Skip to main content

Glossary

This glossary defines the key terms used throughout the MUSE platform. Terms are listed in alphabetical order. Japanese translations are provided for each entry.

Activities​

Japanese: アクティビティ

The concrete actions taken by a program or intervention. Activities are the starting point of a Logic Model — they represent what the program actually does (e.g., delivering training sessions, distributing resources, providing coaching). Activities produce Outputs, which in turn lead to Outcomes and ultimately Impact.

Auto Layout​

Japanese: 自動レイアウト

A tidy-up pass that arranges canvas nodes into a left-to-right flow. Auto Layout runs automatically once, immediately after AI generation, and can be re-triggered at any time from the More menu on the canvas toolbar. Auto Layout only affects visual positions; it does not change titles, descriptions, metrics, or connections.

Canvas​

Japanese: キャンバス

MUSE's interactive workspace for building Logic Models. The canvas is powered by React Flow and allows users to create, connect, and annotate nodes representing the stages of a Logic Model. Users can drag and drop nodes, draw edges between them, and attach Evidence Cards. The canvas view is split into a Canvas tab (for building) and a Recipe tab (for viewing generated Measurement Recipes).

Cautions​

Japanese: 注意点

A short list (1–3 items) of practical caveats attached to each metric in a Measurement Recipe. Cautions describe biases, ethical issues, or data-quality risks that are specific to that metric — for example, self-report bias in a survey question or seasonality effects on attendance counts.

Citation Count​

Japanese: 引用数

The number of times an academic paper has been cited by other papers. In MUSE, citation counts are displayed on external paper cards in the evidence dialog and are used to rank papers by quality. Higher citation counts generally indicate greater influence in the academic community.

Data Collection Method​

Japanese: データ収集方法

The paragraph in a Measurement Recipe entry that describes how the data for a metric is gathered — for example, an online survey, administrative records, direct observation, or web analytics. The Recipe agent derives this from the metric name and the parent Output/Outcome card's context.

Edge​

Japanese: エッジ

An arrow connecting two nodes on the canvas. Each edge represents a causal or logical relationship between two stages of a Logic Model — for example, an arrow from an Activity node to an Output node asserts that the activity produces that output. Edges are directional and can be annotated with supporting evidence.

Effect​

Japanese: エフェクト

The observed outcome or impact of an intervention as reported in a research study. In MUSE, each Evidence Card is tagged with one of five effect categories — Positive, No Effect, Mixed, Side Effect, or Unclear — to summarize what the study found. See the Effect Categories reference for full definitions.

Evidence Card​

Japanese: エビデンスカード

A standardized format for presenting a single piece of research evidence in MUSE. Each Evidence Card captures the core finding of a study as an Intervention → Indicator → Short-term Outcome relationship, along with the effect category and strength-of-evidence rating. Evidence Cards can be attached to nodes on the canvas to ground a Logic Model in empirical research. Note: Evidence Cards are distinct from External Papers — curated evidence is scored by AI, while external papers are shown as reference material only.

External Paper​

Japanese: 外部学術論文

An academic paper retrieved from external databases (such as Semantic Scholar) that is displayed alongside curated evidence as reference material. External papers are not scored by MUSE's AI. They provide supplementary context from the broader academic literature, including TLDR summaries and citation counts to help users quickly assess relevance.

Goal​

Japanese: ゴール

The plain-language description of what you want to achieve, which you provide to MUSE's AI to generate a logic model. A clear, well-structured goal that includes the target population, problem, and desired change produces the best results.

Impact​

Japanese: インパクト

Long-term societal change; the ultimate goal of an intervention. Impact sits at the far right of a Logic Model and represents the broad, lasting difference that a program aims to make in the world — for example, reduced poverty, improved public health, or increased educational attainment. Impact is typically harder to measure directly than shorter-term outcomes and may only become visible years after the intervention.

Intermediate Outcome​

Japanese: 中期アウトカム

Medium-term changes resulting from short-term outcomes. In a Logic Model, Intermediate Outcomes represent the behavioral, social, or systemic shifts that occur after the immediate effects of an intervention take hold — for example, increased economic self-sufficiency following improved financial literacy. Intermediate Outcomes bridge Short-term Outcomes and long-term Impact.

Logic Model​

Japanese: ロジックモデル

A visual representation that maps the causal pathway from Activities to Impact. A Logic Model in MUSE consists of interconnected nodes — Activities, Outputs, Short-term Outcomes, Intermediate Outcomes, and Impact — connected by directed edges. Each causal link can be grounded in Evidence Cards to demonstrate that the assumed relationship is supported by research.

Maryland SMS​

Japanese: メリーランドSMS（科学的手法尺度）

Maryland Scientific Methods Scale. A 0-to-5 rating system used in MUSE to assess the methodological rigor of research evidence. Level 0 denotes non-experimental mathematical models; Level 5 denotes a full Randomized Controlled Trial (RCT). The SMS helps users quickly gauge how much causal confidence to place in a given study. See the Strength of Evidence reference for full level descriptions.

Measurement Recipe​

Japanese: 測定レシピ

A practitioner-ready measurement plan generated by MUSE's Recipe agent from the metrics on a Logic Model. Each Recipe entry contains Measurement Steps, a Data Collection Method, a Frequency, an optional Target Value, and Cautions. Recipes are only generated for Output, Short-term Outcome, and Intermediate Outcome cards. See the Measurement Recipe user guide for full detail.

Measurement Steps​

Japanese: 測定手順

The ordered list (3–6 items) of concrete, imperative actions that a program operator can execute to actually collect the measurement for a metric. Measurement Steps are the core of a Measurement Recipe entry and are written to be executable by an M&E beginner using only basic tools (spreadsheet, mini-survey, attendance log).

Metrics​

Japanese: メトリクス

Lightweight, measurable indicators attached to Logic Model nodes. A MUSE metric carries only a name and an optional description — the operational details (measurement method, frequency, target value, cautions) are generated separately by the Measurement Recipe agent from the name plus the parent card's context. This two-stage design keeps the Logic Model as a lightweight outline and puts the practitioner-ready measurement plan into a dedicated Recipe.

Node​

Japanese: ノード

A card on the canvas representing a single stage in a Logic Model. Each node has a type (Activity, Output, Short-term Outcome, Intermediate Outcome, or Impact), a title, an optional description, and optional Metrics. Nodes are connected to each other by Edges to form the causal chain of the Logic Model. Evidence Cards can be linked to nodes to indicate which research supports the causal relationship flowing from that node.

Node Editor Dialog​

Japanese: ノードエディタダイアログ

The centered, two-column modal used to add a new node or edit an existing one. The left column holds the node type, title, and description; the right column holds the metrics list, where each metric is an inline-expanding card (only one expanded at a time). Opened by clicking Add Node in the toolbar, or by double-clicking / pencil-clicking an existing node on the canvas.

Outputs​

Japanese: アウトプット

The direct products or deliverables that result from carrying out Activities. Outputs are tangible and immediately countable — for example, "100 participants trained" or "50 reports published." Outputs are distinct from Outcomes: they describe what was produced, not what changed as a result. In a Logic Model, Outputs sit between Activities and Short-term Outcomes.

Private Mode​

Japanese: 非公開モード

A workspace setting that keeps its logic models inside the workspace. While it is on, external paper search is not run during AI generation and link sharing is unavailable; turning it on revokes every existing share link in the workspace. See Accounts & Workspaces.

RCT​

Japanese: RCT（ランダム化比較試験）

Randomized Controlled Trial. The gold standard of experimental research, corresponding to Level 5 on the Maryland SMS. In an RCT, participants are randomly assigned to either an intervention group or a control group. Random assignment ensures that — on average — the two groups are equivalent at baseline, allowing researchers to attribute any subsequent difference in outcomes to the intervention itself. RCT evidence provides the highest degree of causal confidence available.

Semantic Scholar​

Japanese: Semantic Scholar（セマンティック・スカラー）

An AI-powered academic search engine developed by the Allen Institute for AI. MUSE can optionally search Semantic Scholar to find academic papers related to causal connections in your logic model. Papers from Semantic Scholar appear as reference material (External Papers) and are distinct from MUSE's curated evidence.

Japanese: 共有リンク

A link that opens a read-only view of a saved logic model without signing in. The owner turns it on with Anyone with the link in the Share dialog and can turn it off again at any time. See Saving & Sharing.

Short-term Outcome​

Japanese: 短期アウトカム

Immediate changes in knowledge, attitudes, skills, or behavior that result from Outputs. Short-term Outcomes are the first measurable effects that participants or target populations experience after an intervention is delivered. They are typically observed within weeks or months of the intervention and serve as leading indicators that the program is on track toward longer-term Intermediate Outcomes and Impact.

Version​

Japanese: バージョン

A snapshot of a logic model created each time you click Save. All versions are kept in the model's history, and restoring one copies it as the newest version. See Saving & Sharing.

Workspace​

Japanese: ワークスペース

The place where saved logic models are stored. Every account has a personal workspace, and shared workspaces can be created to collaborate with invited members, each with the role Owner, Admin, or Member. See Accounts & Workspaces.