Comments on: Google Analytics Bummers http://blog.purevisibility.com/2008/06/google-analytics-bummers/ Mon, 26 Jul 2010 00:04:17 -0400 http://wordpress.org/?v=2.9.2 hourly 1 By: Daniel O'Neil http://blog.purevisibility.com/2008/06/google-analytics-bummers/comment-page-1/#comment-604 Daniel O'Neil Fri, 20 Jun 2008 21:36:45 +0000 http://blog.purevisibility.com/?p=181#comment-604 Hi Jessica, Corellation coefficients are basically comparisons of "fit" between samples. The only problem with that approach is that you don't always have a "fit" to match something to. In the context of Search Engine Marketing what you are usually doing is trying to determine variance across two or more data sets with no implication about historical trends, which suggests a better test would be ANOVA or chi-square tests. However, if you had enough historical data and could compare event "A" to event "B" correlation coefficients would be perfect. Hi Jessica,

Corellation coefficients are basically comparisons of “fit” between samples. The only problem with that approach is that you don’t always have a “fit” to match something to. In the context of Search Engine Marketing what you are usually doing is trying to determine variance across two or more data sets with no implication about historical trends, which suggests a better test would be ANOVA or chi-square tests.

However, if you had enough historical data and could compare event “A” to event “B” correlation coefficients would be perfect.

]]> By: jhullman http://blog.purevisibility.com/2008/06/google-analytics-bummers/comment-page-1/#comment-603 jhullman Fri, 20 Jun 2008 19:24:23 +0000 http://blog.purevisibility.com/?p=181#comment-603 Good question. I find correlation coefficients useful anywhere you're dealing with multiple variables, so they seem like they'd be really helpful on analytics, like anytime you're segmenting data. Yourself? Good question. I find correlation coefficients useful anywhere you’re dealing with multiple variables, so they seem like they’d be really helpful on analytics, like anytime you’re segmenting data. Yourself?

]]> By: Daniel O'Neil http://blog.purevisibility.com/2008/06/google-analytics-bummers/comment-page-1/#comment-602 Daniel O'Neil Fri, 20 Jun 2008 19:01:50 +0000 http://blog.purevisibility.com/?p=181#comment-602 Amen on Median. I'm not so sure about mode; in my experience the data from an analytics dataset is too granular and variable to infer anything about a single value. If you were to recommend that Google put one statistics measure into Analytics, what would it be? Amen on Median. I’m not so sure about mode; in my experience the data from an analytics dataset is too granular and variable to infer anything about a single value.

If you were to recommend that Google put one statistics measure into Analytics, what would it be?

]]>