statistics

Tonight on twitter I caught the buzz about someone losing their google pagerank, due to having had normal links (ie not no follow) in sponsored posts. Hard on the heels of that discussion there was a groundswell of questions. What is pagerank? Why does it matter? What’s a nofollow link anyway*? And so on. And I went on to have an interesting chat about where traffic for blogs come from anyway.

You see, I’m not that convinced that Google and its pagerank are all that important for blogs. Sure, if you want to sell advertising, most companies are probably going to check what your pagerank is, but if you aren’t, why would you be bothered?

Would you be worried about losing your google derived traffic? Does it deliver you good, interested, dedicated blog followers?

It’s my suspicion that most traffic to blogs from search engines falls into two categories. There are the ppl who arrive at a single clearly targeted post – such as for example my camping list. For a while that ranked top of google’s search for a camping list, I think it’s still on the first page. And I’m sure for most ppl it’s quite a useful resource. However, if they are looking for camping resources I doubt the rest of the somewhat eclectic ramblings here interest them, and I assume they are in the portion of traffic which bounces away again.

The rest of the search related traffic is probably slightly surprised at where they find themselves. For whatever reason a post here is thrown up as a good match for what they are looking for, when it manifestly isn’t, and they probably bounce immediately.

Bounce rate is another thing we’re told to worry about. But actually, for a blog, a high bounce rate isn’t so bad. After all, you hope most ppl are passing by regularly, so there should only be one new post for them to read, they read it and then bounce away. You’re looking for a reasonable percentage of recurring visitors, and ppl who’ve arrived from a search engine probably aren’t going to join in with them.

If you’re looking at your stats, there are all sorts of interesting things in there. You can find out what percentage of visitors are new, or which come back. You can find out how long they tend to stay on site, where they come from (around the world), where they came from (referring site). My top referring site is twitter – measure of how much time I spend on there I suspect. But I also get a good number of visitors from posts I write on other sites, such as Emma’s Diary blog. (What, you haven’t read it? But I’m scintillating on there. Rivetting. Pop over and check 🙂 )

And there are other useful places to pick up traffic. Stumbleupon. Pinterest. Facebook. Social networks with user generated links, entirely possibly delivering better quality traffic thaqn search engines? Ever wondered why google is putting so much effort into g+? It’s because they suspect social is the future of the web, and in that world, how is pagerank going to be relevant at all?

*It’s a link that has rel=”nofollow” in it, telling search engines they shouldn’t give weighting to it.

I am writing to express my concern at the statistics used to inform Graham Badman’s review of home education, released 11 June 2009.

The review can be found here as a free download.

and the particularly disturbing section, on which the vast majority of the call to action and most upsetting recommendations are based is the section about child abuse, which was the primary reason for the review in the first place.

Thus:

First, on the basis of local authority evidence and case studies presented, even acknowledging the variation between authorities, the number of children known to children’s social care in some local authorities is disproportionately high relative to the size of their home educating population.

This statement has then been reported in the press as

Children educated at home are twice as likely to be on social services registers for being at risk of abuse as the rest of the population, the head of a government inquiry into home education said yesterday.

and

Graham Badman, the former director of Kent County Council’s children’s services, headed the review. He said the ratio of home-educated children who were known to social services was “approximately double” that of the population at large.

Freedom of Information requests have been made to find the basis of these figures, and resulted in the release of an Annex (pdf):

The statistics within this annex are at the very least problematic, and honestly, they are downright misleading.

First, only 25 of 90 authorities asked responded. It is reasonable to assume that it is likely to be the authorities that perceive a problem that will respond, so the immediate response should be called into question. Why were only 90 of 150 authorities asked in the first place?

We are not told which authorities responded, which seems unreasonable as we are therefore unable to do any research to prove or disprove theories based on their figures.

Next, the median is used as opposed to the mean. There is rarely any good statistical reason to use the median figure, the mean is the average that most people expect to be employed.

Thirdly these figures are then extrapolated to give figures across all local authorities, I do not know of any basis in statistical working to extrapolate from a median of such a small and self selecting sample to give authoratative figures. Even if this was reasonable, we are told that a median of 7 extrapolated across 150 authorities gives 1350 EHE children known to social care – by my calculations it should be 7 * 150 which is 1050.

We are then told that this figure is 6.75%, but as we have been told repeatedly in the report that the known population of home educators is about 20,000 while the actual figure could be as high as 80,000 this is utterly meaningless.

*If* we used 1050 over 80,000 we would come up with a percentage of 1.31 % which is so small as to suggest perhaps the government should instead be shutting all schools down and recommending home education as the best way to safeguard children – surely as reasonable a conclusion as the one that

they have drawn?

I await with interest your response to this complaint – can you advise what you are able to do about incorrect use of statistics in government reports?

Thank you

ps please note, I originally sent this to authority.enquiries@statistics.gov.uk as listed on this page:

of your website, and my email bounced. I think you need to check the email addresses you are giving out.

I have now resent to authority.enquiries@statistics.gsi.gov.uk.

To readers – please feel free to also write and complain 🙂

statistics

Footer