Importance of Data Graphs (& Getting Them Right)
October 7, 2011 § Leave a comment
It is the ability of graphs to suggest theories and provoke questions which make them so important. Observing particular patterns allow us to move from the specific to the general, and up the information pyramid.
Data Content – It makes no sense to use graphs to display small amounts of data where text would suffice.
Data Relevance – Graphs are only as good as their data and their quality.
Graph Complexity – Graphs should be no more complex that the data they portray. Therefore it should be clean of redundant complexity (irrelevant decoration, colour, 3d effects).
A designer knows he has achieved perfection not when there is nothing left to add, but when there is nothing left to take away.
– Antoine de Saint Exupéry
Distortion – If any, should be obvious and with intent. Graphs, especially maps, may distort for abstractive purposes. Never distort data.
Message – If the message is simple, keep it simple. If the message is complex, make it simple.
Graphical Cognition – Some features can be consumed at a subconscious level, whilst others promote intense inspection.
Designing Effective Graphs.
To design effective graphs, we must understand which graphical attributes are most easily decoded.
Pie Charts, as of this writing, appear to be the most disliked forms of statistical graphs amongst information visualizers. They can be hard to decode, but can work well for ‘diving into data’, and when used as a whole in conjunction with other attributes such as color or scale. It is noted that pie charts are only useful for proportions.
Bar Charts, on the other hand, are very useful and can adequately represent all types of data (Nominal, Ordinal, Integer, Ratio/Proportional).
Histograms, can represent Distribution of values. Frequency Histograms use bar height to represent number of observations within a cell, whereas Relative Frequency Histograms use the Area of the bars to represent the Proportion of observations within each cell.
Time Series Data, are a set of observations made at equally spaced points in time.