Pitfalls

Pages in Pitfalls

Pitfalls in metadata: 3 stories

What metadata should contain

The mean versus the median

Distributions

Correlations

Percentages versus percentage points

Normalising data

Correlation is not causation

Ecological fallacy

Small sample size effect

Comparing apples to oranges

Absolute and relative probabilities

File encoding

Missing values

Duplicates, aggregates and totals

Unstandardised data

Data type mismatches

Pitfalls with dates

Outliers

Ambiguity in data

Clean spreadsheets

Breaking scales

Scales in line charts

Width to height ratio

Respecting proportions

Preserving scales for comparisons

Go beyond the defaults

Accessible colours

Colour on colour

#EndRainbow

Perceptual uniformity

Save the pies for dessert

Line charts: double y axes

Line charts: line interpolations

Line charts: perception of differences between lines

Stacked charts

Map projections

Normalising data for mapping

Choropleth classification methods

Disputed territories

Data acquisition

Data quantification

Data transformation

Data aggregation

Anonymisation

Visual representation

Representing people