Introduction

This report critically evaluates the effectiveness of data visualization in media reporting, specifically examining a bar chart from a Daily Mail article that claims the Covid-19 pandemic was nearing its end.


Purpose of the Analysis


Part 1: Critique of the Original Graphic

The Daily Mail chart lacks proper labeling, has a limited timeframe, and omits key contextual data (e.g., absolute case numbers, deaths, and hospitalizations). Using percentage change alone is misleading and distorts public perception of the pandemic’s trajectory. Based on principles of effective data visualization (Tufte, Segel & Heer), the chart lacks integrity and fails to provide a clear understanding of the Covid-19 situation.

Part 2: Improving Data Representation

A line graph using absolute case numbers over two years (2020-2021) provides a more accurate depiction of Covid-19 trends. This visualization links peaks/troughs to policy changes (e.g., lockdowns, “Eat Out to Help Out”). By incorporating contextual annotations and improved labeling, the graph enhances interpretability and transparency. Comparing this to the Daily Mail’s visualization shows how media can manipulate data presentation to push a specific narrative.


Overall Contribution

This work highlights the importance of ethical and accurate data visualization, showing how misleading graphics can shape public perception. It advocates for better transparency in media reporting through a more robust approach to data presentation.





Part 1: Critique of Data

The Daily Mail article, “Chart that shows worst may be over”, uses a bar chart near the top of the page, rooting the narrative in the graphic, thus ‘establishing an overview’ (Segel and Heer, 2010). The article claims: “The end of Britain’s epidemic is in sight”, using the average percentage change of Covid-19 cases from 15 August 2021 to 7 November 2021.

The y-axis is not properly labelled, instead, the label appears at the top of the graph. The x-axis should include a year, along with the source of the data, both of which are absent. Additional context labels explaining Covid rates would improve clarity. The x-axis shows dates (in two-week intervals), while the y-axis represents percentage change of Covid daily rates (factorised in 10% increments, ranging from -20% to +20%), thus providing a ‘storytelling’ tool to ‘reveal stories within the data’ (Segel and Heer, 2010).

The graphic uses two colours strategically to convey impact and ‘specific meaning’ (Madden et al., 2000), making it easily digestible and easy to interpret. The bar chart uses simple colour visualisations, but does not enhance memorability (Madden et al., 2000). Days are easily identifiable, and negative percentage change can be compared day-to-day within the seven-day rolling average. Red was used to express danger (negative change) and blue for safety (positive change). However, recent discussions have highlighted limitations of the RGB (Red-Green-Blue) colour palettes due to their poor perceptual properties (Zeileis et al., 2019). The R package colorspace offers a broader range of colour palettes based on HCL (Hue-Chroma-Luminance), which better aligns with the human visual system.

Each bar represents a single day, allowing the reader to compare day-to-day percentage change, taken as a 7-day rolling average, showing peaks of percentage change (e.g., 24 August and 6 September 2021) and troughs (e.g., 13 September 2021).

Percentage change of Covid cases is too simplistic a metric to declare the end of a pandemic; it is also a misleading representation, which is a “graphical gimmick that uses several visual tricks to exaggerate trends” (Tufte, 1984). Percentage change should be accompanied by additional factors to provide better context, such as:

Further, “context is essential for graphical integrity” (Tufte, 1984), yet the article provides only a partial snapshot of data. There is no single measurement that can accurately indicate the impact of Covid, and the article’s declaration of the pandemic’s end is problematic, as a pandemic is defined as:

“An epidemic occurring worldwide, or over a very wide area, crossing international boundaries and usually affecting a large number of people” (Last, 2001).

The article fails to consider this definition, nor the complexity of declaring epidemic/pandemic status (Singer et al., 2021).





Part 2: Improving Data Representation





Drawing on Coronavirus Data UK, which contains Covid statistics within the UK, this website provides a variety of metrics, including dates, area type, code, and name. To create a more relevant graphic, absolute numbers of new Covid cases per day were chosen, covering two years from 2020 to 2021. In contrast, the article only presented data from 15 August 2021 to 7 November 2021 (without any labelled years).

The line graph presented here creates a clearer and more useful method of comparing data over time. It includes dates on the x-axis and new cases (per day) on the y-axis, which allows the reader to see trends and patterns over a significant period of time. This graph highlights troughs, such as the introduction of lockdowns, and peaks of new Covid cases, for example, during initiatives like ‘Eat Out to Help Out’. The graph provides context and explanations for Covid case trends over time.

It is more helpful to examine absolute numbers when looking at cases of Covid over a longer time-frame to assess whether cases were rising or falling in real terms. This method can then inform future government policy decisions.

The graph includes labels marked alongside a red ‘x’ and associated annotations, offering additional insight and contextualizing the Covid cases over the two years. These labels highlight the impacts of policies such as lockdowns and ‘Eat Out to Help Out’, giving the reader a better understanding of the reasoning behind the trends. As Tufte (1984) states, “data should be clear, detailed, and thorough,” with labelling used to defeat graphical distortion and ambiguity. The graph above adheres to this principle by providing comprehensive labels, helping the reader interpret the data accurately.

Furthermore, dark red lines have been included to show the length of the lockdown periods within each of the three lockdown phases. This allows for seamless transitions between sections, helping to connect different elements through the use of visualizations (Segel and Heer, 2010). The dark red colour, coupled with the x-axis labelling of the lockdown periods, acts as a distinctive visual cue that stands out to the reader (Madden et al., 2000).

Unlike the bar graph in The Daily Mail, the graph in this analysis includes a caption with the source of the data, the region (UK) it covers, and the years it spans, creating a more accountable and reliable graphic. The y-axis is clearly labelled, making the graph more interpretable and reducing the chance of misunderstanding.

The use of percentage change in The Daily Mail graphic meant that results could only be compared to the previous day, making it harder to interpret meaningful trends. Instead of relying on average percentage increase, ‘new cases per day’ in absolute numbers were used in this analysis. This method allows for comparisons across two years, making it easier to distinguish patterns and trends over time. The Daily Mail’s use of percentage change presents a biased interpretation of the data, skewing the findings to suggest the end of the pandemic.