Sue Waters Blog

Latest Statistics Say My Blogs Are……?

| 24 Comments

Unfortunately it is very easy to misinterpret and/or mis-use any type of statistics.

People often look at the number while failing to consider how the statistic was calculated or what it truly means.

About Misinterpreting Statistics

Here’s an example:

Statistics show that 45% of the population can’t read the newspaper.

Shocking literacy rates! Or is it? How many included in that statistic were too young to read, didn’t speak English, had some form of disability etc? What percentage of the entire population was sampled, what method was used, did the method bias the result etc?

Using Statistics in Blogging

So where am I going with this? Well bloggers love statistics and they love to know how they rank against other bloggers. Since Technorati authority is no longer reliable bloggers have looked at other options.

So some bloggers are using PostRank. For example, here are the top blogs on education based on their ranking by PostRank. Trouble is bloggers are looking at the statistics and the number 1-10; not considering how it was calculated, how blogs are ranked against each other using PostRank or what these numbers mean.

Effective use of PostRank

Let me be clear! I love PostRank. It is an incredibly valuable tool for quickly analyzing and comparing all of your blog posts in terms of number of:

  1. Comments
  2. Bookmarked (Delicious, diigo etc)
  3. Twittered
  4. Linked to

All of which helps you reflect how the different post types impact how readers engage with the posts. For example, if your aim is a long informative post you would expect few comments but hopefully lots of bookmarking and/or linking. PostRank helps you work out if you achieved this goal.

Misinterpreting PostRank

But if you are using PostRank to compared your blog’s performance against another blog, or identify the best blogs for a topic than you need to look more closely at their statistics.

In particular look at those eye icons that represent views. What do they mean? Well they are the number of your readers that click the post title in the PostRank widget in your sidebar.

Should high clicking on the PostRank widget in a sidebar make a post (and blog) high ranking?

Below is a screenshot from PostRank. The example on the left is a perfect 10 from another blogger (educational) whose rank on that post is entirely based on click on the PostRank widget. While The Edublogger post had high bookmarking, linking and comments.

And if you’re enjoying this blog, please consider Subscribing for free!

Author: Sue Waters

Edublogs Support Manager @suewaters on Twitter

24 Comments

  1. Too right!

    Just for a bit of fun, here are a few crazier statistics:

    This airline has never had an accident! (Woops, we have no aeroplanes.)

    Exactly 93.42% of every statistic ever published is false.

    102% of the population voted for the president.

    A fair coin tossed at random gives equal numbers of heads and tails: 53% of each.

  2. Since you address a number of concerns here that you expressed on Larry Ferlazzo’s post, I’ll simply link to my comment there that should address most of the analysis misconceptions.

    http://larryferlazzo.edublogs.org/2009/05/16/the-most-popular-blogs-that-might-also-useful-to-educators/#comment-5031

    However, do feel free to ping me again with additional questions or concerns.

    • Hi Melanie, thanks for taking the time to discuss this. I’ve also responded (as you know by email) but have posted my response on Larry’s post under yours.

  3. I think Melanie has covered most of this but I wanted to clarify one geeky point:

    “In particular look at those eye icons that represent views. What do they mean? Well they are the number of your readers that click the post title in the PostRank widget in your sidebar.”

    Its actually the view of a post that we measure not just the click on the link. So if people view a post a lot more – without any other modes of engagement then we want to capture that variation. A single view has an engagement impact of 0.001 compared to a tweet that is 10. So really discounted. Its only there so if views is all you have you have some relative values.

    Thanks for helping us clarify this. I’m sure we will do a better job of describing the details in the future. We pride ourselves on being a transparent meritocratic measurement service so we’d like it to be drop dead simple to understand.

    Cheers,

    Jim

    • Hi Jim, I’m more than happy to engage in geeky talk.

      Yes I probably made that too simple. I kept it simple because majority of them don’t realise what your metrics are saying or what I’m saying. Yes I understand those views are based on people clicking on that link in the PostRank widget to visit a post on the blog.

      Blogs like Larry’s, Richards and quite a few of the other edubloggers I checked out, that use the PostRank widget on their blog, majority of the reader engagement is almost 100 % based on those views. If you look carefully at the metrics that are behind most of their PostRank scores they are mostly views.

      That isn’t telling them that they are doing a better job of engaging their readers and isn’t giving their readers a reliable indication of which posts are the most popular posts.

      As I said in this post and on my comments on Larry’s post I really love many of your PostRank features. It is an extremely valuable tool for bloggers for analyzing their own performance.

      But I don’t like the fact that edubloggers are misinterpreting the meaning of your statistics or the fact that bloggers can easily manipulate how high their blog appears on your topic ranking list just by adding your PostRank widget to their blog.

      • Hi Jim, based on Ken’s comment below, reading your comment again and looking at the stats on PostRank (again) I think have a clear understanding.

        If a blogger has a PostRank widget on their blog, it tracks the views on a post, this give the eye count view? So even if that post isn’t displayed in the PostRank widget its stats are being recorded? Can you confirm that this view of a post should be the same (or similar) to the view being recorded by Google Analytics?

        Can you also explain what the Click metric is?

        Also can you explain how come bloggers like Ken, have posts that rank 10 (http://www.postrank.com/feed/2e0528ccef64191a074d0d746ee7e094) but don’t get displayed in his PostRank Widget (http://newmiddle-earth.blogspot.com/)?

        • 1. “If a blogger has a PostRank widget on their blog, it tracks the views on a post, this give the eye count view?”

          Yes.

          2. “So even if that post isn’t displayed in the PostRank widget its stats are being recorded?”

          Yes.

          3. “Can you confirm that this view of a post should be the same (or similar) to the view being recorded by Google Analytics?”

          Yes.

          4. “Can you also explain what the Click metric is?”

          A click event is recorded when a reader clicks on a link in the top posts widget or when a user clicks on a link in Google reader or generates a similar event from one of our rss reader partners. On postrank.com a view event is related to the “eyes” icon the click is related to the “arrow” icon.

          5. “Can you explain how come bloggers like Ken, have posts that rank 10…but don’t get displayed in his Postrank Widget?”

          We use 2 kinds of Postrank scores: feed-based and thematic. They are both scores out of 10 and reflect the same underlying reader engagement. The two types are used for different purposes, feed-based compares a story with the typical engagement for the feed and thematic compares it with other stories possibly from other feeds.

          Story engagement is the weighted sum of the readers interactions: 5 tweets * 7 +4 comments * 10 = 75 engagement points.

          To calculate the feed-based postrank we normalize the story engagement based on the typical engagement for the feed at the time. If a typical story receives an engagement of 30, then a score of 75 means the postrank should be much higher than a typical 5.0 say a 7.8.

          If you start a blog your engagement is probably low. So an 8.0 might reflect say 50 points of engagement. As you build readers the average engagement will hopefully go up. Now to get that same 8.0 you need to get more engagement. An 8.0 on my blog is a different amount of engagement from a 8.0 on the Huffington Post.

          Feed-based Postrank is useful if you want the top 20% from a feed. Or want to compare your performance relative to your own posts. If you were a swimmer tracking your race times you wouldn’t compare race times from when you were 10 years old to your race times today. Feed-based postrank exists for the same reason.

          Contrast this with Thematic Postrank. In this case we compare engagement with an arbitrary set of stories, possibly from many feeds. Think search results across feeds, posts in a related topic or stories from a single feed over a long time period. In this case we don’t have a relevant historical context to normalize with.

          We determine the story engagement in exactly the same way but when we normalize the engagement to a 1-10 score we divide by the median story (a 5.0) then logarithmically scale the stories below to 1.0 and above to 10.

          Thematic is useful for comparing across contexts. We chose to use thematic in the top posts widget because we are comparing stories over a long span of time – say your last 50-100 posts.

          Hope this helps.

  4. Kia ora Sue!

    I’m always amazed at the assumptions that are made with things like PostRank. You introduced me to PR at a time when it was still called AideRSS.

    The eye icons you speak of, that represent views, will obviously not register a number if the Post Rank never shows on the widget. For instance, on my blog I have deliberately adjusted the widget so that only 5 post titles are displayed at any time. That means that even a reasonably high ranking Middle-earth post on PR may not get any contribution towards the number in the eye icon unless it ranks 5th or higher in the list in order to be displayed so that it’s clicked in the widget . This seems a bit daft to me for it’s a strange and artificial way of ranking the popularity of a post. I don’t intend adjusting the PR widget on my blog BTW.

    I got a response from PostRank’s Melanie Baker last year when I spoke of the disparities between displays, including the anomalous ranking that my Index Page acquired when it was polling a straight PR10. While the ranking was a possibility and it probably was getting that sort of attention, it was also an absolute nonsense since the post had never appeared in RSS listings, and never will, due to the way I introduced a backdating of its posting date to a time when I didn’t even have a blog.

    The disparity of the ranking as shown in the PR Firefox Extension can be -ve or +ve and can be out by as much as 2 units on the PR scale.

    Also, I know all about what contributes to the scale, but that doesn’t help me with what the scale relates to. I’ve never managed to work out just exactly what that scale (the PR 1 – 10 scale) is based on mathematically – whether it’s logarithmic or what, or if it is a comparative rating within the blog, or with other blogs – the jury is still out in my mind with this.

    Also, you have one at the moment that rates PR 9.2 and the next two are at 8.4 and 8.0. How do these compare with my top ratings of 8.1 and 8.0? Are they all measeured on the same scale? I wonder.

    The disparities of PR numbers are also quite anomalous. At the moment, for instance, I have three posts that show on the extension that they’re polling a straight 10 but only two of them reach 8 on the PR widget, which makes me suspicious.

    Catchya later
    from Middle-earth

    • Heya Ken,

      Done more research tonight (and now have a big headache).

      Based on your comment, Jim’s and analyzing Larry’s Google Analytics account my understanding is that “the eye statistic represents pageviews of a post which are only calculated if you add a PostRank Widget to your blog.”

      So PostRank is recording views of posts on your blog and using it as a Stats to include in calculating the post’s Rank a bit like how Google Analytics works except PostRank also includes other metrics.

      For example, with Larry’s post on “My verdict on twitter”:

      1. Google Analytics – 302 pageviews
      2. PostRank – 215 pageviews

      There is a slight different in counts between GA and PR but many of his other posts had minimal differences. As I said previously considering all the stats that PostRank are monitoring they do a reasonable task of giving fairly reliable statistics.

      Off course this means that I have to think some more (sigh).

      Take for example my post on The Edublogger “Check Out This Two Page Blog Guide For Parents!”

      This post has a current score of rank of 2.2 and most of the engagement making up that is 16 Delicious counts.

      But if you looks at Google Analytics it has 726 pageviews for the month and ranks 7 in Analytics for the month.

      Technically speaking, having the view stats in the ranking, is probably giving the post a more balanced metric. And where, off course, Melanie, Jim and Ilya clap their hands that Sue finally gets why they might have included it. Does that mean that I might add your widgets to my sidebars? Maybe — thinking about it.

      Yes I still have an issue of people not understanding what the stats means or that bloggers can manipulate their ranking. However that is not that much different from them not understanding most of the stats that bloggers can use.

      PS I believe I know the answer Ken, to the different score values between different dates. Based on the article that Melanie suggested I read my understanding is say you had a post with a 10 2 months ago — that post was ranked compared to past performance. Now 2 months later you are a better blogger, and you have improved your engagement to get a 10 now means your post has to preformed that bit better to achieve the 10.

  5. Thanks Jim.

    So just confirming to a blogger needs to have the PostRank widget on their blog if they want to tracks the views on a post and have this included in their stats? (Just making sure that I have that fact confirmed properly so I can change my information on Larry’s post).

    Now the feed based vs thematic.

    When you go to PostRank you are seeing your feedbased and the numbers calculated are based on typical engagement at that time.

    The PostRank widget use Thematic which is why your 10 scores in the feedbased don’t necessarily become included. It is also why why you can have one score on the feedbased and slightly different on the PostRank widget.

    But surely the thematic that the PostRank widget is using is your median story across all your blog posts that PostRank has been tracking? As it can’t be using a median based on lots of bloggers values since then you would see that some people scores were really low on their widgets.

    Now I’m assuming that the ranking of blogs by topic is based on thematic? http://www.postrank.com/topic/education?

    I know Melanie and I have discussed this but both my blogs theedublogger.edublogs.org and aquaculturepda.edublogs.org each are categorized with the topics blogging and education. She says you can have blogs showing up in more than one topic area. However http://theedublogger.edublogs.org/ is only on the blogging list while http://aquaculturepda.edublogs.org/ is on education. Have I done something wrong when adding the topic categories? As The Edublogger is as much education as blogging.

    • 1. “So just confirming to a blogger needs to have the PostRank widget on their blog if they want to tracks the views on a post and have this included in their stats?”

      Yes.

      2. “The PostRank widget use Thematic which is why your 10 scores in the feed-based don’t necessarily become included. It is also why why you can have one score on the feedbased and slightly different on the PostRank widget.”

      Correct.

      3. “Surely the thematic that the PostRank widget is using is your median story across all your blog posts that PostRank has been tracking? As it can’t be using a median based on lots of bloggers values since then you would see that some people scores were really low on their widgets.”

      We consider the last 50 stories in the feed for the top post widget. No stories from other feeds are included in any way.

      2. “I’m assuming that the ranking of blogs by topic is based on thematic? http://www.postrank.com/topic/education?”

      That could make sense but the answer currently is that Topics are feed-based so that you see the best from all the feeds in the topic. We aren’t really done/happy with the Topic content consumption model yet. Depending on the phase of the moon I thing thematic or feed-based make sense. Sorry if this sense inconsistent. :-/

      3. “http://theedublogger.edublogs.org/ is only on the blogging list while http://aquaculturepda.edublogs.org/ is on education. Have I done something wrong when adding the topic categories? As The Edublogger is as much education as blogging.”

      You are free to submit your blog to any topic that makes sense for it. If it fits in education and blogging then by all means file it under both. We fully expect for blogs to be classified in many topics. In fact this crowd sourced classification data is really handy information to have. Topics can be treated like tags in this way.

      Thanks so much for the opportunity to get all this information out there. Don’t let the questions stop! Our intent is to create a transparent, meritocratic analytics system that makes sense. I know we are not as clear as we need to be in this area, yet but we are working to get there. These exchanges are invaluable to that end.

      Cheers,

      Jim

      • Thanks Jim

        1. “We consider the last 50 stories in the feed for the top post widget. No stories from other feeds are included in any way”

        Does that mean you calculate the medium for the PostRank widget ranking based on the last 50 stories and this is then compared with all the posts in your entire feed that PostRank is referencing? To come up with the top 10 post that are displayed?

        So for example, you could have a post from 12 months ago appear provided your feed had been in PostRank for 12 months?

        2. Back to ranking of blogs by topic.

        When blogs are ranked against each other by topic. For example, education it is compared blogs against each other using their feed based ranking. You are obviously comparing the ranking of each blog against their rank the week before (and are showing if improved or decreased ranking).

        But I still remain confused how you might be comparing those relative ranks of the blogs to determine who is number 1.

        For example if you look at Education this week (http://www.postrank.com/topic/education)

        Livescience is No. 1. Is the fact that it is pushing a lot of posts out daily the main reason why it ranks in this spot? Similar for Inside High Ed which is N0. 3 spot. Once again pushing out higher volumes of post.

        Seems there is stats in the metric that are making this ranking less reliable.

        But laughing I see now The Edublogger has been included in Education and it has jumped up the list.

        3. Meritocratic analytics system that makes sense

        I do feel for you with that as I know how bloggers struggle with understanding what statistics means. Which is why I ask so many questions because part of my role is to get to a point where I understand it adequately to be able to explain to others.

        Meanwhile I shall go back to Larry’s post to explain that I was incorrect with the views data and to explain what it means (we hope).

        • 1. “Does that mean you calculate the medium for the PostRank widget ranking based on the last 50 stories and this is then compared with all the posts in your entire feed that PostRank is referencing? To come up with the top 10 post that are displayed?”

          We only consider the last 50 stories in a feed int he top posts widget. That lets older stories drop off after a while. 50 was chosen becasue that’s the typical total number of posts /year for small bloggers where 1 year is about right while more prolific bloggers usually want to showcase their best stuff. As a zero configuration widget 50 is good at satisfying abroad range of users.

          2. “You are obviously comparing the ranking of each blog against their rank the week before (and are showing if improved or decreased ranking). But I still remain confused how you might be comparing those relative ranks of the blogs to determine who is number 1.”

          I apologize for being confusing on this point. We do indeed compare week over week performance. We do that based on the unnormalized feed engagement scores. So in this sense the feed ranking is not feed based. Its just the raw underlying feed engagement so that would be more akin to thematic.

          3. “Livescience is No. 1. Is the fact that it is pushing a lot of posts out daily the main reason why it ranks in this spot? Similar for Inside High Ed which is N0. 3 spot. Once again pushing out higher volumes of post.”

          Yes when comparing raw feed engagement post velocity is a factor becasue it gives an audience more to engage with. In practical terms a publisher needs to have a sizable audience to match a high volume of posts so its usually reflects an authentically good blog. I could kick out 20 posts per day on my blog but that would stretch my audience pretty thin and as a result my engagement/post would go way down. There’s obviously some wiggle room in the middle ground. We’re hoping to build on what we have now to offer publishers more detailed analytics that show more of these nuances.

          4. “I do feel for you with that as I know how bloggers struggle with understanding what statistics means. Which is why I ask so many questions because part of my role is to get to a point where I understand it adequately to be able to explain to others.”

          Thanks!

        • Hi Jim,

          As Ken says I really appreciate the time both you and Melanie have spent trying to enlighten us.

          Personally I think we all need to go to a pub, sit down over a few drinks (LOTS), which will make it clear to all us of involved.

          1. Last 50 stories
          Totally understand and makes perfect sense.

          2. Ranking of blogs against each other
          Now I do need to make it clear that I don’t want, or need to know, Colonial Sanders secret recipe for Kentucky Fried Chicken but I would like to have a better conceptual understanding of it.

          Trying to read between the lines of what might be saying each week you monitor the stats of the blogs. To determine the ranking of a blog you would probably be adding together that blog’s story engagement for a week based on all metrics measure on all posts in the feed (or just the 50 posts?).

          For example it might be Monday – Sunday:
          Blog 1:
          2000 views x 0.001 + 93 links x 0.001 + 100 bookmarking x 10 + 30 tweets x 7 + 15 comments x 10 = Blog 1 Engagement

          Blog 2:
          300 bookmarking x10 + 60 tweets x 7 + 100 comments x10 = Blog 2 Engagement

          The blog with the higher engagement ranks higher.

          3. The statistics
          I know it is probably not possible, or practical, and perhaps not statistics that others would like. But if I look at the type of Widget I would like to have on my blogs it relate to the ability to customise the widget to show the stats I wanted.

          For example, I will love to show which posts had the most Delicious/Diigo saves because that tells my readers these posts are posts people like to refer back to.

          Same with having an ability to show posts with the highest comments – I have posts that I would want to showcase because of the amount of information shared in the comments.

          These are types of features a person like me would be willing to pay for especially if I was also able to download and easily analyze these statistics.

  6. Tēnā kōrua e Sue, kōrua e Jim!

    I’ve been silent but not unobservant.

    Wow! This is some complex, intricate stuff.

    I wonder about it’s intricacy, for it takes educated and experienced minds to understand what the whole thing means and how it works (I’m still trying to work it out).

    I’m not knocking it – I’m just observing here. I use PostRank and also support PostRank on my blog.

    Put it this way, Jim – if I were to write a post to explicitly unpack all this, it would require some considerable thought and crafting to write the post that others could understand what it’s all about (I speak only for myself here – Sue might have a more lucid idea of how it all comes together).

    If you catch my drift, there’s a lot that confuses the blogger who’s into this sort of thing. I wonder at the meaning it all has to those who are less enthusiastic about how it all works and just want to look at ‘ratings’.

    I thoroughly appreciate your time spent with us on this matter. I also thank Melanie Baker for her inputs on Sue’s and my own blog.

    Catchya later
    from Middle-earth

  7. Kia ora tātou!

    Okay. I’ve done my scientific experiment to prove that Jim is quite right – not that I didn’t believe him. I just wanted to confirm that it all works the way I thought it did.

    I published 8 posts (to date) in June. All except two (2) posts were published while the PR widget was displayed on my blog. I removed the widget after I’d published 3 posts.

    The third post had just started climbing in its rating when the widget was removed. It did not improve its rating until the widget was reinstated, when it shot to a PR rating of 10 within a few days.

    The two posts that followed were published during the time the widget was removed from my blog. Not only did they have no data recorded against them during that time, but they didn’t register anything after the widget was reinstated, and still remain at PR displayed rating of 1.

    The other five (5) posts shot to a PR displayed rating of 10 within a few days of the publishing date of each, all of which occurred smoothly either before the widget was removed or after it was reinstated.

    I know I’m the world’s greatest blog post writer :-), but to be able to write posts that give a rating pattern of 10, 10, 10, 1, 1, 10, 10, 10 to order must be verging on genius level.

    Not.

    I don’t think the observed results were coincidental.

    Catchya later

  8. Kia ora Sue!

    Pattern is now 10, 10, 10, 9.6, 10, 10, 10, 10. Whether this is as a result of retrospective catchup or subsequent accumulation, I’ve not got a clue. :-)

    Catchya later

  9. Pingback: PostRank’s Ten “Most Engaged” Education Blogs | Larry Ferlazzo's Websites of the Day...

  10. Pingback: Tips For Measuring Post Success: Part 1 Bookmarking | The Edublogger

  11. and plays, and win great prizes

  12. Pingback: What you wanted to KNOW about blogging! | The Edublogger

  13. Pingback: What you wanted to KNOW about blogging!

  14. Pingback: What you wanted to KNOW about blogging! | Education

Leave a Reply

Required fields are marked *.