Tree Maps

1. Introduction

This tool makes use of qualitative information. This is in the form of important distinctions or differences that people see in the world around them. Important differences are those that make a difference. Differences that make a difference can be described as information (Bateson, 1979).

The term "tree map" is taken from the field of cladistics, a methodology for the classification of living organisms into a nested hierarchy of categories based on relative commonality of structure (e.g species, families, genera etc).

Tree Maps help us deal with some problems commonly encountered when dealing with qualitative information:

1. "The Apples and Pears" problem: How to make comparisons between complex entities. Clue: Locate the choice in a context, and bring the observer into the picture.

2. "The Trees and Forest" problem: How to see an overall order but also to recognise local diversity at the same time.Clue: Link the analysis of the macro to the micro.

2. How to develop a Tree Map

The Basic Structure

A Tree Map is a nested classification. It can be represented as a diagram that grows from a single trunk and ends with many leaves. A simple example is given below.

Generating a Tree Map

Normally this method is used with one respondent. Some ways of using it with more than one person are discussed below. Follow the following steps:

1. Identify the respondent's area of expertise or knowledge that you want to explore. For example, knowledge of animal diseases held by a paravet, or knowledge of local NGOs held by a NGO staff member working on a NGO capacity building project.

2. Generate a list of actual cases which contains a wide variety of examples. For example, a list of cases treated by the paravet in the last month, or a list of NGOs know to the NGO staff member. Aim for at least twenty case and ensure there is variety. Write the name of each one down on a separate card. Cases might be events (treatment provided) or entities (clients). Having the names of actual cases is not essential for the development of a Tree Map but it does help a great deal by grounding the exercise in an observable reality.

3. Place all the cards in one pile (A) and ask the respondent to tell you about some of the differences between all these cases. The purpose of this question is simply to generate awareness of the large number of differences that exist.

4. Ask the respondent to sort all the cards into two piles of any size (B&C), according to the most important difference between the cases represented on the cards. Emphasise that importance can be as seen from his/her position. There is no requirement here to generate an "objective" description. Also emphasise that a distinction is important if it makes a difference. Because respondents may casually offer a difference simply to oblige the interviewer it is important to check its significance by asking "What difference does this difference make ?" If one can't be identified suggest to the respondent they consider if there are other differences which might be more important.

5. Write down a description of the reported difference at points B&C. Take each pile at B&C in turn and repeat stage 4 above. Repeat this process with all piles until there is only one example left in each pile. These become the leaves (end of branches) of the Tree Map.

6. In some cases there may be more than one example left in a pile but the respondent may not be able to identify an important difference between them. Don't force them to do so, but simply note that no further difference could be identified. It may be useful to do all the above steps in the order of the groups of cards that the respondent feels they know most about first, and least about later.

Working with multiple respondents

If you want to involve more than one respondent in the development of a Tree Map there are two options. The first is "quick and dirty". Simply ask the above questions to the group, and let them discuss the matter before making a collective choice. The result is that we have diagram representing a collective construct but we dont know who has contributed what, and to what extent each person agrees with the final product. The second option is more time consuming, and still not completely articulated (Readers contributions are welcome !). The proposed steps are as follows:

1. Present respondent A's Tree Map to a second respondent (B), with the labels attached to each branch, but without the names of the cases found at the end, at each leaf. Give the cards containing all the cases to respondent B. Ask them to proceed from the trunk upwards, sorting the cards according to the differences already labelled on the tree by respondent B, according to where they feels each card should go. Ideally there should be a strong correlation between the two sets of classifications made on the same tree structure. This can be tested by visual impression. A method of making a statistical test of the correlation is being investigated. Contributions on how to do this analysis would be welcomed !

2. Place the distinctions listed at each junction on the tree by respondent A on a second set of cards and ask respondent B to develop a new tree, using those distinctions only (plus the same set of cases). The two structures can then be compared visually. I am currently looking for some mathematical way of comparing the similarity of the two structures.

3. A third and more open possibility is to give respondent B the same sets of distinctions and cases as used by respondent A, but allow them to introduce any additional distinctions they feel are necessary. The two can be compared in terms of of the percentage of common distrinctions, out of the total number of distinctions used.

Future developments in group based methods could explore two different approaches. Collective Tree Maps could be developed serially (e.g. one respondent after, in a chain) or in parallel (e.g all respondents moving one step at a time).

3. Using Tree Maps

General uses

Value can be obtained from Tree Maps at two stages:(a) during the creation of the Tree Maps, and (b) through comparisons made between parts of the structure once it has been created.

1. During creation of a Tree Map the main use is as a enthographic tool: understanding people's view of the world. Ther are two form of usage:
   1. Identifying the distinctions that people see as important. This is evident in the differences cited, and the differences they are seen to make. It is also evident in how early in the exercise they are cited. The method makes use of binary classificiations (two sub-categoires only at any one time). This forces the respondnet to prioritise, to choose between a number of potentially important distinctions. There is a common sense caveat here: What you hear will be context dependent (especially on the perceived nature of the relationship between interviewer and respondent). But differences that respondents cite, but may not believe are important, may be harder for them to explain (what difference they make). This helps make them detectable.

   2. Identifying the limits to people's knowledge: When respondents cannot identify differences between two or more entities the limits to their knowledge seem to have been reached. Not knowing what people know about can be important, especially when they might be expected to, or claim to have, expertise in that area.

2. After creation there are two other types of usages. One is based on proceeding from trunk to leaf, the other is based on proceeding from leaf to trunk.
   1. Planning based discussions: Starting from the trunk, respondents can be asked questions such as "How will your work in the next six months with this group be different, compared to this group (those labelled at the first junction) ?" Then move to the next junction on each branch, one a time, and re-iterate the same question. Lack of a difference suggests lack of a strategy, and points to areas where it may need to be be more articulated. Other related planning questions which can be asked, from trunk to branch, are "Which of these will be the most immediate priority ?", "Which of these will you be spending the most time with ?", "Which of these will present the most problems ?", Which of these do you want to scale up / expand ?", etc. The answer, and associated rationale, can be added as a further annotation to the Tree Map, at the appropriate junction.

   2. Evaluation based discussions: These proceed in the opposite direction, from leaf to trunk. Starting from two adjacent leaves respondents can be asked questions such as " Which of these two groups was most successful in the last six months ?". Then move to the next two most adjacent leaves and re-iterate the same question. Then compare the two leaves which were identified as the most successful and which share the same larger branch. Reiterate the process again with the other leaves. In each case "move" the most successful case down the tree to the next junction, to meet the most successful case moved down from the next most adjacent branch. As above, the choice, and associated rationale, can be added as an extra annotation to the Tree Map, at the appropriate junction.

      Success overall can be defined in two ways. At any one junction it will be in terms of which of the two types that exist there (as represented by particular cases). The types take on a wider generality and relevance the closer they are to the trunk. When viewed in terms of particular cases (examples) success could be seen in terms similar to a tennis tournament. The number of grades of success or performance would equal the number of junctions along a branch from trunk to leaf. In the diagram above there are three, beyond the original leaf itself. Comparisons would need to take into account the fact that some branches may be longer than others.

Specific Applications

1. Assessing the impact of capacity building activities: When supporting capacity building work with individuals or whole organisations, we might expect that this assistance, either in the short or long term, would make a difference to the person or organisations relationships with their clients. One attribute of that relationship is responsiveness . The service provider might be more sensitive to the differences between clients needs. They may be more up to date in their knowledge about their various clients' needs. The important differences they see between their clients (that they think are important) may be more reflective of their clients concerns, and not just their own. Much of this information is available, in the first instance, in the form of knowledge the service provider has (or does not have) about its clients. This knowledge is in effect a proxy indicator. Tree Maps are a way of mapping a service provider's knowledge of their clients. How differentiated is their map of the differences between clients, how up to date is their knowledge of each "leaf" and branch ? How much do the descriptions of difference reflect client versus provider concerns ? This information can be verified by independent observation and follow-up contacts with clients. Are services visibly differentiated (rather than homogenous) ? How frequently have they been modified ? Do clients views on relative value of services provided correlate wiith those of the provider ?

2. It should be possible to use Tree Maps as a means of doing a stakeholder analysis in a development project. This could initially be from the perspective from one observer, most likely a stakeholder themselves. Firstly, a list of cases reflecting the maximum possible variety of stakeholders would be identfied. The process would then start at the trunk, with the respondent identifying "the most important distinction between all the stakeholders in the project". Then each initial category of stakeholder could be progressively differentiated until all cases were located as a leaf of their own. Alternately the analysis could be attempted without prior naming of specific cases, and then differentiated to the degree of detail that was required.

   An alternate process would make use of multiple participants, involved on a rolling basis. The first respondent, (possibly the one person seen as the most important) would be asked to identify the first branch in the tree. People identified as most representative of each of those two branches (possibly by the first respondent) would then be asked to identify the next two branches (x2), from their point of view. People representative each of those branches (x2) would then be asked to identify the next two branches (x4), etc. Some branches may join back toether again, when the same person is identified as a representaive of an important sub-group by two different branches of the tree that has been developed. This should not be a problem. It makes the tree more heterarchical than hierarchical, and provides more flexibility of interpretations when, and if, experiences is aggregated back down the tree structure to the trunk

3. Scenario planning: This is a practice widely associated with Shell. It involves developing a number of alternate views of the future and then identifying how the organisation concerned would react differently to each different scenario. Tree Maps of different scenarios could be developed by starting off with a question: "What is the most important difference in views within this organsiation about what it should be doing over the next 10 years ?". Within each new branch created by the answer, the question could be repeated, but pre-phrased in terms of "Within this vew of the future...". Once the Tree Map is constructed a planning based set of questions could be pursued, as described above. As above, both individual and group based versions of this exercise could be designed.

4. Qualitative monitoring: A staff member of an NGO may be working with many groups or villages. Their categorisation of these groups or villages can be documented using the Tree Map method. This structure then provides a framework within which comparisons can be made by that person between the different items of news they may have heard from each of these locations. News is defined here as a form of difference, a change from the previous condition. News from two sources on immediately adjacent branches can be compared in terms of its relative importance, as seen by the observer (the NGO staff member who owns the classification). The most important news (thus defined) can then be compared to the most important news from the next most adjacent branches. Complex information can be managed by making one-at-a-time comparisons. Comparisons will start at a very local level (adjacent leaves) and move to towards more global comparisons (branches next to the trunk). In organisations the organogram itself can form a similar type of tree structure and be used for the same purpose. A participatory monitoring system involing a large number of staff, designed on the this basis is described in a paper available at http://www.swan.ac.uk/cds/rd/ccdb.htm

An Invitation

If you are interested in using this tool in any setting I would be happy to provide informal advice and comment. For further information or permission to reproduce this paper email Rick Davies at rick@shimbir.demon.co.uk or fax 44 1223 841367

Return toRick Davies' Home Page or to MandE NEWS