Topic Five Workbook: Visualising the Evidence Study Notes (Pre-Class)

This workbook reading emphasizes reframing chart creation as a means of evidence-based communication.
Visualisation transforms raw data into interpretable evidence by:
- Reducing complexity
- Enabling comparisons
- Revealing relationships
Different chart types serve various purposes:
- Bar and column charts for categorical comparisons
- Clustered columns for grouped comparisons
- Scatter plots for relationships
- Line charts for trends over time
Conditional formatting allows for quick pattern recognition within tables.
Dashboards combine multiple visualisations providing a holistic view for decision-making.
Design principles ensure interpretability: clarity, focus, simplicity, and logical layout.
Effective visualisation transforms structured data into actionable evidence rather than a neutral presentation.

Previous topics covered measures of central tendency, variance, and correlation.
Created various charts in Excel, including:
- Bar charts
- Clustered column charts
- Scatter plots
Initial focus may have been on technical exercises rather than practical applications of these techniques in organizations.
Visualisation is not merely about chart creation; it is a critical process for transforming data into actionable evidence.
Data alone does not inform decisions; it requires structure through visualisation, allowing patterns, differences, and relationships to become evident.
This section shifts focus from “doing” to “choosing”: understanding the appropriateness of different visualisations and their impacts on interpretation.
Key questions to consider:
- When is each visualisation appropriate?
- What does each visualisation reveal or conceal?
- How do visualisation choices shape interpretation?

Real-world data is often complex, containing many observations across multiple variables.
Visualisation simplifies data by summarizing key features.
Example: A bar chart summarizes average values across categories; including standard deviations reveals volatility.
Without simplification, decision-makers risk cognitive overload, making decision-making difficult.

Many decisions rely on comparing alternatives, such as the performance of different stores or property values between cities.
Visualisations facilitate comparisons, allowing for immediate visual differences rather than scanning numerical data.
Effective comparison requires careful design; cluttered charts or unclear labels can obscure important differences.

Key insights often arise from examining relationships between variables.
Examples:
- Whether income relates to property prices.
- Whether population density impacts public transport usage.
Scatter plots enable direct observation of relationships, where statistical measures alone may fall short.

The three functions—reduction, comparison, and relationship detection—improve decision quality.
Effective visualisation leads to better insights, allowing for informed decisions, while poor visualisation leads to misinterpretation and poor decisions.
Visualisation significantly impacts how evidence is understood and utilized.

Importance of selecting the most appropriate visualisation type to represent evidence.

Among the most common visualisations, suited for comparing categories (e.g., average sales across stores).
Strength lies in clarity—height or length differences facilitate easy comparisons.
Effective bar/column charts must have:
- Clearly labelled categories
- Ordered values
- Removal of unnecessary visual elements.

Enable grouped comparisons by displaying multiple variables for each category.
Limitations include visual clutter due to clustering and challenges in detecting relationships between variables.

Display pairs of values from two continuous variables, revealing:
- Direction of the relationship (positive or negative)
- Strength of the correlation
- Form (linear or non-linear).
Essential for discovering relationships, not merely for calculating correlation.
Highlight the limitation of relying solely on numerical measures, allowing for detection of non-linear patterns.

Different visualisations may yield different interpretations of the same dataset.
Example: The same data shown as a clustered column chart versus a scatter plot can lead to different insights.
Visualisation choices dictate visibility of relationships, where poor choices can hide meaningful patterns.

Ideal for displaying sequential data over time, highlighting trends, growth, or fluctuations.
Effective for identifying patterns in data flow (e.g., sales trends).
Important considerations:
- Horizontal axis representing time
- Consistent scales to avoid misleading representations
- Clarity by limiting number of lines to avoid confusion.

Allows data visualisation within tables using techniques such as:
- Colour scales (heatmaps)
- Data bars
- Icon sets.
Enables pattern recognition directly in data without separate charts—useful for large datasets.

As the visualisation toolkit grows, the challenge shifts to selecting the right visualisation type for the data story.
Key questions to match:
- What kind of values are being compared? (Bar/column charts are useful)
- What relationships are being explored? (Scatter plots are ideal)
- What trends are evident? (Line charts display trends over time)
Proper combination of visualisation types enriches the data narrative, making it comprehensible and actionable.

Defined as a curated collection of visualisations designed to answer specific questions or support decisions.
Each chart in a dashboard must be included deliberately to contribute meaningfully to a central question (e.g., performance comparisons, sales trends).

Dashboards benefit from the combination of different chart types to answer diverse analytical questions, enhancing overall data exploration.

Effective dashboards need clear focus, avoiding unnecessary complexity.
Emphasis on simplicity to aid interpretation.
Use clear titles and labels, ensuring intuitive understanding without extensive explanations.
Logical layout guides viewing order for effective narrative flow and comprehension.

Statistics underpin the creation of dashboards by summarizing data accurately:
- Measures of central tendency: mean, median, mode.
- Measures of dispersion: variance, standard deviation.
- Measurements of linear relationships: covariance, correlation.
Statistics ensure that data represented in dashboards is accurate and insightful, enabling decision-makers to act confidently based on correct information.

Emphasizes an essential shift in perspective: the goal is not simply to create charts, but to effectively select visualisations that support specific claims about data.
Visualisations are non-neutral, actively shaping interpretation and understanding within evidence-based decision-making.
Data must be structured, interpreted, and effectively communicated for it to represent evidence adequately.
Statistics is essential, providing a robust foundation for data visualisation and the development of trustworthy insights.