Micah Fisher-Kirshner originally shared:
Diving into the summary analysis of +Moz Search Engine Ranking Factors, this year`s study is better than last year`s with how to express what the correlations mean.
The use of association at the start rather than the use of strong/high correlations gave me a small smile since it avoids incorrect assumptions on what the data means.
This does not mean that the study was flawless as standard correlation ranges was not used nor were they consistent with their correlation definitions (i.e.: a 0.2 or 0.3 correlation is not near a "decent correlation" #4 and a correlation 0.05 higher does not suddenly become a "reasonably strong correlation" #6), yet it provided a more cautious explanation for the deeper data points than years past.
Quick thoughts on the high level summary article
It would have been nice to have some top changes in a visual format vs a fully written post, could see some things stand out better that way.
Diving into the details of https://moz.com/search-ranking-factors page
The thing that stabs me in the heart the most is the use of a scatterplot icon for "View Correlation Data" which then proceeds to show bar charts... talk about dashing some hopes there!
I generally wish they had some visual comparisons for year over year data, would make for interesting trending analysis.
Making correct data accuracy or analyses is important, so I get worried when I see oddities in the results such as when I saw that # of nofollow backlinks has a higher correlation than of follow backlinks; something I`d recommend noting needs further analysis. Same goes for spam scores that show a positive correlation (yes, I know it`s low, but for sake of their preferred correlation ranges, might as well note something to look into more).
All in all
Look forward to seeing what additional analyses they provide in the coming days and weeks (whether from a curiosity standpoint or from a critiquing post).
Announcing the 2015 Search Engine Ranking Factors Study
Micah Fisher-Kirshner originally shared:
Diving into the summary analysis of +Moz Search Engine Ranking Factors, this year`s study is better than last year`s with how to express what the correlations mean.
The use of association at the start rather than the use of strong/high correlations gave me a small smile since it avoids incorrect assumptions on what the data means.
This does not mean that the study was flawless as standard correlation ranges was not used nor were they consistent with their correlation definitions (i.e.: a 0.2 or 0.3 correlation is not near a "decent correlation" #4 and a correlation 0.05 higher does not suddenly become a "reasonably strong correlation" #6), yet it provided a more cautious explanation for the deeper data points than years past.
Quick thoughts on the high level summary article
It would have been nice to have some top changes in a visual format vs a fully written post, could see some things stand out better that way.
Diving into the details of https://moz.com/search-ranking-factors page
The thing that stabs me in the heart the most is the use of a scatterplot icon for "View Correlation Data" which then proceeds to show bar charts... talk about dashing some hopes there!
I generally wish they had some visual comparisons for year over year data, would make for interesting trending analysis.
Making correct data accuracy or analyses is important, so I get worried when I see oddities in the results such as when I saw that # of nofollow backlinks has a higher correlation than of follow backlinks; something I`d recommend noting needs further analysis. Same goes for spam scores that show a positive correlation (yes, I know it`s low, but for sake of their preferred correlation ranges, might as well note something to look into more).
All in all
Look forward to seeing what additional analyses they provide in the coming days and weeks (whether from a curiosity standpoint or from a critiquing post).
