Announcement

Collapse
No announcement yet.

How to lie with statistics, pt. 8,243,721.5

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    How to lie with statistics, pt. 8,243,721.5

    Top truth bending by the New South Wales ministry of health:

    #2
    How to lie with statistics, pt. 8,243,721.5

    Well, 43,000 nurses is a lot like "no nurses".

    Comment


      #3
      How to lie with statistics, pt. 8,243,721.5

      Did they start recruiting really tall nurses in 2011/12?

      Or maybe the ones they had before were just further away.

      Comment


        #4
        How to lie with statistics, pt. 8,243,721.5

        That's proper outstanding, that is.

        Hey kids, the closer you zoom in, the bigger things get.

        Brilliant.

        Comment


          #5
          How to lie with statistics, pt. 8,243,721.5

          They're not lying with statistics.

          They're lying with the misleading graphical representation of statistics.

          (the sort of thing you'd find in a "[strike]management[/strike] executive summary")

          Comment


            #6
            How to lie with statistics, pt. 8,243,721.5

            Lying with axes. Which reads more aggresive than it sounds!

            Also, notice that the last entry is a projection.

            Comment


              #7
              How to lie with statistics, pt. 8,243,721.5

              They only fibbed with the y-axis. It's fine otherwise.

              Comment


                #8
                How to lie with statistics, pt. 8,243,721.5

                You know, if that graph started from zero on the y-axis, it would be about 60 cm high.

                Comment


                  #9
                  How to lie with statistics, pt. 8,243,721.5

                  Axis is your flexible friend.

                  Comment


                    #10
                    How to lie with statistics, pt. 8,243,721.5

                    Can't we get the Mail excited about this?

                    How the Golden Generation of HEROES died fighting the Axis only to see this kind of abuse?

                    Comment


                      #11
                      How to lie with statistics, pt. 8,243,721.5

                      That graph is misleading, I would say, because the people-figures look like they're representing numbers, apart from what the axes and figures are telling you.

                      I've made a lot of graphs for investment banks. We do a lot of split-axes graphs, but generally, would start the axes like that. It really depends on whether they wanted to accentuate the change or not. But I really wouldn't use people-figures like that.

                      Comment


                        #12
                        How to lie with statistics, pt. 8,243,721.5

                        This is very lawyerly, but I think the question is whether pitch book standards of "accuracy" should be applied to public administration.

                        If that happens, we are all in deep trouble.

                        Comment


                          #13
                          How to lie with statistics, pt. 8,243,721.5

                          Oh I agree. I was really taking issue with SP's point about there being nothing wrong with it (technically).

                          Comment


                            #14
                            How to lie with statistics, pt. 8,243,721.5

                            I read him as only referring to the X-axis, which I don't have a problem with (assuming that this wasn't done at the end of 2013 or later).

                            Comment


                              #15
                              How to lie with statistics, pt. 8,243,721.5

                              I can't see where June is on that graph.

                              Comment


                                #16
                                How to lie with statistics, pt. 8,243,721.5

                                Second on the left on the fourth row, next to Sharon

                                Comment


                                  #17
                                  How to lie with statistics, pt. 8,243,721.5

                                  ursus arctos wrote: I read him as only referring to the X-axis, which I don't have a problem with (assuming that this wasn't done at the end of 2013 or later).
                                  Oh, I thought he meant it had to be foreshortened otherwise you couldn't see any difference unless it was really high.

                                  Think I need more coffee.

                                  Comment


                                    #18
                                    How to lie with statistics, pt. 8,243,721.5

                                    MsD wrote: Oh I agree. I was really taking issue with SP's point about there being nothing wrong with it (technically).
                                    I forgot to include quotes.

                                    Comment


                                      #19
                                      How to lie with statistics, pt. 8,243,721.5

                                      Ok, sorry, but I do have to work with problems of visual data representation - and although it's mostly been in banking, it's also been in NGOs, representing sensitive things like mortality. It's hard to show a significant change clearly without distorting or misleading, although your narrative, labels and titles should make it clear.

                                      Comment


                                        #20
                                        How to lie with statistics, pt. 8,243,721.5

                                        Southport Zeb wrote: Second on the left on the fourth row, next to Sharon
                                        Of course. It's hard to make any of them out when you put them in uniform.

                                        Comment


                                          #21
                                          How to lie with statistics, pt. 8,243,721.5

                                          MsD wrote: Ok, sorry, but I do have to work with problems of visual data representation - and although it's mostly been in banking, it's also been in NGOs, representing sensitive things like mortality. It's hard to show a significant change clearly without distorting or misleading, although your narrative, labels and titles should make it clear.
                                          I don't really see the justification for creating a graph based on only five data points, especially when the relative differences in magnitude are so small.

                                          Personally, if my intention was not to mislead the readers, I'd publish this data as a table.

                                          Comment


                                            #22
                                            How to lie with statistics, pt. 8,243,721.5

                                            A directional arrow between each set of columns with the change percentage e.g "+5%" would clarify a bit. That's what I'd do, probably, but in a serious report (where you want people to really absorb the info), yes, a table, maybe in addition to a graph.

                                            Comment


                                              #23
                                              How to lie with statistics, pt. 8,243,721.5

                                              Well, clearly the whole point of the exercise was to mislead the public, so a table would never have "worked".

                                              What's just as bad is that I guarantee you that the team that produced it sold it as an "infographic".

                                              Comment

                                              Working...
                                              X