7 responses

  1. Jon Peltier
    October 29, 2009

    If you have many series, you could split the chart into multiple panels, as in the example in How to Build a Simple Panel Chart.

    Reply

  2. Ross
    October 29, 2009

    Hi Guys,

    Should this chart be a table…

    Reply

  3. AoverT
    October 29, 2009

    Nice post. You’ll find interaction charts are used a lot in the biomedical sciences to show when changes in particular outcomes, let’s say the effect of a drug, are dependent on other factors, for example age or sex.
    Jon, I absolutely agree that panels make for a really time efficient and data rich method for visualising interactions across many variables.
    Ross, I’m not sure about a table beingt he best option. To me, the chart would help the reader understand the nature of the interaction much better than a table. A table with a sparklines/microcharts may be even better. I suppose it comes down to what question you’re answering with the chart. Are you more interested in the exact figures (in which case I totally agree with you) or is the main story the interaction (in which case I’d go with the charts or table with sparklines).

    Reply

  4. Chandoo
    November 3, 2009

    @Jon.. good suggestion.

    @Ross: yeah, you can make a table, As AoverT pointed out, tables lack visual appeal, so chart might more effective to prove a point.

    @AoverT: Very good remarks. Thanks for sharing your experience.

    Reply

  5. Chris
    November 13, 2009

    Chandoo,

    you wouldnt’ know how to do that version of interaction chart easily in excel
    http://flowingdata.com/2009/11/10/do-we-need-more-teachers/sat-scores/
    Do I miss sth or is it really necessary to fake the axis through dummy series? i’m not seeing the forrest for the trees right now

    Reply

    • Chandoo
      November 13, 2009

      @Chris… I think you have to fake the axis thru dummy series. Good link btw.

      Reply

  6. Harish Negi
    October 18, 2016

    Oh, it is cool information. Will try it. thanks

    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top
mobile desktop