PowerPoint, evidence, visualization and charts (I’m just annoyed…)

This week’s reading covered Edward Tufte’s critique of PowerPoint, his warning about charts, tables and graphs and Ben Schmidt’s analysis of history dissertations.  In a nutshell, Tufte HATES PowerPoint.  In  Beautiful Evidence, he says it is presenter oriented not audience oriented.  It reduces information to bullet points and limits the amount of text/charts/graphs that can be put onto each slide.  That means that information gets broken up and makes it more difficult to analyze.  He states “…it harms the quality of thought for producers and consumers of information.”    In PowerPoint presentations, information appears and disappears quickly making it difficult to follow the trail of the presenter’s reasoning and for the audience to do their own reasoning.  It reduces arguments to lists which leave out details, narratives and relationships.  He argues that a written report, printed out for each attendee (he refers mostly to meetings, not necessarily conferences) contains the most detail and information, is easier to follow and allows each person to take the information with when he/she leaves.   In Visualizing Evidence, Tufte uses the incident of a cholera outbreak in an area of London in 1854 to illustrate a point about charts and graphs.   Cholera is transmitted via infected water but before this outbreak it had been believed to be transmitted by air.  A physician tracked the cause of the outbreak to a particular well pump and convinced the London city council to cut off access to that well.   The cases of cholera dropped dramatically and returned to normal levels.  Charts showing the number of cases by week seemed to point to stopping access to the well as the main reason for the end of the outbreak.   But reading the full report from the doctor showed that other factors such as reduced population (many people had left the area by the time access was cut off, so a reduced population would also account for fewer cases) might account for the break in the epidemic although cutting off access probably helped prevent a recurrence of the cholera.  The point is that charts and graphs and tables contain limited information and that information can be easily manipulated depending on how it is grouped, so Tufte says it’s best to assume that any charts/graphs/tables that appear in a presentation are the best of all the possible results chosen expressly to advance the presenter’s case.

Ben Schmidt’s articles are about his inquiry into what years historians write about the most and what years are turning points in history.   He looked at American history dissertations submitted over a 120 year period and looked at the years included in the titles to see which time periods were most written about .  In his graphs, he found spikes at the years of the Civil War and the two World Wars (not really a surprise).  In his next analysis, he made an assumption that writers would pick years with round numbers – ending in either zero or five and then assigned a variable to years that ended in other numbers.  He then graphed the results showing spikes for certain years and how they should up more times than might be expected based in his assumption.  The years that coincided with big historical events popped up more often (what a surprise!)

Frankly, I’m annoyed!  I do agree with Tufte about being skeptical of charts, etc.  It’s far too easy to manipulate the data to make it fit a predetermined conclusion.  Also, his points about PowerPoint are well-made – it is a fairly inflexible format with a limited ability to convey technical or complicated information, BUT it is (in my opinion) incumbent upon the presenter to choose the method that best conveys all the information necessary and shame of people on those fields that rely on complicated and technical information for choosing a method that isn’t up to the task.    As for Ben Schmidt’s “investigation” – it seems pointless to me.  Historical events/periods/eras, etc don’t conform to nice, neat sets of time.  It seems logical to me that years associated with significant events are, of course, go to show up more often in dissertations than those that don’t.  His scope of inquiry was rather limited as he only choose dissertations with numbers in the title (what percentage of the whole is that?) and his assumptions of what years would be chosen is absurd.  I think he was just wasting time and trying to put off working of his own dissertation.  So between Tufte’s diatribe and Schmidt’s pointless charts, I’m just annoyed.

PS Check out the September 21 Dilbert comic strip (brilliant!) http://www.dilbert.com/