# Likert Scales

Source: https://www.potatoannotator.com/docs/annotation-types/likert-scales

Likert scales allow annotators to rate items on a numeric scale, commonly used for sentiment intensity, agreement levels, and quality ratings.

![Likert scale rating interface](/images/docs/likert-scale.png "Likert scale with labeled endpoints for rating tasks in Potato")

## Basic Configuration

```yaml
annotation_schemes:
  - annotation_type: likert
    name: agreement
    description: "How much do you agree with this statement?"
    size: 5
    min_label: "Strongly Disagree"
    max_label: "Strongly Agree"
```

## Configuration Options

### Scale Size

Define the number of points:

```yaml
size: 5  # 5-point scale (1-5)
size: 7  # 7-point scale (1-7)
size: 10 # 10-point scale (1-10)
```

### Endpoint Labels

Label the scale endpoints:

```yaml
min_label: "Not at all"
max_label: "Extremely"
```

### Middle Label

For odd-numbered scales, label the midpoint:

```yaml
size: 5
min_label: "Strongly Disagree"
max_label: "Strongly Agree"
mid_label: "Neutral"  # Shown at position 3
```

### All Labels

Label every point on the scale:

```yaml
size: 5
labels:
  1: "Strongly Disagree"
  2: "Disagree"
  3: "Neutral"
  4: "Agree"
  5: "Strongly Agree"
```

## Common Scale Types

### Agreement Scale (5-point)

```yaml
- annotation_type: likert
  name: agreement
  size: 5
  min_label: "Strongly Disagree"
  max_label: "Strongly Agree"
```

### Frequency Scale

```yaml
- annotation_type: likert
  name: frequency
  size: 5
  labels:
    1: "Never"
    2: "Rarely"
    3: "Sometimes"
    4: "Often"
    5: "Always"
```

### Quality Scale

```yaml
- annotation_type: likert
  name: quality
  size: 5
  min_label: "Very Poor"
  max_label: "Excellent"
```

### Importance Scale

```yaml
- annotation_type: likert
  name: importance
  size: 5
  min_label: "Not Important"
  max_label: "Extremely Important"
```

### Satisfaction Scale

```yaml
- annotation_type: likert
  name: satisfaction
  size: 5
  min_label: "Very Dissatisfied"
  max_label: "Very Satisfied"
```

## Keyboard Shortcuts

Enable number key shortcuts:

```yaml
- annotation_type: likert
  name: rating
  size: 5
  keyboard_shortcuts: true  # 1-5 keys select ratings
```

## Display Options

### Horizontal Layout (Default)

```yaml
- annotation_type: likert
  name: rating
  size: 5
  display: horizontal
```

### Show Numbers

Display numeric values:

```yaml
- annotation_type: likert
  name: rating
  size: 5
  show_numbers: true
```

## Multiple Likert Scales

Rate multiple aspects:

```yaml
annotation_schemes:
  - annotation_type: likert
    name: clarity
    description: "How clear is this text?"
    size: 5
    min_label: "Very Unclear"
    max_label: "Very Clear"

  - annotation_type: likert
    name: relevance
    description: "How relevant is this text?"
    size: 5
    min_label: "Not Relevant"
    max_label: "Highly Relevant"

  - annotation_type: likert
    name: quality
    description: "Overall quality rating"
    size: 5
    min_label: "Poor"
    max_label: "Excellent"
```

## Multirate (Matrix Rating)

For rating multiple items on the same scale:

```yaml
- annotation_type: multirate
  name: aspect_ratings
  description: "Rate each aspect"
  items:
    - "Clarity"
    - "Accuracy"
    - "Completeness"
    - "Relevance"
  size: 5
  min_label: "Poor"
  max_label: "Excellent"
```

This displays as a matrix where each row is an item and columns are scale points.

## Semantic Differential

Rate between two opposite concepts:

```yaml
- annotation_type: likert
  name: tone
  size: 7
  min_label: "Formal"
  max_label: "Informal"

- annotation_type: likert
  name: sentiment
  size: 7
  min_label: "Negative"
  max_label: "Positive"
```

## Required vs Optional

Make the rating required:

```yaml
- annotation_type: likert
  name: rating
  size: 5
  required: true
```

## Default Value

Set a pre-selected value:

```yaml
- annotation_type: likert
  name: rating
  size: 5
  default: 3  # Pre-select middle value
```

## Full Example: Survey Task

```yaml
annotation_task_name: "Content Quality Survey"

annotation_schemes:
  - annotation_type: likert
    name: clarity
    description: "How clear and understandable is this content?"
    size: 5
    min_label: "Very Unclear"
    max_label: "Very Clear"
    keyboard_shortcuts: true

  - annotation_type: likert
    name: accuracy
    description: "How accurate is the information?"
    size: 5
    min_label: "Very Inaccurate"
    max_label: "Very Accurate"
    keyboard_shortcuts: true

  - annotation_type: likert
    name: usefulness
    description: "How useful would this be for the target audience?"
    size: 5
    min_label: "Not Useful"
    max_label: "Very Useful"
    keyboard_shortcuts: true

  - annotation_type: text
    name: feedback
    description: "Any additional feedback? (Optional)"
    required: false
```

## Best Practices

1. **Use odd numbers** for scales with a neutral midpoint
2. **Use even numbers** to force a direction (no neutral option)
3. **5 or 7 points** work best for most applications
4. **Label endpoints clearly** to anchor the scale
5. **Be consistent** across your project with scale direction
6. **Consider keyboard shortcuts** for faster annotation