TCC 315/PSY418 Class Discussion 2/11/97

Handout: CAD Comparison Graph:

WhatÕs Wrong with it?

* mismatch between categories & companies in upper and lower graph

* arrangement does not highlight what they want to show

* scale inordinately accurate for an estimate of comparison

* what is 100% successful test?

Which model should you buy?

* It is difficult to decide because of the 3-D presentation of data

What is the purpose of a graph:

* to display information quickly so that it is easily interpretable.

* to present information experientially.

Do you even need both the top graph and the bottom table?

Note that there is dominance in the choices:

Alternatives (Mechanical Desktop and Microstation Modeler) do worse in all categories than the others.

This means that a rational chooser will never pick the dominated alternatives.

Assignment (for next class): How would you redraw this graph?

Use of stem and leaf plots:

Can capture much experiential data that would otherwise be lost if you just computed a mean and standard deviation from the data.

The stem and leaf plots (or even histograms) give a more intutitive interpretation.

How to represent information badly in order to ÒfoolÓ or manipulate information

* Labor productivity vs. Japan: U.S. productivity constant; 3 data points cast in a 3-D representation in order to make the information appear more substantial.

* Public and private elementary schools: hiding the data in scale.

* Projections for the supply of U.S. Energy: no correspondence between representaion and the actual numbers

* US Trade Exports with China and Taiwan: changing the scales on two graphs side-by-side to create an illusion of comparability. Also, flipping color representation between graphs.

* Purchasing power of the diminishing dollar: Using an area representation to create the appearance of shrinkage instead of a linear representation.

* Views on Taxes: Omission of scales; choosing scales to magnify appearance of changes

* New YorkersÕ Trust of Newspapers: Changing scale in mid-axis.

* Life expectancy: Obscuring the data by alphabetizing it or by pairing it next to other groups.

* Commission payments to travel agents: burying data by obscuring it with tiny labels.

* Earnings per share and dividends: using a 3-D graph to obscure data that would be better presented in 2-D.