A1-NateAnthony
From CS294-10 Visualization Fa07
[edit] bad (from Good magazine):
criticized graph
- contrast of precision with apparent meaninglessness
* precision: * three significant digit precision of teacher salaries by state * apparent meaninglessness: * colors for each state * graphic depictions of teachers
- repetitive labels: “above avg. worker's salary”
- no order to the data points
- U.S. not comparable to Canada:
* vacation days paid? * annual salary?
deconstruction:
- data model:
* relational: * state codes: nominal characters * statistical: * teachers above avg. worker's salary: ordinal floating point * avg. salary: ordinal floating point
- conceptual
* teachers above avg. worker's salary: percentage * avg. salary: dollars
- data types:
* state code * percentage of teachers above avg. worker's salary * avg. salary in dollars per year
- dimensions and measures
* dimensions:
* domain(state codes): two characters
* independent variable
* domain(salaries): greater than zero
* measure:
* range(percentage of teachers above avg. worker's salary): 0-100
* dependent variable
- encoding:
* labels * state codes as characters * percentage of teachers above avg. worker's salary * avg. salary in dollars * ordered by avg. salary in state (although not at all apparent on first glance) * apparently random encoding: color and depictions of teachers
my version:
- this version reveals that AK was duplicated in the original version! (One of the data points presumably representing AR.)
deconstruction:
- 2 dimensions: zones (on a map of the U.S.)
- shading represents teacher pay
== better (also from Good magazine):
==
well-done:
- “ordered” colors
- visual sense of order of magnitude difference across income classes
- meaningful order of delineation
- informative labels
* lots of information on one page
- visual differentiation of pie graphs from histogram through style
criticism:
- unclear meaning of shading in pie graph
- inaccurate scaling?
