A Practical Guide to the Why-Because Analysis Method

Performing a Why-Because Analysis

Thilo Paul-Stve
September 21, 2005
This guide to the Why-Because Analysis (WBA) method concentrates on
the detailed steps necessary to perform a WBA. The data to create this guide
and the flow charts were determined by Hierarchical Task
Analysis [Paul-Stve 05]. The flow charts follow the IBM Flowcharting Techniques guide, which complies with the requirements of the ISO standard
[IBM 69].

Overview
A Why-Because Analysis (WBA) [Ladkin 01, Ladkin 02] starts with gathering information about the incident (Figure 1). This information is then used to construct either a
List of Facts (facts listed alone) or a Why-Because List.
The construction of the Why-Because Graph (WB Graph, WBG) starts with determining the mishap (the top node). Then the necessary causal factors (NCF) that finally
led to the mishap are determined, using the Why-Because List, until a chosen level of
detail is reached. Finally, the quality (correctness and explanatory completeness) of the
WBG is assured by detecting and correcting errors. A report can then be written using
the WBG.
The WBA process is factored here into eight subprocesses, explained using flowchart
notation.

Gather Information
As shown in Figure 2, the first step is information gathering. First, the sources of information must be identified. These can be, for example, witness reports, responsible
authorities, or applicable documentation. (It has proven useful to get printed copies of
the material.)

The quality of the information must be assessed. Checking the sources and doing
some background research helps. If a team is performing the WBA, the information
material can be discussed. Finally, the useful information material is selected.

Determine the Facts

The selected information material is read again in-depth to identify the statements that
concern the course of events. These statements are split into single events. Presumptions must also explicitly be identified in order to extract the facts! (Figure 3) There
are at least two ways to arrange the facts, both shown in Figure 4: one may apply the
Counterfactual Test earlier (to form a Why-Because List) or later (one creates then a List
of Facts).

Create a List of Facts

A List of Facts is a collection of all the facts that might be relevant to the incident.
Every fact determined is written down with (at least) a serial number, a brief description
suitable for a title, and a reference to its origin.

Create a Why-Because List

A Why-Because List incorporates information about the facts and their relations to each
other. These relations are expressed in Why-Because pairs of facts. First, every fact is
noted with a serial number, a description, and its reference. When all facts have been
recorded this way, the Counterfactual Test is applied to every pair. If there are any
discrepancies, the Why-Because List has to be corrected. Finally the List is checked for
completeness and consistency and again corrected if necessary.

Create an Auxiliary List of Facts

An Auxiliary List of Facts is optional, but often helpful to an understanding of the
incident (see Figure 5). After having created a List of Facts or a Why-Because List,
a classification system, such as selection according to time (to create later a Timeline)
or according to the actors involved (to create later a Time-Actor Diagram, or TAD), is
chosen. Then every fact is filed under its class. Again, every fact is noted with a serial
number, a description, and its reference. If there are facts that do not fit the classification
system, the classification must be adapted.

Determine the Mishap / Top Node

The first task in creating a WB Graph is to determine the unwanted event that constitutes the incident, the mishap. This will be the top node of the WB Graph. To determine
the mishap, the facts collected in the List of Facts or Why-Because List are reviewed and
assessed (Figure 6). Often, the mishap will be obvious, especially in transportation system accidents. But in some cases, for example in computer security incidents in which
many interests are involved, it may not be so easy to identify the mishap event.
The mishap is the event or circumstance that most directly caused the loss of resources, e.g., lives or money, that constitute the accident. These facts can be as obvious
as "AC impacts mountain", but it can often become difficult to tell what makes up the
accident. If working in a team, discussing the facts in the group is helpful.
The mishap is inserted as the top node in the WB Graph with a descriptive label and
a reference to the List of Facts or Why-Because List.

Determine the Necessary Causal Factors

Determining the Necessary Causal Factors (NCFs) is an iterative process starting with
the mishap, the top node of the WBG (see Figure 7).
The following procedure is iterated until done. For every fact that is represented by a
node N already in the WB Graph, the child nodes, the necessary causal factors (NCFs)
are determined either from the facts found in the List of Facts or from the pairs of facts
found in the Why-Because List, as follows:
the List of Facts is reviewed and the Counterfactual Test is applied between N
and each other node in the List of Facts; or
all pairs in the Why-Because List are selected which have N as the first item. The
second item of the pair is then an NCF of N.
When working in a team, discussing the selection in the group is helpful.
The NCFs are added as child nodes of the node representing the examined fact with
a descriptive label and a reference to the List of Facts or Why-Because List.
This procedure is iterated until the desired level of detail is reached, or until every
node appears in the graph.

Quality Assurance and Correction of the WB Graph

After having determined all NCFs to reach the desired level of detail, the Causal Completeness Test is applied. The graph is thoroughly inspected to ensure that the incident is
described sufficiently, and that there are no errors. This step is most successful if carried
out in a face-to-face team meeting (see Figure 8).

If inadequacies or errors are found, they are corrected by changing or adding causal
relations, removing nodes, or adding nodes. Adding nodes requires carefully extending the List of Facts / Why-Because List and then returning to the process of determining NCFs (Figure 9).
If it is determined that the quality of the WBG must be improved, the Counterfactual
Test should be applied once again to check the causal relations. When the entire WB
Graph has been checked in this manner, it is finished.

Glossary
Auxiliary List of Facts (auxLoF) Auxiliary List of Facts are optional. The facts are
arranged according to a classification system, such as timestamp or actor involved.
Auxiliary List of Facts help to gain a better understanding of an incident. The facts
are notated with a serial number, a short description, their class, and a reference
to their source.
Causal Completeness Test (CCT) A technical criterion for determining sufficiency of
causal explanation. The CCT applies between a collection A1 , A2 , ....., An of facts
and a fact B. The CCT is satisfied when (a) each Ak is an NCF of B; and (b) the
Causal Sufficiency Criterion holds between the set A 1 , A2 , ....., An and B. The technical definition may be found in [Ladkin 01].
Causal Sufficiency Criterion The Causal Sufficiency Criterion between a set of facts
A1 , A2 , ....., An and a fact B is that, given the world as it more or less is, it is impossible for B not to have happened if all of the Ak have happened. That is, had the
world been just sufficiently different that B did not happen, then at least one of
the Ak (not necessarily the same one for each different circumstance) would not
have happened either. The technical definition may be found in [Ladkin 01].
Counterfactual Test (CT) The criterion for determining a Necessary Causal Factor.
Given two facts, A and B, CT asks whether, if the world had been just sufficiently
different that A had not happened, whether B would have happened anyway.
If B would not have happened in this situation in which A did not happen, the
Counterfactual Test is passed, and A is a Necessary Causal Factor of B.
List of Facts (LoF) The List of Facts contains the significant facts that are causal factors of the incident. The facts are notated with a serial number, a short description,
and a reference to their source.
Necessary Causal Factor (NCF) A fact that causally affects the occurrence of another
fact in the course of events of the incident. This is determined by applying the
Counterfactual Test. In Why-Because Graphs, NCFs are represented by child
nodes.
Topnode The top node of the Why-Because Graph represents the failure of the examined system (mishap).
Why-Because Graph (WB Graph, WBG) The Why-Because Graph shows as edges the
causal relations between the facts, shown as nodes, that led to the failure of a system.
Why-Because List (WB List) The Why-Because List contains the facts that are causal
factors of the incident, arranged in pairs consisting of a necessary causal factor
and its effect. Every single fact is notated with a serial number, a short description,
and a reference to its source.

References
[IBM 69]

IBM. Flowcharting Techniques, GC20-8152-1 edition, 1969. Available

from http://www.fh-jena.de/kleine/history/, accessed 20 September 2005.

[Ladkin 01]

Peter B. Ladkin. Causal system analysis, volume RVS-Bk-05-01.

RVS Group, University of Bielefeld, 2001. Available at www.rvs.unibielefeld.de Why-Because Analysis.

[Ladkin 02]

Peter B. Ladkin. WBA Home Page. Available at www.rvs.uni-bielefeld.de

Why-Because Analysis, June 2002.

[Paul-Stve 05] Thilo Paul-Stve. Formal Task Analysis of Graphical System Engineering Software Use. Rapport technique, RVS Group, Faculty of Technology, University of Bielefeld, March 2005. Available at www.rvs.unibielefeld.de Publications Theses Written in the Group.

incident occurred

Gather Information
(GI)

Determine Facts
(DF)

use List of
Facts?
yes

no

Create List of
Facts (CL)

Create Auxiliary List

of Facts (CA)

Create WhyBecause List (CW)

clear picture of
course of events /
participants?
no
yes
Determine Mishap /
Topnode (DM)

Determine Necessary
Causal Factors (DN)

Assure Quality
(AQ)

Correct WB-Graph
(CW)

create
report

finished

Figure 1: Why-Because Analysis Overview

incident occurred

Gather Information
identify sources of information (consult
(GI)
expert sources and responsible authorities,
do internet research, ...)

get copies

information
material

read information material

check sources of information material,

background investigation

document
sources

sift through
information material

yes

working in a
team?
no

discuss information material

select useful
information material

Determine Facts
(DF)

Figure 2: Gather Information

information
material

Gather Information

information
material

Determine Facts
(DF)

read selected
information material

identify statements
concerning the course
of events

split statements into

single events

sort out
presumptions

use List of
Facts?

Figure 3: Determine the Facts

notated
facts

Figure 4: Create List of Facts / Why-Because List

10
Create Auxiliary
List of Facts (CA)

List of
Facts

notated
facts

no

Determine Mishap /
Topnode (DM)

yes

clear picture of
course of events /
participants?
no

(facts) with serial number,

description, and reference

Create List of Facts

(CL)
write down the remaining events

yes

use List of
Facts?

Determine Facts
(DF)

yes

check for
completeness and
consistency:
sound?

yes

do counterfactual
test: sound?

do counterfactual
test

no

no

determine and write down

cause and effect pairs

correct Why-Because List

Create Why-Because List

(CW)

WhyBecause
List

clear picture of
course of events /
participants?
yes
no
WhyBecause
List

Create Auxiliary List of Facts

(CA)

choose classification system:

time, actors, ...

review List of Facts /

WB-List

List of
Facts

adapt classification system

file every fact under its class

with serial number, description,
and reference

every fact filed?

no
yes

Determine Mishap /
Topnode (DM)

Figure 5: Create auxiliary List of Facts

11

Auxiliary
List of
Facts

Create Auxiliary
List of Facts
(CA)

no

clear picture of
course of events /
participants?
yes

Determine Mishap /
Topnode

review facts

WhyBecause
List

assess facts

List of
Facts

yes
discuss facts in group

working in a
team?
no

determine the fact, that

most directly caused the loss of
resources, e.g., money or lives,
that make up the accident

add fact as topnode to

Why-Because Graph
with describing label
and reference to List of Facts /
Why-Because List

Determine Necessary
Causal Factors (DN)

Figure 6: Determine the Mishap / Top node

12

WhyBecause
Graph

Determine Mishap /
Topnode (DM)

WhyBecause
List

Determine Necessary
Causal Factors
(DN)

CW 5

DN 1:
review List of Facts /
Why-Because List

assess facts

List of
Facts
yes
discuss facts in group

WhyBecause
Graph

working in a
team?
no

identify all necessary causal

factors for examined node

contemplate node to
examine further

add factors as child nodes to

examined node with describing label
and reference to List of Facts / WhyBecause List

WhyBecause
Graph

desired level of
detail reached?
no
yes

Assure Quality
(AQ)

CW Correct WB-Graph
CW 5: extend List of Facts / Why-Because List

Figure 7: Determine the Necessary Causal Factors

13

Determine Necessary
Causal Factors (DN)

WhyBecause
List

Assure Quality
(AQ)

CW 3

AQ 1:
do causal
completeness test

CW 4
AQ 2:
incident described
sufficiently?

List of
Facts

no

yes

CW 1

look at WB-Graph

WhyBecause
Graph

sift through
causalities of graph

yes
discuss causalities
of graph

working in a
team?
no
AQ 3:
errors? unsound?

yes

no

AQ 4:
do counterfactual
test

improve quality?
yes

CW 2

no

yes
create
report

finished

AQ 5:
counterfactual test
passed?

CW Correct WB-Graph
CW 1: detected wrong or missing causal relation?
CW 2: further factors required to explain causal factor?
CW 3: change or add causal relation
CW 4: remove node
CW 5: extend List of Facts / Why-Because List

Figure 8: Assure the Quality

14

no

Figure 9: Correct the WB Graph

15

yes

no

CW 2:
further factors
required to explain
causal factor?

no

AQ 5

yes

yes

detected superfluous
node?

no

CW 1:
detected wrong
or missing causal
relation?

AQ 3

AQ: Assure Quality

AQ 1: do causal completeness test
AQ 2: incident described sufficiently?
AQ 3: errors? unsound?
AQ 4: do counterfactual test
AQ 5: counterfactual test passed?
DN: Determine NCFs
DN 1: review List of Facts / Why-Because List

AQ 4

WhyBecause
Graph

AQ 2

List of
Facts

WhyBecause
List

CW 5:
extend List of Facts /
Why-Because List

CW 4:
remove node

CW 3:
change or add
causal relation

Correct WB-Graph
(CW)

DN 1

WhyBecause
Graph

AQ 1