Ladder of Inference

t is vital in any group and/or organization to establish and maintain a reliable exchange of valid and verifiable information about important problems and issues. This requires the ability to discriminate among four types of information: description, inference, attribution, and evaluation. A description is a (hopefully objective and reasonably accurate) report or account of an experience or observation. An inference is a conclusion derived from beliefs or what are thought to be facts. An attribution is an ascribed, inferred, or assumed cause, characteristic, or motive of another person. An evaluation is a determination or judgement about the value or "goodness" of a statement or action by another person. The Ladder of Inference Model is a very useful tool for helping individuals become more aware of and discriminate among these four very different types of information and their use in communication.

The Ladder of Inference Model from Action Science is a representation of different ways that individuals make sense of and deal with everyday events. Individuals select and process certain aspects of events, and introduce elements from this processing into their thinking, feeling, and interactions. These elements include inferences, attributions, and evaluations that may have considerable error relative to objective observations of the same events. The further an individual moves or extrapolates from the actual, original data (i.e., the verbatim words spoken and observable actions made by individuals), the greater is the potential error. This model can be useful in helping individuals reduce such errors and the resulting interpersonal problems.

We can consider various numbers of steps on the ladder of inference, starting with the data (the actual statements and actions) and moving progressively further away from the data, e.g., as illustrated in the following steps (read from bottom to top):

5. EVALUATION (of the other person's statements and/or actions)

4. ATTRIBUTION (assumed cause or motive of the other)

3. INFERENCE (interpretation & conclusion about what we think happened)

2. SELECTIVE & PARTIAL FOCUS (on some part of what was said or done)

1. DESCRIPTION (accurate recounting of the observed actions and/or verbatim statements of the other)

0. DATA (verbatim words said or the specific actions taken)

In the following illustration of this model, we consider a situation in which two individuals, X and Y, were participants - and then we consider a range of possible interpretations and responses by X to the actions of and verbatim words spoken by Y (i.e., "the data"). Briefly, the data are:

X and Y are both VPs, reporting to the president of a company. In an executive staff meeting X has just made a proposal to develop a new line of business. Y leans forward and speaks, rather loudly: "Certainly the company needs some new business options. This is a creative, interesting idea, but I have a lot of questions. What is the basis for your conclusion that this project would break-even in less than one year?"
Now, let's consider a range of possible ways that X might make sense of this brief interaction. Four different possibilities are summarized below, in order of increasing distance or extrapolation from the original data. These possibilities are referred to as different steps up the "ladder of inference," a model in which increasing extrapolation beyond the original data is represented by taking additional steps up the ladder.

1. X could possibly describe (report objectively and accurately) what Y said and did (step 1 above); however, it is likely that X would operate at one or more steps removed from the verbatim data to select and derive meaning (make sense) of what happened. It is likely that X will at least move to step 2 on the ladder of inference, in which X selects a portion of Y's observable actions and verbatim words for attention, e.g., X might select and focus on Y's statement:

"... but I have a lot of questions..." and that Y was speaking loudly.
2. X may move further beyond the data to step 3 on the ladder of inference. At this step X might infer or attribute meaning, which may be different from the verbatim statements and likely goes beyond the common cultural meaning of the utterance. Inferences at this level are quite specific to the individual. One possible example of X's thinking could be:

Y is trying to make me look bad and shoot down my proposal.
3. X may move even further beyond the data to steps 4 and 5 of the ladder of inference by developing conclusions, including attributions about Y's motives and evaluations of Y's actions and utterances. One possible example is:

Y is a {expletives deleted} lazy bureaucrat who wouldn't know a good idea if it hit him in the face! He's not willing to hustle and make things happen, but doesn't want anyone else to make him look bad by their accomplishments. The president should have fired him years ago!
It should be clear that, each time X moves further up the ladder of inference, she/her moves further from the actual data about what occurred in the event and, therefore, he/she is more prone to error. Also, as one moves further up the ladder, it is increasingly likely that the inferences, attributions, and evaluations of different participants will differ. For example, an alternative inference at the step 3 (which might be the inference made by a different individual observing the same event) is:

Y is asking some important questions that X didn't address adequately in his presentation. Y is really looking after the company's interests and future.
The Ladder of Inference Model can be used to help individuals recognize the kinds of inferences they are making, the assumptions implicit in these inferences, the conclusions they lead to, and the effects that acting on these inferences have in the individuals' organizational settings. Then, it can help individuals consider that there are other alternative inferences, learn to inquire and check out potential inferences, and ultimately act in more effective ways. For example, individuals can be helped to slow down and focus on the inferential steps and implicit assumptions they are using in abstracting conclusions from the original data of an event. Usually, these inferential processes are done quickly, skillfully, without awareness - so an individual may need assistance in reconstructing his/her implicit steps and reexamining the inferences and attributions made along the way. This kind of off-line analysis can help individuals learn about their typical response patterns and become more skillful in recognizing and avoiding such ineffective patterns as they deal with future events.