---
title: "Making Sense of Business Process Descriptions: Graphical vs. Textual (2012)"
type: source
tags: [bpm, bpmn, use-cases, understandability, pragmatic-quality, empirical, experiment]
authors: [Ottensooser, Avner; Fekete, Alan; Reijers, Hajo A.; Mendling, Jan; Menictas, Con]
year: 2012
venue: "Journal of Systems and Software 85(3):596–606"
kind: paper
raw_path: "raw/Process Frameworks & BPM/Making Sense of Business Process Descriptions- An Experimental Comparison of Graphical and Textual Notations (2012).pdf"
sources: []
key_claims:
  - "Controlled experiment with 196 students across USYD, TU/e, WU Vienna (later HU) comparing BPMN diagrams to Cockburn-style written use cases (UC) for process understanding."
  - "H1 (UC increases understanding) and H5 (UC followed by BPMN beats UC alone) supported at p<0.01 for BOTH business-analyst proxies (trained) and business-user proxies (untrained)."
  - "H2 (BPMN increases understanding) supported only for trained BA proxies (p<0.0001); NOT supported for untrained BU proxies (p=0.1540)."
  - "H4 (BPMN contributes more than UC) only weakly supported (p≈0.054) for BAs; not for BUs."
  - "Reading UC first then BPMN yielded the maximal scores across all reader communities — a joint textual+graphical presentation dominated either alone."
  - "Untrained readers do NOT gain significant understanding from BPMN on their own; training in the notation is required before graphical models communicate to them."
  - "Participants who often worked with flowcharts/BPMN reported stronger agreement with graphical over textual — familiarity effect on preference."
created: 2026-04-15
updated: 2026-04-15
---

# Making Sense of Business Process Descriptions (Ottensooser et al. 2012)

## Summary
Ottensooser, Fekete, Reijers, Mendling & Menictas report a randomised, controlled, within-subjects experiment testing whether **BPMN diagrams** or **written use cases** (Cockburn-style structured text — trigger, primary actor, main success scenario as numbered steps, extensions) better convey a business process to readers. 196 university students in three cohorts (TU/e n=74+55, USYD n=19+22, WU Vienna/HU n=13+13) served as proxies for business analysts (BA — trained in process modelling) and business users (BU — untrained). The domain was a hypothetical financial-services company; artefacts were six models of comparable complexity. Each participant answered a placebo questionnaire (domain knowledge only), then QSet1 after the first artefact, then QSet2 after the second (opposite notation), enabling measurement of **primary** (single-notation) and **secondary** (joint) contributions.

## Method
- Independent variable: treatment order (BPMN-first vs. UC-first).
- Dependent variables: primary contribution = QSet1 − Placebo; secondary contribution = QSet2 − QSet1.
- Hypothesis testing via one-sided Wilcoxon sign-rank tests (confirmed with t-tests); Shapiro-Wilk normality satisfied.
- Cockburn written use case = structured natural-language template ([[frameworks/written-use-case]]).

## Headline findings (Table 2, §4.3)

| Hypothesis | BA proxies | BU proxies |
|---|---|---|
| H1: UC increases understanding | 0.0000 ✓ | 0.0054 ✓ |
| H2: BPMN increases understanding | 0.0000 ✓ | 0.1540 ✗ |
| H4: BPMN > UC (primary) | 0.0542 ~ | 0.7729 ✗ |
| H5: UC→BPMN > UC alone | 0.0066 ✓ | 0.0003 ✓ |
| H6: BPMN→UC > BPMN alone | 0.9114 ✗ | 0.4521 ✗ |

**Key takeaways:**
- **Textual use cases worked for everyone.** Both trained and untrained readers significantly improved domain understanding from reading UC.
- **BPMN worked only for the trained.** Untrained business users showed no significant gain from BPMN alone.
- **Joint presentation, UC first, dominated.** Reading UC then BPMN improved scores significantly for both groups; the reverse order (BPMN first, then UC) did NOT significantly improve over BPMN alone (H6).
- Self-reported familiarity with flowcharts/BPMN correlated (r≈0.17, p≈0.019) with stronger preference for graphical notation — a notation is a skill, not an intuitive medium.

## Interpretation
The authors conclude that **BPMN alone is not a sufficient communication medium for stakeholders without formal training**, and recommend either (a) training readers before exposing them to BPMN, or (b) pairing BPMN with an upfront written use-case presentation. The maximal-score configuration across all reader communities was **UC → BPMN**.

## Connections
- Directly refines [[concepts/process-model-quality]] §pragmatic quality with empirical evidence.
- Supports the validation rule "translate BPMN to natural language before showing to a domain expert" from [[sources/2018-dumas-fundamentals-of-bpm-ch5-discovery]] §5.1.2, §5.4.2 — see also [[syntheses/interview-structuring-for-process-models]] §4.1.
- Adds nuance: for *comprehension* (not only validation), **written text first, then diagram** is empirically superior — not merely an acceptable fallback.
- Authors already in the wiki: [[entities/hajo-reijers]], [[entities/jan-mendling]].
- Introduces artefact [[frameworks/written-use-case]] (Cockburn 2000) *(referenced-not-ingested for Cockburn book itself)*.