Data Visualization in Environmental Public Health

Data Visualization in Environmental Public Health: From Data to Insight

COMMON PITFALL WHY IT MATTERS

ENVIRONMENTAL PUBLIC HEALTH EXAMPLE

ANALYTICAL PITFALLS Comparing

County A identified 180 critical violations during 1,000 inspections, while County B identified 150 during 500 inspections. Based on counts alone, County A appears to have more critical violations. After accounting for the number of inspections, County A had 18 critical violations per 100 inspections, compared with 30 per 100 inspections in County B. The rates show that critical violations were identified more frequently in County B. A visualization shows higher asthma complaint rates near industrial sites and uses the title “Industrial Sites Cause Higher Asthma Complaints.” While the data show an association, they do not establish that industrial sites caused the increase. Other factors, such as traffic, population density, housing conditions, or reporting patterns, might contribute to the observed relationship. A visualization shows that 82% of inspections were completed on time, without indicating that the agency's target is 95% or that last year's performance was 90%. This information changes how the data are interpreted. A mosquito surveillance dashboard viewed in August still displays data last updated in April and indicates low mosquito activity. Because current surveillance data are missing, staff incorrectly assume activity remains low and make decisions based on conditions that no longer reflect the current situation. Inspection records are missing facility-type information for many establishments. A chart comparing violations by facility type only includes records with complete classifications, making some categories appear to have fewer violations than they do.

Raw counts are useful when the total volume matters. Counts can produce misleading comparisons, however, when groups differ in population size, number of inspections, number of participants, time observed, or another measure of opportunity or exposure. Rates place groups on a common scale by dividing each count by an appropriate denominator. The observed relationship between two variables does not establish that one caused the other. Visualizations should communicate associations accurately and provide enough context to prevent overinterpretation.

counts without using a common

denominator: Using counts when rates are more appropriate Confusing correlation with causation

Presenting findings without context or benchmarks

Without historical trends, goals, benchmarks, or relevant comparison groups, viewers can struggle to determine whether a result is meaningful.

Displaying outdated data

Outdated information might not reflect current conditions or priorities and can lead to inappropriate conclusions or decisions.

Using incomplete, inaccurate, or inconsistent data

Data quality problems can produce misleading visualizations even when the visualization itself is designed correctly.

Questions to Ask Before Finalizing Every visualization should answer: So, what? Before finalizing the visualization, consider the following questions. • Who is the intended audience, and what level of detail do they need? • What is the primary message or insight the visualization should communicate? • Is the visualization clear and accurate? • What context or limitations does the audience need? • Is the visualization accessible, and does it use readable text, appropriate color contrast, and clear labels? • What decision, discussion, or action should this visualization support?