Training: searching statistics on ons.gov.uk

Image

5 February 2012, CILIP HQ (organised by CILIP Information Services Group)

Notes on the day

Geoff Davies, Implementation Manager at the ONS, gave a run-through of the navigation of the newly redesigned ons.gov.uk. Recent improvements include new search functionality, additional synonyms and acronyms and better navigation.

  • Several new elements on the homepage will be useful for headline figures – the “carousel” in the centre which announces the latest big releases, and the Key figures panel on the right which is a quick way of accessing the most up-to-date stats for GDP, unemployment etc.
  • The UK Publication Hub (link at bottom of landing page) holds all government data, not just that held by ONS.
  • ONS YouTube videos give explanations of big releases, and the new interactives are a good way of interrogating data.
  • Links to the previous site are obsolete, so if you’ve saved a URL it won’t redirect to the new site, but all the statistical releases have been carried over, so they will be there if you dig deep enough.

Geoff then outlined the basic structure of the ONS site, which is a simple nested hierarchy:

  • Business area (section) folder -> each publication has a folder -> calendar entry for each edition -> edition folder -> all content “nuggets” released on that date eg. charts, data tables, summary, statistical bulletin etc.
  • Every edition published to the site has a separate release page, which goes live on the publication date (the release calendar includes future publications). Everything relating to that release is accessible from the page – datasets and reference tables are listed at the bottom of the page, and contact details for a named person responsible for that release are to the right.
  • The redesigned theme pages, which are launching shortly and will be rolled out gradually across each theme, are simplified and easier to understand, and much more visual than the current text-based version. A moving carousel, in the centre, gives the most recent data. They are a work in progress and will be improved as more pages are updated.

Geoff gave a quick run-through of the navigation tabs across the top of the site:

  • Browse by theme – alphabetical index of themes -> individual theme pages, with the most relevant or important content at the top.
  • Publications – chronological list, with filters on the right to narrow down content.
  • Data – chronological list, search for datasets and reference tables here (not available in publications list).
  • Release calendar – all releases, chronologically, including future releases (the landing page only includes big releases). If you click through to a release page there’s a link to all editions at top right, to access previous data.
  • Guidance and methodology – gives background on the ONS and data collection, classifications etc.
  • Media Centre – includes official statements and releases, and letters correcting misinterpretations of stats in the media.
  • About ONS – most useful is the ad hoc research undertaken by ONS, which isn’t searchable in the publications indexes. Go to Publication Scheme under What We Do, then Published Ad Hoc Data on the left.

Continuing problems with the site

The main issue users have raised since the redesign is difficulty in finding content. The ONS has decentralised publishing, which means each department is responsible for their own releases (around 460 staff contributing to the site). This has led to inconsistency, as some staff are reluctant to change old methods or not interested in web standards, and some are just too busy. The ONS are working on solutions:

  • training staff on how to tag content with six or seven most useful keywords (too few, or too many irrelevant ones, mean weaker search results), and improving the metadata.
  • publishing support team to help departments who are too busy or uninterested.
  • health checks are run on content regularly.
  • there is pressure from management to conform to the new standards.

Practical examples

We ran through some real search queries for tips on searching the site, with assistance from a member of the customer services team (whose name I missed, sorry!). The main advice was to search through the release calendar using filters as necessary (selecting ‘last 5 years’ clears future releases from the list), and to use the ‘all editions’ link on each release page to locate time series data.

Unfortunately, the practical examples just proved that the search functionality of the site still needs improvement (if a roomful of information professionals struggles to find data you have a problem!). Advising users to call the customer services team with any queries is helpful but no use in a high pressure environment where data is needed within hours, not days – what I really needed were ways of finding the stats myself.

Reflections

  • The redesigned ons.gov.uk site is much cleaner and simpler than the old version, and easier to navigate, but it’s still difficult to actually find specific data. It’s a shame the ONS didn’t take advantage of having a room full of information professionals to interrogate the system further and to make notes of improvements needed.
  • Some of the problems the ONS are facing are familiar – they’ve decentralised uploading of content, but some staff are reluctant to adopt new techniques and others are over-keen and tag excessively. This is true of other new technologies being adopted across many library sectors (certainly it applies to social media in the news industry). It’s an issue of good training and perseverance with the new standards, and having support from management is vital.
  • Some issues with the redesign are similar to those we’ve experienced in relaunching our intranet recently – lack of redirects from old pages, decentralising, need for training.

Applying what I learned

  • The key figures and carousel on the front page of ons.gov.uk will be incredibly useful for finding the most recent headline data quickly (a common query).
  • The new theme pages will be very useful once they are launched, as a quick way to access key figures on a topic (another common query).
  • I’ll bookmark the ad hoc data page as an extra location to check for data.
  • The training also offered some good ideas on how to ensure consistently good content and metadata, which we could apply to any new roles that our department undertakes.

Umbrella conference: Ian Rowlands, The Google Generation

Intentionally or not, Ian Rowlands built on Nigel Ford’s talk that preceded him, reinforcing the point that researchers and practitioners need to work more closely together.

The original CIBER research that Rowlands and colleagues launched at UCL consisted purely of deep log analysis (DLA), studying comprehensive data showing how people use the web. But although the research was rigorous and reliable (reflecting what the participants did, not what they remembered selectively), it gave no context – researchers could determine what web users did but not why.

Now, CIBER is combining the ‘science’ of research with the ‘art’ of practice by surveying web users to find out about their behaviour, and what the DLA tells them. The hope is that by combining log analysis data on users’ search sessions with questionnaires on how users see their web behaviour, researchers can begin to make value judgments on how we use the web, and help to reshape info systems.

Ian Rowlands was my tutor at City when I did my Masters (not that he’d remember me!), before the team moved to UCL, so some of the content was familiar. It’ll be interesting to see the results of the behaviour survey and whether information providers need to change the way systems are navigated in the wake of Google.

I think as practitioners we sometimes like to think we know best, but what makes sense to a trained researcher isn’t always clear to everyone! I’ve definitely been guilty of that in the past. Meeting user needs is surely key for info pros, even if their approach seems counter-intuitive to us. And if there really is a gap in research knowledge now Google is dominant we are surely best placed to train users.

Notes on the presentation:

DLA tells us how we search the web:

  • horizontal web seeking – tend to skim across info, don’t go deep – stickiness is an issue
  • navigating – users spend more time looking around a site than they do looking at the content
  • power browsing – rapid clicking, impatient, very short dwell times

Rowlands says there is a “sense of ‘moral panic’ – that young people are not as disciplined as we were, that we’re losing the sense of the library as a place of quiet contemplation.” Nicholas Carr posed the question, “Is Google making us stupid?” – is it changing the way we think? Rowlands says there is some evidence to suggest that “our brains are changing with our emersion in the digital space”.

If the evidence shows there is a Google Generation – defined as those born after 1993 who have grown up with Google – who experience digital libraries differently to previous generations, then libraries need to adopt different learning strategies now that they are reaching university.

  • don’t read the manual first – learn by having a go
  • not good at deep research, evaluating info once they’ve found it
  • not good at using other technology (misperception that they’re totally tech-savvy)

However, a British Library study showed that there is no significant difference between the generations, and that we have all changed in the wake of Google-type web searching:

  • the apparent facility of technology is not true, and is very basic
  • research characterised by speed and lack of evaulation
  • prefer simple solutions, routes eg Google, Yahoo
  • don’t find library-sponsored resources intuitive
  • don’t understand their information needs

Web behaviour surveys

CIBER paired with BBC Lab UK to conduct a web behaviour test, which is still ongoing. Stanford are also looking at the psychology side.The survey looks at two questions:

  1. Testing Google ‘first-click’ behaviour – habit of selecting first result, not delving deeper, lazy searching – trying to see how influential rank is vs the info content – are they making an informed choice, evaluating the info? Is there a perceived authority in the info provider if it’s a trusted brand? Or are they just trusting the algorithm of Google? Users are provided with results to health questions in a random order, stripped of the links and references; and then with the reference given, to see if behaviour changes.
  2. Are there demographic and age factors linked to searching – terms used, time spent on task, confidence in results? Do generations negotiate the web differently? Do we all “walk individually or in groups through digital space”? And if methods are different depending on your demographic, “could we shape an info system to reflect this”?

Results are expected next year.

Umbrella conference: Prof Nigel Ford: Technology, personalisation & librarians

I was late arriving for the first session of the day (sorry) so I missed the start of Nigel Ford’s presentation. His discussion of pathways (as applied to europeana.eu) made sense though – surely in an ideal world what every system needs is an infinite number of routes through the information, to suit “an infinite array of users”?

Providing the most obvious, commonly used ones should meet the needs of most customers, but by enabling them to make their own paths as well you don’t exclude those who want to search off the beaten track, have the knowledge to run complex searches or whose search methods vary from the norm.

Ian Rowlands’ analogy of the college that built the paths eighteen months after the buildings, so that they followed the routes taken by users and not the other way around, really rang true (I think I’ve heard it before, but then he was my tutor at City so maybe I have!).

Ford’s rallying call for researchers and practitioners to work more closely together rang true too. Research doesn’t really influence the way I work at all, change rarely happens and I’m sure there are better ways of doing what I do. But reading research takes time and, as Ford said, practitioners tend to focus on the bottom line.

The ‘art versus science’ analogy (and I missed the author of the study) is a good one – researchers are all about robustness, vigorous data and models that don’t necessarily apply to the real world, while practitioners are good at real world usefulness but without basing it on solid research.

As Ford said, we “need both elements of our knowledge…more interaction to achieve critical mass”, or “progress will be difficult”.

Online: Tom Roper on David Nicholas’s address to the Cilip AGM

I was too slow to grab a ticket for the Cilip AGM but luckily Tom Roper has blogged his views on the talk of the evening, David Nicholas (an old lecturer of mine at City university before he defected) on the Google generation.

As Tom summarises:

Librarians don’t bang our drum enough: as information and its retrieval becomes more and more complex, we can help, if we can show how we can help people and organisations achieve the outcomes they want… We cannot count on loyalty anymore.

I’ve definitely noticed the effect Google has had on search practices at work. A few years ago, the first thing a trainee was taught was Boolean searching; we realised a few weeks ago, to our embarrassment, that our current trainee (who has been with us since September) didn’t know the term.

Journalists used to whacking words into Google and receiving an immediate answer are much less patient when deep searching is needed, and much less receptive to learning advanced search techniques. They’re also much more likely to favour non-authoritative sources that offer quick answers than taking the time to find reliable sources (Wikipedia is increasingly cited as a source in its own right).

I don’t think David Nicholas is entirely right (Tom makes a good point that a lot of content is disappearing behind paywalls, making it harder not easier to locate) but librarians do need to develop new strategies to enable the Google generation to find the information they need in an over-saturated web, rather than attempting to train users in old, and sometimes out-of-date, searching styles.