Designing Tests for the Power of Nine

Package for Re:source - Deliverable 3

Date: 21/03/2001


1. Background

In its Deliverable for the Library and Information Commission entitled The Power of Nine – A Preliminary Investigation into Navigation Strategies for the New Library with Special Reference to disabled People, Stringer and Carey proposed that the taxonomy of on-screen option arrays should not exceed nine on the basis of two complementary considerations:

  • A maximum of nine would make screen arrays compatible with the telephone numeric key pad
  • Nine conforms to the optimum options formula of 7 + or - 2.

2. Discussion

2.1 humanITy's first consideration was that the Power of Nine thesis should be scientifically tested for effectiveness against other search strategies:

  • Arrays of above and below 9 in symmetrical format: e.g. 4 or 16 in 2 x 2 and 4 x 4 arrays
  • Asymmetric arrays
  • Alphanumeric and chronological arrays
  • Graphics-representational arrays, e.g. maps, outline drawings.

2.2 humanITy also considered a variety of options for testing sites:

  • National Sports Database
  • NHS database
  • Arts and Humanities Data Service, King's College, University of London
  • Community Information Database.

2.3 A sports database was initially considered to be very attractive because taxonomy would simply be by subject or by geography, e.g.:

  • Geography: x Regions, no more than 9 (North-West, East Anglia, London) > x localities in a Region, no more than 9 (Liverpool, Ipswich, South-East) > x centres within a locality, no more than 9.
  • Sporting: Classification by typed > individual/professional - team professional - individual/amateur - team amateur > water - track and field - involving animals - pitch/field ball games

On balance, it was agreed that this did not present a rich enough challenge to the thesis but that it might be used later to test 'low level' operations.

  • NHS Database: humanITy has been holding extensive discussions with the NHS Information Authority and has discovered that current NHS databases suffer from the disadvantage that they have been primarily designed from the point of view of NHS personnel. Patients require a completely different perspective and intensive work is required in this area. It is not determined yet whether current clinician-oriented databases can be adapted for patient use or whether new databases will have to be constructed from scratch.
  • Arts and Humanities Database: This was the second most favoured option because it presented a variety of problems within a rich context. It allowed, in particular, one element to be classified in a wide variety of ways, demonstrating the rich possibilities of hypertext for overcoming 'tree structure' taxonomy. However, it was ultimately determined that the very complexity would make it difficult to carry out tests which could adequately isolate causal factors in performance.
  • Community Information Database: Community databases offer a wider variety of content than a sports database but there is not so much taxonomical complexity as there is in the case of a humanities and arts database. There is also increasing evidence that people logging on to the Internet want access to information about their own localities. A further factor was humanITy's connections with the Sussex Community Internet Project (SCIP) which seems to be an ideal local partner.

3. Conclusion

3.1 It was therefore determined, on the basis of Section 2 above, that humanITy should test out the Power of Nine hypothesis on a community database connected with SCIP.

3.2 In respect of the discussion in l. above, it was determined that there should be a comparison of the Power of Nine as an access strategy but that this should be compared with the current options available on the chosen site, rather than setting up comparative tests for different options.

4. Process

Because of the Deliverable contract and deadline constraints, some of the detailed work on Section 3 reporting discussions with community contacts has not yet been completed. This will be included in the 2001/2002 follow-up Deliverable to Re:source at no cost.