The concept of proxies and their validity when attempting to study an immeasurable statistic was interesting to me.  Data visualizations are meant to give the person consuming the information an objective, omniscient feeling.  Proxies in their definition are subjective in that they are a correlation but rarely if ever provide a causation.  The example provided in Ways of Knowing was the use of facial expressions for emotions.  Facial expressions are supported to be useful tools for measuring this, but although this may be one of the most productive ways of measurement, I am not sold on it’s fruitfulness.  How can one say that a proxy is accurate when the intended data is subjective?  Emotion is a feeling unique to each person.  One person’s maximum sadness or anger will be different from another persons.  Also, their physical expressions will differ from others.  Should there be parameters on what can and can’t be proxied?  It seems that it is not up to the audience to decide this, as a data visualization assumes validity and accuracy to readers.  Is it imperative that researchers be held ethically responsible for proxy use in data visualizations?

In a way I feel that these questions are futile because proxies will always be used because they are often the most effective way of knowing.  Perhaps the best course of action would be to treat proxies in the same way we do representation.  A proxy is a reference and not a reality.