Data Visualization: Principle and Best Practices

After collecting and organizing your research data, it is now the time to address your study’s research questions and hypotheses through the presentation of relevant data. Presenting data involves using various graphical techniques to show the reader the relationship between different data sets visually, to emphasize the nature of a particular aspect of the data, or geographically “place” data appropriately on a map.

Data visualization is the method of illustrating data as charts or graphs to aid in analysis and interpretation. These visuals make complex, large, and varied data sets more accessible and interactive with different stakeholders. 

Now, how do you choose the right chart type? 

The single most crucial step in data visualization revolves around what knowledge the research wants to impart to the reader. With the amount of data available today, it is very easy to mislead someone with a line chart, bubble chart, or scatter plot. This can happen if presenting data is not appropriate for the content, and there are also instances that the charts interpret the content differently. Now, let’s discuss the common chart types based on what situations they can be used in.

  • Comparison and Ranking
    • Bar Chart
      • It represents quantitative values with the length of the bars, which makes the differences of values more obvious.
      • It is ideal for qualitative or quantitative variables that have been categorized.
      • Vertical bar charts are best used when comparing values in time series data or when the magnitude is emphasized rather than the movement of a time series.
      • Horizontal bar charts are best used when using categorical data with long labels for a specific period.
        • Best Practices
          • Start the y-axis at zero. This is most commonly done to have a proportioned size of bars. There will be smaller differences between the bars.
          • Set the spaces between bars at approximately half the bar’s width.
  • Arrange the bars according to the magnitude whenever necessary.
  • Do not overdo the colors of the bars.
  • Trends over Time
    • Line Chart
      • It gives a quick view of the trend, acceleration, deceleration, and volatility of time series data. 
      • This can be used to answer these types of questions such as:
        • How many languages have gone extinct from 1990–2010?
        • How different were the wolf populations in Minnesota, USA, during the 1980s and the 2000s?
      • It allows you to look at multiple data sets and emphasizes the movement of the points in the time series.
        • Best Practices
          • Include a zero baseline, if possible.
  • The spacing of the time series units must remain constant.
  • If the curves do not overlap, you may label the lines directly.
  • Maintain a 2:3 or 3:4 height-to-width ratio.
  • Part-to-Whole Comparison
    • Pie Chart
      • It is the most popular yet most misused chart.
      • It is best used when the proportion (or percentage) holds more importance concerning the total.
      • It is applicable for data sorted into categories for a specific period.
        • Best Practices
          • Visualize no more than five categories.
          • Order slices correctly. Place the largest slice in the 12 o’clock position, then arrange the order in a clockwise manner.
  • Make sure all slice sizes add up to 100%.
  • Don’t use multiple pie charts for comparison. This will take more time for the brain to process the data. Instead, do a stacked bar chart. 
  • Stacked Bar Chart
    • It is a variation of the bar chart wherein each bar represents a whole, and the segments in each bar represent different categories of the said whole.
  • Correlation
    • Scatter Plot
      • It is ideal for showing clusters, patterns, and relationships.
      • It can be used for two or three quantitative variables.
        • Best Practice
          • When showing a trend, use trend lines but do not compare more than two.
  • Use quadrants or dividers when appropriate.

Image by Benjamin Nodar

  • Bubble Chart
    • It can be used to visualize a data set with two or four dimensions.
      • Dimensions 1 and 2: actual numerical values (visualized in the axes)
      • Dimension 3: representation of each bubble (larger bubble, larger value)
      • Dimension 4: data categories (use of different colors or hues)
    • This chart isn’t as widely used as others. You could include a short description of the charts to be understandable.

Image by Our World in Data

  • Geographical Data
    • Map
      • It is used for spatial or geographic analysis.
      • Thematic maps use the attributes of a location and represent the data using visual elements and relationships in the data.
        • Best Practice
          • For quantitative data with only positive values, choosing a color scheme at most two-color hues that would range from light to dark is recommended.

Image by Philippine Statistics Authority

Visual data presentations are effective communication media that show and convey data and information. As journal editors and reviewers scan through these presentations before reading the entire text, their importance cannot be disregarded. Let Journal Lab help you increase your manuscript’s likelihood of publication with our expertly tailored services for your specific research needs. Our Formatting service will tailor your manuscript’s format and overall layout to meet your style guide requirements. Data visualization will also be applied coherently and effectively convey the message of your manuscript. All you need to provide are (1) the name of your target journal/style guide and (2) the final draft of your manuscript. ​

Summary

PURPOSETYPE OF CHART
Comparison and RankingBar Chart
– It is ideal for qualitative or quantitative variables that have been categorized.
Trends over TimeLine Chart
– It gives a quick view of the trend, acceleration, deceleration, and volatility of time series data
Part-to-Whole ComparisonPie Chart
– Best used when the proportion (or percentage) holds more importance with respect to the total.
Stacked Bar Chart
– A variation of the bar chart wherein each bar represents a whole, and the segments in each bar represent different categories of the said whole.
CorrelationScatter Plot
– Ideal for showing clusters, patterns, and relationships.
Bubble Chart
– used to visualize a data set with two or four dimensions.
Geographical DataMap
– used for spatial or geographic analysis.

References

Nodar B. (n.d.) Using Quadrant Plots to Improve Program Performance. Isixsigma. https://www.isixsigma.com/tools-templates/graphical-analysis-charts/using-quadrant-plots-improve-program-performance/#:~:text=A%20quadrant%20plot%20is%20a,their%20efforts%20when%20measuring%20performance.

Hyerle, D. (2000). Thinking maps: Visual tools for activating habits of mind. In A. L. Costa & B. Kallick (Eds.), Learning and Leading with Habits of Mind: 16 Essential Characteristics for Success. (pp. 149–174). ASCD. http://www.mcoe.edu.my/Uploads/WMSTC2013_habits_of_mind.pdf

In, J. & Lee, S. (2017). Statistical data presentation. Korean Journal of Anesthesiology, 70(3), 267–276. https://dx.doi.org/10.4097%2Fkjae.2017.70.3.267

Jensen, E. (2008). Brain-Based Learning: The New Science Of Teaching and Training. Pearson Education. 

Monitoring, Evaluation, Accountability and Learning for Development and Humanitarian Professionals (MEAL DPRO Starter). (2019). Data Analysis, Visualization and Interpretation. https://mealdprostarter.org/n-data-analysis-visualization-and-interpretation/

Our World in Data. (2021). Life expectancy vs. GDP per capita, 2018. https://ourworldindata.org/grapher/life-expectancy-vs-gdp-per-capita

Philippine Statistics Authority (PSA). (2019). PSA releases the 2015 Municipal and City Level Poverty Estimates: Map of the 2009, 2012 and 2015 municipal and city level poverty incidence estimates by poverty level. https://psa.gov.ph/sites/default/files/Map_2015%20SAE.jpg

Royal Geographical Society. (2021). Section 3 – Data Presentation. https://www.rgs.org/CMSPages/GetFile.aspx?nodeguid=043e3992-2e9a-4b5d-86ae-72398ce0c93b&lang=en-GB

Stobierski, T. (2021, January 28). Bad data visualization: 5 examples of misleading data. Business Insights – Blog. Retrieved December 27, 2021, from https://online.hbs.edu/blog/post/bad-data-visualization

World Health Organization. (2020). Philippines Coronavirus Disease 2019 (COVID-19) Situation Report #42. https://reliefweb.int/sites/reliefweb.int/files/resources/WHO%20PHL%20SitRep%2042_COVID-19_30Jun2020.pdf