-
Laddering: An introduc0on
Gordon Rugg
This document is copyle: by Hyde & Rugg Associates. You are welcome to use it in whole or in part for any non-‐commercial
purpose, including academic lectures, provided that you reproduce the copyle: statement with the material that you use.
Hyde & Rugg Associates
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Laddering is a technique that was invented by Hinkle in the 1960s, as a way of clarifying a person’s goals, beliefs etc in a simple, systema0c and powerful way. At least three main branches of laddering have developed since then. One is in clinical psychology and counseling, where laddering began. Another is in market research, pioneered by Reynolds and Gutman. A third is in compu0ng, where it is used in requirements acquisi0on and so:ware evalua0on. Laddering fits well with graph theory, which is based on simple concepts, but is extremely powerful, and is used in a wide range of fields. It’s par0cularly useful in combina0on with other methods, such as think-‐aloud or card sorts. The version of laddering described in this tutorial is based on the type of laddering that is used in compu0ng. It has an explicit grounding in graph theory. It’s par0cularly good for: • Systema0cally unpacking what people mean by subjec0ve or technical terms • Iden0fying and unpacking different viewpoints/facets within a body of knowledge • Finding out how people categorise things • Systema0cally mapping people’s goals and values
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Contents This tutorial introduces basic laddering. It covers the following topics: • An overview of laddering’s origins, and its underpinnings in graph theory • A brief overview of graph theory • How to use laddering to unpack a person’s categorisa0on • How to use laddering to “boWoming out” explana0ons into objec0ve terms • How to use laddering to unpack a person’s goals and values • How to use laddering to help a person iden0fy alterna0ve routes to the same goal • How to analyse the results from laddering • A brief overview of facet theory, for handling mul0ple viewpoints in knowledge • Hints & 0ps • Summary and references
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Graph theory: A three layer tree Graph theory: the basic concepts A lot of human knowledge is organised into hierarchies like the one opposite. In graph theory terms, this is a tree. A key feature of a tree is that none of the branches re-‐join lower down. The tree consists of nodes (the circles) which are joined by arcs (the lines). The lower diagram opposite shows what’s known in graph theory as a net, where some branches re-‐join. The version of laddering described in this tutorial is based on trees, as opposed to nets.
The basic concepts
Copyle: Hyde & Rugg www.hydeandrugg.com
Graph theory: A net
-
Trees, leaves and bo;oming out Graph theory integrated with laddering Trees end in leaves where the tree can’t be subdivided any further. In the diagram opposite, the black nodes in the boWom layer are leaves. So, if you’re unpacking someone’s explana0ons of subjec0ve or technical terms, the unpacking doesn’t go on in an endless loop; it reaches leaves, where you have specific tangible terms. This is known as bo;oming out in laddering. The last main graph theory concept for the moment is directed graphs. For instance, “X is the child of Y” involves a direc0on of rela0onship, whereas “X is related to Y” is the same in both direc0ons. Laddering usually involves directed graphs.
The basic concepts
Copyle: Hyde & Rugg www.hydeandrugg.com
Graph theory: A directed graph
-
Main applica>ons of laddering Laddering overview We have found laddering par0cularly useful for clarifying three main areas of people’s knowledge: • The mental structuring they use to organise
things into types or categories • Subjec0ve or technical terms that they use • Their goals and values For each of these three purposes, we’ve developed a customised nota0on that lets you record your ques0ons and the other person’s answers swi:ly, accurately and consistently. It’s simply not prac0cal to try drawing their knowledge as a tree diagram – the diagram soon becomes far too large to fit onto even a large sheet of paper. The diagrams opposite show the nota0on for each purpose. There’s more detail in the following pages.
The basic concepts
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Laddering down through subcategories The diagram opposite shows how you can ladder downwards through the categories and subcategories in turn, via spoken ques0ons (on the le:) and wriWen record (on the right). In this example, the first probe (i.e. ques0on) produces three categories, namely cars, trucks and motorcycles. [Hint: don’t suggest other categories – you’re trying to find out what categories they’re using, not what categories you think they should be using.] The next probe unpacks cars, and elicits three subcategories: saloon, hatchback and sports. (You keep going un0l the respondent runs out of categories; you don’t need to stop at three.) There’s then a decision to make: whether the next ques0on should be about subcategories of saloon, or go back a step to subcategories of trucks, and then of motorcyles. When laddering, you have to make this type of decision a lot. We usually work through all the main categories first, then each subcategory in turn, going down one level of the graph at a 0me. This is breadth-‐first explora0on, as opposed to depth first; in depth-‐first you drill down as far as you can ini0ally, and then work outwards. It’s your choice.
Basic laddering on categories (1)
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Graph from laddering in progress
Graph showing three levels in colour
The previous page showed how you can ladder down from vehicle to elicit someone’s categorisa0on of vehicles (into cars, trucks and motorcyles), and then ladder further down through those categories. \ It produced three subcategories of cars and two subcategories of trucks, and was about to elicit subcategories of motorcyles. The graph for this would look like the one opposite. You could con0nue the laddering un0l you had boWomed out all the branches of the tree, or you could boWom our only some branches – it’s your choice. The lower image shows how you can start doing quan0ta0ve analysis of your results. There are three levels in the graph in the image. The top level (dark blue) contains one node; the middle contains three nodes (mid blue) and the lowest level contains nine nodes (light blue). Experts in a field typically have deep, rich trees. Novices typically have shallow, lopsided trees. There is more about analysis of laddering results in a later sec0on.
Basic laddering on categories (2)
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Laddering on explana>ons Laddering on explana0ons is very similar in process to laddering on categories. The example opposite is based on a real case. I was trying to find out why some students were gefng lower marks than expected in their wriWen work. When I laddered down, I found that the students were focusing on the flow of the wri0ng, at the expense of the references and technical concepts that would have earned them the marks. Laddering on explana0ons o:en produces unexpected findings like this, that make sense of previously-‐puzzling behaviours. Laddering on explana0ons also tends to produce surprises from ques0ons where you think you know what the answer will be.
Basic laddering on explana0ons
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Bo;oming out an explana>on The example on the previous page showed how laddering elicited an explana0on that made sense of what the students were doing. It’s usually a good idea to keep going un0l the explana0on boWoms out. Usually this occurs when you either reach a number, or you reach a concept that can’t be explained any further with words, such as the name of a colour or shape. In the example opposite, the explana0on boWoms out with a number – two or three references per page. [For brevity, we’re assuming that the concept of references is boWomed out separately.] The most interes0ng answers o:en appear at the boWoming out level. BoWoming out can unpack subjec0ve terms such as too many into objec0ve terms, such as numbers that come from the par0cipants themselves. This is invaluable in fields such as so:ware and product design, where user feedback o:en has a lot of subjec0ve terms. It’s also invaluable for quan0ta0ve analysis.
Laddering on explana0ons: BoWoming out
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Upward laddering: The star>ng point
Upward laddering: Going up another level
Laddering is very useful for elici0ng people’s goals and values. It’s par0cularly good at uncovering unexpected higher-‐level goals which give new insights and solu0ons. The illustra0on shows a real example. It starts with a choice between two arbitrary items. The ini0al choices soon converge on a small number of higher-‐level nodes, regardless of where you start. This example shows how using a teapot as the ini0al item swi:ly leads into concepts that can be used for IT support. The double-‐headed arrow is the symbol for “Which of these would you prefer and why?” The china teapot is preferred, because it’s not expensive. The next upward probe is where things start gefng interes0ng. You might expect “Not expensive” to be chosen because of “not was0ng money” but instead the higher-‐level goal is “not showy”. The person then went on to explain that they value family tea, as an informal bonding experience, where not being showy is a key feature.
Basic upward laddering on goals and values
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Laddering up on goals and values
Laddering down on alterna>ve routes
Let’s imagine that you’re using laddering to iden0fy ways of helping housebound people via computer-‐based support. The upward laddering in this example has iden0fied family bonding events as a goal or value for the respondent. Now that you know this is important to the respondent, you can start looking for other ways of achieving that goal, where computers might help. The lower illustra0on shows how you can do this via downward laddering on categories (“types of”) from that goal. We already know that aJernoon tea is one type of family bonding event, from the upward laddering, so we put that into the list of responses without needing to repeat the ques0on. The probe elicits two more answers: walking the dog and looking at family photos. The looking at family photos response immediately raises the possibility of using so:ware to view the photos, perhaps via Facebook or similar social network sites.
Laddering to find alterna0ve routes to a goal
Copyle: Hyde & Rugg www.hydeandrugg.com
-
An even spread of knowledge
Sparse, shallow knowledge
Very focused deep knowledge, sparse elsewhere
Qualita9ve analysis Laddering can give answers about people’s reasoning, as with the students who were deliberately not using many references in their wri0ng. Some0mes that’s all the answer you need. Laddering can also give answers about how people categorise their world, and about what maWers to them, as in the teapot example earlier. Combining qualita9ve and quan9ta9ve analysis You can combine qualita0ve and quan0ta0ve analysis by e.g. looking at how many respondents define something in the same way. By boWoming out their explana0ons, you can also check whether they all mean the same thing by the same term, or not.
Quan9ta9ve analysis You can measure the depth, breadth and richness of the trees you elicit, by coun0ng levels and nodes. This can be useful for assessing how much knowledge someone has of a topic, and where there are gaps in their knowledge. The illustra0ons opposite show some paWerns that o:en occur.
Laddering: Basic analysis of results
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Using laddering as an incidental method We o:en use laddering for quick clarifica0on of points that arise when we’re using another method. For instance, if someone is giving feedback about some so:ware, and they men0on that it’s clunky, then we’d probably use laddering to clarify exactly what they mean by clunky. This frequently produces unexpected new insights, and usually produces useful, clear results. This is par0cularly helpful when you boWom out an explana0on, and discover that the respondent has a clear, sensible leaf-‐level explana0on that’s completely different from what you were expec0ng. We use upwards laddering similarly, to clarify the respondent’s goals rela0ng to a point that arises in conversa0on. O:en, the higher-‐level goal gives very useful new insights.
Using laddering as a main method Laddering is good for inves0ga0ng subjec0ve concepts rigorously and systema0cally, based directly on the respondent’s own categorisa0on and defini0ons. It’s useful for finding out what people actually want, par0cularly when you ladder up more than one level. This takes you beyond the responses that you’d get from ques0onnaires, into the higher-‐level goals and values. It’s then o:en possible to see clusterings of values across individuals (which is why it’s used in market research). For instance, one of my students found that respondents who liked a par0cular product fell into two groups: one group liked it for func0onal reasons, while the other liked it as a fashion statement. The implica0ons for marke0ng etc were very different for each group. Laddering is also very useful for clarifying what precisely people mean by subjec0ve terms, which can help with the design and choice of products and services for them.
Ways of using laddering
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Facet 1: Place of manufacture
Facet 2: Cost
The answers to a ques0on o:en depend on the viewpoint involved. This has been studied systema0cally in various fields, such as informa0on science, where it takes the form of facet theory. The two diagrams opposite show how laddering downwards from the same star0ng point can produce very different classifica0ons, depending on the viewpoint/facet involved. If we use the facet of place of manufacture, then item A and item B are both in the class of USA. However, if we use the facet of cost, then A and B go into different classes. When you’re laddering, you need to be sure that you know which facet you’re laddering on, so that you don’t end up with a tangled mess. This is another reason why it’s beWer to take notes as a collec0on of small separate chunks, rather than try to do one huge diagram. With separate chunks, it’s a lot easier to spot when different facets are involved, and to make sense of what’s happening. Respondents usually tell you the relevant facets by asking you which viewpoint you want them to use when answering your ques0on. You can ask them to list the relevant viewpoints, and then to tell you which viewpoint they’re using.
Laddering and facet theory
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Two sample chunks in a transcript Hints and 9ps • Laddering works well in combina0on with other
techniques, to clarify points that they raise. • When using laddering, you have to make
frequent decisions about where to probe deeper and where to stop. This is mentally wearing, so budget rest 0me between sessions.
• Vary the peripheral phrasing around the probe, so you don’t look robo0c, but be very careful to keep the core probe phrasing the same, so you don’t accidentally dri: into a different facet.
• Don’t probe into the other person’s personal space without ethical clearance.
• Leave plenty of white space in your paper transcripts; otherwise, the transcripts become horribly cluWered.
• Transcribe the exact words that the respondent uses; don’t paraphrase.
• If the respondent talks too much, slow them down by insis0ng on transcribing every word they say.
• Keep it simple. Don’t get too ambi0ous; start by just going up and down a couple of levels.
Laddering: Hints and 0ps
Copyle: Hyde & Rugg www.hydeandrugg.com
-
Summary and conclusion Laddering is a fast, flexible method that elicits the respondent’s knowledge and defini0ons and then represents them in a simple but rigorous nota0on. It’s par0cularly useful for clarifying people’s higher-‐level goals and values, and for clarifying what people mean by subjec0ve or technical terms. This o:en helps to iden0fy and rec0fy misunderstandings. Laddering is a good complement to other methods such as card sorts or think-‐aloud technique, which are good at elici0ng informa0on, but not so good at unpacking the broader context for that informa0on. Using basic laddering is easy, but advanced laddering needs prac0ce, and a clear understanding of the underlying theore0cal founda0on. This is one reason that we prefer to do laddering manually, rather than using so:ware, which tends to constrain you and to be less flexible than laddering with pen and paper. If you o:en find yourself wondering why someone is doing something, or what they mean by something, then laddering will probably help you find the answer.
Some references and links Fransella, F. & Bannister, D. (1980), Inquiring man, Harmondsworth: Penguin. A very readable introduc9on to Kelly’s Personal Construct Theory, on which laddering is based. Hinkle, D.(1965), The change of personal constructs from the viewpoint of a theory of construct implica>ons. Unpublished PhD thesis, Ohio State University. In Fransella, F. and Bannister, D. (1980), Inquiring man, Harmondsworth: Penguin. The text that originated laddering. Reynolds, T.J. & Gutman, J. (1988) Laddering theory, method, analysis and interpreta0on. Journal of Adver>sing Research, February/March 1988, pp 11-‐31. A good, thorough overview of the variety of laddering used in market research and related areas. Rugg, G., Eva, M., Mahmood, A., Rehman, N., Andrews, S. & Davies, S. (2002) Elici0ng informa0on about organisa0onal culture via laddering. Informa>on Systems Journal , 12, pp. 215-‐229. Laddering in a field usually viewed as very subjec9ve. Rugg, G & Petre, M. (2006). A Gentle Guide to Research Methods. This contains a tutorial chapter on laddering, with worked examples, including analysis.
Summary, conclusion, references & links
Copyle: Hyde & Rugg www.hydeandrugg.com