Comments on: Dating Site Photo Effectiveness

By: Jeff Weir

Jeff Weir — Sat, 27 Feb 2010 03:53:29 +0000

I love the categories in these graphs. They really beg the question as to how effective a photo would be of you doing something interesting with a friendly animal in bed while on holiday.

By: uberVU - social comments

uberVU - social comments — Fri, 05 Feb 2010 07:14:10 +0000

Social comments and analytics for this post…

This post was mentioned on Twitter by Jon_Peltier: New on the PTS Blog: http://zz.gd/e7aba7 Dating Site Photo Effectiveness…

By: Jon Peltier

Jon Peltier — Tue, 02 Feb 2010 23:14:12 +0000

Here is what Jeff is talking about.

By: Jeff Weir

Jeff Weir — Tue, 02 Feb 2010 20:07:34 +0000

Jon…you’re going really dotty on us. Do a Google image search on Yayoi Kusama…she’s got an exibition here in Wellington New Zealand that I’m sure you would approve of

By: Jon Peltier

Jon Peltier — Tue, 02 Feb 2010 15:10:17 +0000

Ikkeman - The gridlines in the dot plot can be made more pronounced to help with identification, or the labels can be placed next to the points per Derek's suggestion. If you decide to use bars, just use bars, but don't clutter them with markers as well. If clearly marked, bars showing Deviation from Average also work. There are many ways to skin this cat, and there are multiple ways to do so effectively. Gary - I don't like Secondary Axes in Charts much myself, but this is one example of Secondary Axes that Work.

By: Gary

Gary — Tue, 02 Feb 2010 14:33:40 +0000

Actually the original (maybe turned 90 degrees) works great if you just relabel the axis to “Deviation from Average…” That way the zero-based bars are irrelevant. I like this more than the dot plot, actually, since there’s only one series. Maybe use data labels to indicate the actual effectiveness percentage values. I don’t like double-axis-ing it so much.

By: Thom Mitchell

Thom Mitchell — Tue, 02 Feb 2010 14:02:47 +0000

You may be a victim of your own effectiveness or powers of persuasion: you have made me a believer in bar charts with a zero base. I prefer the ability to identify small *relative* differences, rather than spreading out the data points of a dot plot to fit a horizontal scale of smallest [numerical] width, i.e., the range of the values. Color me a bar chart fan!

By: Stružák

Stružák — Tue, 02 Feb 2010 10:39:36 +0000

If there is one thing I really admire on the dot plot charts, it must be the fact, that it can be easilly made and read using just one or two colours. Well done!

By: derek

derek — Tue, 02 Feb 2010 10:25:23 +0000

If you need the labels to be associated with the dots, you can label the dots directly, although I prefer then lined up on the left.

By: ikkeman

ikkeman — Tue, 02 Feb 2010 08:03:22 +0000

I do think the dot plot runs the risk of loosing the relation between the data label and it’s data point.
It’s already in play in the chart above, but as the list of data labels expands, it becomes harder to determine which point correlates to which label.

The Gridlines help, but I think the barchart offers better readability.
Maybe a thin-bar chart with big dots at the end of each bar?