roster-first approach is usually cleaner

📊 Best Suited Study Types

Crop lists in agricultural surveys with light per-crop follow-up
Asset or inventory listing with a small number of detail questions
Ad-hoc field observation logs
Any open-ended item listing where the enumerator naturally records one item fully before moving to the next

📁 Raw Data Appearance

Exports as a single repeat sheet — one row per item. The item_name column holds the enumerator-typed name alongside that item's detail answers in the same row.

Enumerator-Controlled Roster Loop

A true two-pass, DHS-style roster. All item names are collected first in an uncounted repeat (via the same "Add Group?" prompt as Format 6), then a second repeat — automatically sized to match however many names were entered — asks the detail questions for each named item, pulling back the matching name by position.

✅ When To Use

When the complete list of items should exist before any details are collected
When something else needs to reference the full list — e.g. selecting a household head from the roster just entered
When there is substantial detail-gathering per item across multiple question modules

⚠ When NOT To Use

When items are few and detail-gathering is simple — Format 6 is simpler and sufficient
When there's no need to see the full list before asking about individual items
When the survey is short and a single pass is enough

📊 Best Suited Study Types

Household member rosters — DHS/LSMS-style demographic surveys
Livestock or asset rosters followed by detailed per-item modules
Agricultural plot rosters with detailed per-plot data collection
Any study needing the full item list available before diving into per-item detail

📁 Raw Data Appearance

Exports as two separate repeat sheets — a roster sheet (one row per named item) and a details sheet (one row per item, indexed to match the roster by position). Requires joining the two sheets by position when analysing.

The Enubiaka XLSForm Repeat Group Standards | By Marvis Onyenwenu Enubiaka