© 1979-2003 by Ludwig Benner, Jr. .All rights reserved.

Guide 2

TASK GUIDANCE FOR ORGANIZING AND
ANALYZING INVESTIGATION DATA

For Use During MES-Based Investigations



Table of Contents

Blue text indicates clickable links in the version posted at the www.starlinesw.com Web site

Go to Guide: 0 1 2 3 4 5 6 7 8 9 10 | Contact Webmaster

Starline Software Ltd. Home Page

ORGANIZING DATA

An investigator's next challenge is to organize the building blocks into a description of the process being investigated. This organization challenge generally stated is to:

  1. Place EBs on the Time/Actor matrix in their proper sequence
  2. Link interactions to show flow of actions needed to produce outcome
  3. Apply logic tests to the actions displayed
  4. Define gaps in flow of actions
  5. Update display to reflect latest data and analysis

Procedures for this task are contained in this Guide. They cover manual and software-supported procedures.

INTRODUCTION

Guide 1 described how to look for and document ACTIONS as Event Blocks during an investigation. This Guide describes how to organize, relate and test MES Event Building Blocks (EBs) to identify INTERACTIONS. This task is required to develop an objective, consistent description and explanation of what happened.

All these functions take place on special actor/action matrixes, a key data organization and analysis element of the MES investigation technology. These matrixes provide for the progressive organization., display and analysis of data as the data are acquired during an investigation. As each new data item is added to the matrix, the description and explanation of what happened become more complete and coherent. MES matrixes introduce new opportunities to improve investigation efficiency, effectiveness, timeliness and value. This Guide describes the tasks required to prepare and quality check MES Matrixes[1].

MES Matrixes have many uses. They can be used to

show investigation progress during an investigation, and communicate the knowns and unknowns as the investigation proceeds. (MES is a formal "progressive analysis" process.)
define and flag remaining uncertainties about the phenomenon described, in a "research defining" sense.
facilitate, structure and control speculation and testing to resolve gaps and uncertainties.
determine and test relevance and validity of data used, and logic of the process.
discover, define and assess needs disclosed by the occurrence, and their significance relative to expected outcomes.
structure identification, definition and assessment of alternatives to help manage future efforts more effectively.
design a monitoring plan that identifies whether the expected results are being achieved after action is taken.
do objective quality control checks of investigation work products.
aid in the preparation and presentation of training about operations
give designers detailed feedback to improve system designs
share new information about accidents with similar activities, to help them improve performance
provide valid data for analyses of trends and other statistical studies
build a knowledge base about a system and how the system functions
provide inputs to upgrade operating and procedures manuals to reduce future risks.
provide a basis to evaluate codes, standards and regulations

This examples also show how the preparation of good Matrixes can contribute to a comprehensive organizational learning process based on efficient, objective, timely and integrated investigations of all kinds of occurrences in the organization or activity..

GUIDANCE OBJECTIVE

The objective of this Guidance is to explain how an investigator can organize and analyze investigation data by developing, testing and quality checking MES Matrixes. Matrixes are a progressive data organizing and analysis tool, with which an investigator can achieve consistent, valid work products efficiently and quickly. The Matrixes become "events flow charts" describing what happened. Ideally, an MES Matrix provides a description of an occurrence in a form that allows others to reproduce it, much like a musical score or an actor's script enables reproduction of a symphony or play. This Guide describes how to place, analyze, relate, test, and present observed or recorded data during investigations to arrive at complete, replicable, easily communicated and valid explanations of the occurrences.

Completed Matrixes describing what happened will look generally like Figure 2-1, which shows the progression of an process producing an accidental outcome. In Matrixes, each change event is an EB created from data observed and documented during the investigation[2].

Figure 2-1 General Appearance of MES-based Matrix.

EBs on the final Matrix have links to show interactive relationships among EBs that produced the outcome. Uncertainties, if any, are indicated with a "?" convention, either in the EB or along the links.

APPLICABILITY

MES Matrixes can be used in all investigations, but they are most valuable during investigations of unprecedented, large, complicated or mysterious occurrences. For Level 1 investigations, they can be sketched to develop the narrative sections of forms. For Level 2 investigations, they facilitate a demonstration that the events are presented in logical sequence. For Level 3 investigations, they provide a basis for discovering and defining problems disclosed by the evidence For Level 4 investigations, they also help organize, direct and control the tasks of however many investigators are engaged in the investigation, and screen unsupported theories about what happened, as well as discipline speculations.

The MES Matrix can also be used to define new or existing system operations, and then analyze those systems for potential problems and action needs.

DATA REQUIRED

The MES Matrix organizes data investigators acquire, step-by-step in a progressive way that defines the remaining data still needed as the investigation progresses. This procedure accommodates all relevant data from observers, documents, and other things, as the data are acquired. The procedure specifies criteria investigators can use to identify data relevant or irrelevant to a specific investigation. The procedure determines whether or not data acquired is necessary and sufficient through a specified testing procedure.

DATA DISPLAY

Data acquired during an investigation must be recast into Event Building Blocks (EBs) before it can be used to build an MES matrix. Each new EB is placed on a Matrix as it is acquired, to capture and organize the data. Time and actor coordinates on the Matrix guide the placement of each EB relative to each other EB. Concurrent precede/follow and necessary/sufficient logic tests define the interactions among EBs required to produce the outcome of interest. When completed, the Matrix will describe what happened, and the interactions will define why it happened, with gaps in the EB flow showing remaining uncertainties[3].

MATRIX DEVELOPMENT

Matrixes can be developed manually or on computers. Part I describes manual matrix development. Part II describes computerized development.

Remember, your goal is to develop a flow chart of what happened in a way that explains why it happened, using the data available. Remember also that you are using a dynamic matrix. By design, it will grow and contents may shift significantly over time as more data are acquired during an investigation. It is also a matrix that helps define additional questions to ask and data to seek as it develops.

PART I MANUAL MATRIX DEVELOPMENT.

Manual MES Matrix preparation procedures during investigations require:

1. Event Building Blocks created from the data acquired (See Guide 1).

2. A Matrix layout space, such as a large clear wall, large chalk board or a roll of butcher paper on which to lay out the EBs as they become available. A clear floor are can also be used.

3. A method for capturing the content of interim and completed matrixes, such as a camera, large xerographic or blueprint copier or coded EB filing scheme e.g., an alphanumeric designation of matrix intersection for each card.

4. If butcher paper is used, a marking pen or chalk to draw linking arrows between EBs; otherwise use masking tape for links.

PROCEDURE FOR PREPARING THE MES MATRIX

Position EBs on the MES Matrix as they become available. There is no theoretical limit to the number of EBs you can add -if you can make room for them on your workspace. Manual implementation handles 30 actors easily, computers handle many more.

1. Create the Matrix workspace.

Select a work surface on which you will prepare your matrix. Prepare a name card (3M "POST-IT") for each actor named in any EB you put on the Matrix. Make a new name card containing only the actor's name as you make the first EB for any actor. Start the MES Matrix by placing all name cards in a column down along the left edge of your work surface. Place each EB into the row defined by the name card with the EB actor's name on Matrix.

NOTE: Refer to Guide 1 for detailed information about preparation of EBs. EBs must conform to the specifications in that Guide before they can be positioned correctly on the Matrix.

2. Position EBs on the Matrix

Place each EB you create on the actor's horizontal row. Position it in its correct time and spatial sequence relative to all other EBs already on the Matrix. Expand the Matrix to the left or right when you add EBs, and down when you have to add an actor row.

Use the left edge of the EB to position the EB under the time it started. Keep the time scale flexible, and reposition the EBs as they are added to keep the relative timing intact throughout the investigation.

NOTE: As the Matrix develops, you will find it necessary to insert new EBs between existing EBs, requiring you to expand the time-line scale. Feel free to do so at any time. Using stick-on cards makes it easy to move the events around during the Matrix development process.

3. Test each EB as it is placed on Matrix

As each event is put into place on the actor's line, mentally compare the position of your EB against the next preceding EB and the next following EB on that row, to assure it is positioned in its correct time and spatial sequence.

NOTE: This is the first data ordering step. If it fits sequentially in its logical time and spatial order, it is ready to be placed into the Matrix -- tentatively. A tentative estimated duration can be indicated on the EB or Matrix when needed. Note each tentative entry with a "?" to show it requires confirmation or possible future correction. The duration of EBs may become relevant when two or more EBs occurred or were occurring simultaneously during the occurrence.

Sometimes it becomes necessary to place a time line with tick marks along the top of part of the matrix, to ensure that the timing of events is distinguishable, or to show the role of time in the scenario. Any time line should be considered tentative until the final Matrix testing is completed.

4. Link causally-related EBs.

This is the first of several logic tests for your matrixes. If the actor influenced subsequent actions during the occurrence (a change maker), or reacted to another actor/change maker by doing something, the EB may be relevant. As EBs are added, draw tentative arrows (lines with pointers) from one EB to any later EB that it changed. Linking EBs with arrows creates "EB pairs" or "events sets." Causal links between EB pairs may be shown in one of the following forms.

Figure 2-2. Causally-linked EB Sets

Precede/follow logic links may produce (1) causally-linked EB pairs, or (2) converging EB sets or (3) diverging EBs sets. Uncertainties are indicated by a question mark (?) between the EBs, as in (4). Uncertainties are not objectionable if they were pursued during the investigation, and if they are faithfully represented in the text of the report. (See also IQC procedure in Guide 10.)

NOTE: The linked events sets result from the initial (sequential) logic testing of your EB 'flow chart.' These linked EB sets form the starting point for later completeness testing and the development of needs and management action options. If timing is important, time ticks may be added along the top of the Matrix.

5. Pinpoint gaps on Matrix.

If you can't draw links between EBs, the gaps between unlinked EBs point to potential unknowns in your understanding of what happened. Thus the Matrix defines what you need to find out as your investigation progresses. Where links are suspected to exist, an arrow with a ? shows uncertainty that should be a candidate for further investigation. The question marks can be used to assign investigation tasks as the investigation progresses, reducing the elapsed time for the investigation, and reducing investigation costs.

NOTE: The linked events sets are the results of the initial (sequential) logic testing of your EB 'flow chart.' These linked EB sets form the basis for later completeness testing and the development of needs statements and candidate actions that will follow. If you can not establish links from the beginning EB(s) to the last EB or outcome, you know you have gaps in your understanding of the occurrence that may need further investigation.

6. Bridge gaps on Matrix.

Try MES-Trees, (see Guide 3), simulations, tests or reenactments (see Guide 6) to bridge gaps in your logical EB flows if the gap prevents you from understanding and describing what happened and why it happened for your purposes.

NOTE: MES-Trees use EBs on both sides of the gap to bound speculation. This helps control irrelevant speculation, and helps identify specific data needed to verify each hypothesis. Change Tracking (Guide 4) and the Energy Trace and Barrier Analysis (Guide 5) can also be helpful for this purpose. The Test Planning Guide (Guide 6) describes procedures to ensure that any expenditures for simulation, testing or reenactments address specific data needs for your Matrix, reducing the risk of unfruitful efforts.

7. Perform Necessary and Sufficient tests.

This procedure tests each EB set or pair to determine the sufficiency of the event flows represented to produce the occurrence outcome.

NOTE: This testing procedure reviews all the EB sets on the Matrix
to determine the necessity and sufficiency of the EBs and links shown.

A Test each EB pair for the necessary relationship, then B test if linked EBs are sufficient to produce the right EB (event Z), then C review all the links on the matrix. Begin as follows, starting with the right-most EB set (the outcome) on the matrix. Follow the if-then questions on the decision tree procedure below for each EB pair on the matrix.(EB or Event Building Block is expanded term for Event Block or EB)

Recent research has disclosed that use of a necessary/sufficient GATE (N/S GATE) into each event block to indicate incomplete testing or understanding provides increased rigor to this procedure. This gate is now included in Investigation Catalyst software. The form of the gate is filled arrowhead on the links into the event block. When arrows are created they arrowhead is empty. After all the necessary and sufficient predecessor events have been coupled to the event block, the empty arrowheads into the EB are filled in. Empty arrowheads indicate either a need for more investigation, or an uncertainty that can't be resolved because of a lack of data.

8. Clean Up Matrix.

Upon completion of the Necessary and Sufficient testing procedures, you should remove all the extraneous, unlinked EBs and notations that are irrelevant to the description of what actually happened from the Matrix. What remains should be a network of linked EBs that lead from the first event in the scenario to the outcome. That network should be the fullest description possible from the observed data and structured hypotheses. The next step is to subject the finished matrix to a final quality control examination.

A completed matrix will have these general elements, before the recommendation development process is undertaken

MES MATRIX QUALITY CONTROL REVIEW

  1. After a matrix is checked for the necessary and sufficient logic flow, initiate a final quality control check of the work. Recheck EBs against the criteria in Guide 1. As you gain experience, you will begin to spot your own signals that an 'event' in a Matrix is really not an event at all, but rather is a conclusion, state or other unacceptable entry.
  2. After a Matrix is completed, have someone who does not know anything about what happened to review the Matrix for logic errors and out-of-sequence events. Also determine the completeness of the image or mental movie they can develop, as a further Q C step.
  3. Check to be certain that all the sources referenced in the EBs are verified against the EBs, and are on hand and retrievable.
  4. If the Matrix review indicates problems or misunderstanding, ask the persons who should know the operations to look at the Matrix and verify your entries and the logic of the EB flow and links. This review usually produces valid events flows and descriptions, or additional EBs after the are subjected to the necessary and sufficient tests.
  5. Review each ? on the Matrix and verify that all sources of data have been exhausted, or that the decision not to pursue remaining uncertainties, gaps or unknowns has been accepted by the investigation sponsor (the person paying for it.)
  6. After the quality control checks have been completed, the investigator(s) should sign the Matrix to denote that it has been checked and found satisfactory, and that it is finished.

Upon completion of the final quality control steps and the signing of the Matrix by its developers, the Matrix is ready for subsequent uses.

PART 2. COMPUTERIZED MATRIX DEVELOPMENT

The process of developing EBs, positioning them on the matrix, linking them and testing sets for cause-effect relationships is is supported by Investigation Catalyst software, from Starline Software Ltd. That software provides Tutorials, software help and rules the help investigators. However, users should be acquainted with the manual implementation described above to guide their use of the software.

Computer graphics applications have developed to the point that they can be used to create animated descriptions of the behavior of objects in accidents. In the aviation field, particularly, the behavior of an aircraft in flight has been reconstructed from fight data recorders and displayed as a computer-generated movie. In significant cases, where the behavior of objects is important to display, the "mental movie" created by the investigator can be "transferred" to a computer using these computer graphics animating applications. The displays can also be used to aid in hypotheses. Look for expended use of this capability in the future.

Once made, these animations of the occurrence can be used for training, design considerations, and other purposes.

Users are invited to provide feedback to the webmaster on software applications that would be useful to other users.

ENDNOTES

[1] See also Hendrick, K. and Benner, L., INVESTIGATING ACCIDENTS WITH STEP, Marcel Dekker Inc., New York, NY 1987. The acronym STEP (from Simultaneous Timed Events Plots) has been used to describe these matrixes. However, while descriptive of the format, STEP does not communicate the full breadth of functions served by the matrixes, so the broader term MES technology-based Matrixes or MES MATRIXS is the term of choice used in this set of Guides.

[2] The format shown in this Guide differs from the format presented previously, because these Guides distinguish the description development tasks separately from the needs and recommendation development tasks. A completed marked up matrix is illustrated in the Recommended Actions Development Guide 8.

[3] MES Matrixes differ from Events and Causal Factors (E&CF) charts because MES shows only actions - no conditions - on the matrix. Conditions do not produce the next change; the action(s) that produced the conditions produced it, and they should be the primary interest of the investigator. E&CF charts are technically obsolete today.

Go to Guide: 0 1 2 3 4 5 6 7 8 9 10