statrefs home‎ > ‎Main‎ > ‎Fundamentals‎ > ‎Common Errors in Statistics‎ > ‎

Eyeballing the Data



What is "eyeballing the data"?

Drawing conclusions from data by simply looking at the data, either in raw data form, or as expressed in graphs or charts. 

Graphical representations of data along with descriptive statistics should most always be a part of the initial review and analysis of data.  But most of the time, conclusions should not be based solely on a visual review of the data, but should also be supported by a test of hypothesis or other equivalent method.
 

The eyeballing game:  http://woodgears.ca/eyeball
 

Here are some notes, albeit tinged with a bit of sarcasm, on the perils of simply looking at data or simple graphics to draw conclusions.  The link below is to the R Help mailing list, and a thread that discusses the IOTT (inter-ocular trauma test).
 


Other terms that might describe the same issue:



Visualizing financial data
  • Vincent Granville
Many time series charts seem to exhibit a pattern: an up-trend, apparent periodicity, a stochastic process that seems not to be memory-less, and so on.  Look at the following picture, representing stock price simulations. Do you think there is an up-trend? 

Detecting Patterns with the Naked Eye
www.analyticbridge.com

To add to that image, I have added a triangle chart pattern:  http://www.flickr.com/photos/38563808@N00/11315285494/

A few years ago I wrote a program to add a number of common chart patterns to price series. What amazed me was how many patterns could be overlaid on a series, all conforming to strict rules of construction. Not only do our eyes play tricks, but we can easily write software to enhance those pattern finding illusions.






Comments