Google analytics reported visitor loyalty (probably unique IP addresses?) for one week in July 2008 as significantly* BIGGER than during one week in January 2007.
January 2007 (July 2008) :
- 8 (22) visitors visited between 7 and 14 times.
- 11 (27) visited 15-25 times.
- 11 (21) visited 26-50 times.
- 0 (32) visited 51-100 times.
Up to 29 (101) visitors (unique IP addresses) , other than my good-self, return frequently enough for me to assume they drop-by on a daily basis. Out of pure, unfettered, cussedness I am also assuming that at least half of these loyal visitors are naughty, naughty, spam-bots or or other bots of an icky nature, as opposed to pleasantly pert bots. This assumption still leaves me with about 50 regular, daily, visitors who may actually be people!
* Significance in a formal Statistical sense identified by using Excel’s t-test function for a one-tailed, independent groups t-test that lead to the rejection of the null hypothesis, h0, p< 0.001
h0 ‘= there are no more people reading my blog regularly in July 2008 than in January 2007’
The result is statistically very powerful but I have low confidence levels in it because of the low signal-noise ratio introduced by the way the variable (a loyal blog reading person) is operationalised (unique IP address) that introduces a lot of noise mostly from bots.
Even worse than low statistical confidence is my inappropriate test-selection. Inappropriate because although the data fulfills some of the assumtions of the independent groups t-test e.g. parametric, it is sufficiently naughty to potentially violate other assumptions such as truely independent groups.
In summary, we can probably ignore the statistical significance of the numbers because of all the non-number related issues.
Statistical escapades put aside, I am still convinced that the Wendy House has quite a few more regular readers now than in January 2007.