Explanation Satisfaction Scale

Description

The Explanation Satisfaction Scale is a short questionnaire on which people who have used an AI system's explanations rate how satisfying and sufficient they found them; Hoffman, Mueller, Klein and Litman developed it in the DARPA Explainable AI programme to capture the user's judgement of an explanation, as distinct from a researcher's judgement of its quality. Each item is a statement about the explanation, for example that from it the user understands how the system works, or that it is satisfying, has sufficient detail, seems complete, tells the user how to use the system, is useful to their goals or shows how accurate the system is, and each is rated on a five-point scale from 'I agree strongly' to 'I disagree strongly'. Participants complete it after they have worked with the system and received explanations, usually once for each explanation condition in a study that compares conditions; the published form sets no scoring rule or threshold, and the implementation by one of its authors in the OpenScales library scores it as the mean of the item ratings. Versions differ: the authors' 2018 report gives eight items, their 2023 journal article seven (it drops the item on knowing when to trust the system), and the form on the XAITK site nine (it adds an item on reliability), with a note that researchers may use a subset. The score measures how users judge an explanation, not whether they understood it or decide better with it, and the authors present it beside separate measures of mental models, trust and task performance.

Example Use Cases

Explainability

A research team comparing counterfactual images with saliency heatmaps (LIME and layer-wise relevance propagation) for a chest X-ray pneumonia classifier gives lay participants the scale after they have predicted the classifier's decision on a set of images. It chooses which explanation to develop on prediction accuracy and uses the ratings only as a second check, so that the explanation people prefer is not assumed to be the one they understand better.

A team deciding whether causal explanations of a reinforcement-learning agent's strategy in a real-time strategy game are worth the extra modelling gives participants the scale after each explained gameplay video. It compares the causal explanations with a simpler explanation and with no explanation on four items (understanding, satisfaction, sufficient detail and completeness), and pairs the ratings with a test of whether participants can predict the agent's next action.

A lender piloting a new letter that explains which factors led to a credit refusal asks pilot applicants to complete the scale after reading either the new letter or the current one. It uses the items on sufficient detail and usefulness to decide whether to add the sentence, held back so far, that says what change would have led to approval.

Limitations

A high rating does not show that people understand the system or decide better with it. Hase and Bansal found no evidence that participants' ratings of explanation quality predicted whether the explanations helped them predict the model, Buçinca et al. found that subjective measures did not predict performance on the real decision task, and the scale's authors note that people can feel satisfied while their understanding is piecemeal or wrong.
The published evidence for the scale is small: content-validity ratings by 35 attendees of a DARPA programme meeting, an internal consistency (Cronbach's alpha) of 0.86, and eight volunteers rating good and bad explanations of everyday technologies, where the difference was large but only marginally significant. The 2018 report's mean for good explanations (42.75) exceeds what eight items scored 1 to 5 can sum to, so the form tested is unclear, and the XAITK page asks for reliability data.
There are no norms or thresholds, so a score means something only against another condition in the same study, such as the current explanation or no explanation. A mean rating reported on its own, such as 4 out of 5, cannot support a claim that an explanation is satisfactory.
Versions with seven, eight and nine items are in circulation, item wording differs between them, and the form invites researchers to use a subset; Madumal et al., for example, report four items. Scores from different studies are therefore not comparable unless the same items and wording were used, and each study should report exactly which it used.

Resources

Research Papers

Measures for explainable AI: Explanation goodness, user satisfaction, mental models, curiosity, trust, and human-AI performance

Robert R. Hoffman et al.•Feb 6, 2023

Metrics for Explainable AI: Challenges and Prospects

Robert R. Hoffman et al.•Feb 1, 2019

The question addressed in this paper is: If we present to a user an AI system that explains how it works, how do we know whether the explanation works and the user has achieved a pragmatic understanding of the AI? In other words, how do we know that an explanainable AI system (XAI) is any good? Our focus is on the key concepts of measurement. We discuss specific methods for evaluating: (1) the goodness of explanations, (2) whether users are satisfied by explanations, (3) how well users understand the AI systems, (4) how curiosity motivates the search for explanations, (5) whether the user's trust and reliance on the AI are appropriate, and finally, (6) how the human-XAI work system performs. The recommendations we present derive from our integration of extensive research literatures and our own psychometric evaluations.

Software Packages

OpenScales: Explanation Satisfaction Scale (EXPSAT) scale definition

Structured format for representing questionnaires and scales, and a repository of freely-usable and distributable scales. - stmueller/OpenScales

Documentations

Explanation Satisfaction Scale (XAITK)

Kitware Inc•Jan 20, 2025

Explanation Satisfaction is an evaluation of explanations by a user. The reference is to the explanatory value to the user.

Related Techniques

Name	Description	Assurance Goals
Think-Aloud Study of Explanation Use	A think-aloud study of explanation use asks a small number of participants to say what they are thinking while they use an AI system's explanations to do a realistic task, so that the design team can see how the explanation is read, used, ignored or misread. It applies the think-aloud method from cognitive psychology, which rests on Ericsson and Simon's account of verbal reports: people can voice what they are attending to as they work without changing their thinking, provided they are not asked to report things they would not otherwise attend to, such as the reasons for their actions. In a session, a facilitator gives a short practice task, then sets tasks in which the participant sees the model's output and explanation and makes a decision, and while the participant works gives only neutral reminders, such as 'what are you thinking?', when they fall silent; audio and screen are recorded. The recordings are transcribed and coded, often by thematic analysis, into findings such as parts of the explanation that were misread, parts never looked at and questions it left unanswered, and the team changes the design and tests it again. The output is qualitative, design-stage evidence about how an explanation is understood: it can show that a misreading occurs, but not how often it occurs among users or whether their decisions improve.	Explainability
Forward Simulation Test	A forward simulation test checks whether people who have studied a model's explanations can predict the model's output for inputs they have not seen, so it tests with people the understanding an explanation method is assumed to give, where fidelity checks only compare the explanation with the model by computation. Doshi-Velez and Kim proposed it as a human-grounded evaluation of interpretability: participants are given an explanation and an input and must state what the model will output, whether or not that output is correct. In the controlled design of Hase and Bansal, participants study a set of examples with the model's outputs, predict the outputs for a separate set of test inputs, then study the same examples with explanations added and predict the same test inputs again, and the gain in accuracy is credited to the explanations (a separate group that studies the examples without explanations guards against a practice gain); the test inputs are kept apart from the explained examples so that explanations cannot give the answers away, and are chosen so that the model is right and wrong equally often, so that predicting the true label scores no better than chance. A counterfactual variant, in Hase and Bansal's form, shows an input, the model's output and its explanation, and asks what the model will output for an edited version of that input, with edits chosen so that half change the output. The result is the change in the share of correct predictions, with a confidence interval, for each explanation method or design compared; it is evidence that people can anticipate the model on the kind of input tested, not that they make better decisions with it.	Explainability Transparency
Appropriate Reliance Measurement	Appropriate reliance measurement tests whether people who receive an AI system's advice follow it when it is right and keep their own judgement when it is wrong, and whether an explanation, or another aid such as a confidence score, moves them towards that pattern. It uses a two-step decision study on a classification task, the only kind Schemmer et al. define it for: for each task item the participant first decides alone, then sees the AI's advice, with or without the explanation, and makes a final decision, on a set of items with known correct answers drawn so that the AI is wrong on a known share of them. Only items on which the first decision and the advice disagree are counted, because when they agree the final decision cannot show which source the person relied on. Schemmer et al. define two rates over these items: relative AI reliance, the share of items with a wrong first decision and correct advice on which the person switched to the advice, and relative self-reliance, the share of items with a correct first decision and wrong advice on which the person kept their own answer. Comparing the two rates between an explanation condition and a control condition shows whether the explanation reduces under-reliance (rejecting correct advice), reduces over-reliance (accepting wrong advice) or merely raises agreement with the AI, which overall accuracy or agreement with the AI does not show on its own.	Explainability Safety