2011 UK Census

Traditional output models have been based purely on a fixed output approach that collects certain variables into predefined packages. Continuing developments see a transition to incorporate a larger degree of user defined flexible output that lets users select and package the precise data that they want, in the way that they want, from given sets of variables often held in cubes.

Plans for the overall dissemination model for 2011, and any role that flexible output may play, are still in development. The degree and shape of user interest, and other crucial factors such as disclosure control, will play a critical role in the precise form of the planning. To help begin to gather information to inform this, the online output questionnaire sought to gauge a level of user interest in such developments by asking:

Question Thinking about a 2011 Census Web dissemination system, how useful would you find, and what level of priority (high, medium, low) would you give the facility to create user defined tables from given sets of variables?

Table 11.1: Usefulness of the facility to create user defined tables

Table 11.2: Priority of the facility to create user defined tables

94 per cent of users responding say that user defined flexible output will be either 'very useful' or 'of some use', and just over half the total think that this should be a high priority. There is little variation in views across the different groups of Census users, so clearly there is a positive, widespread and consistent opinion of, at the very least, the potential offered by flexible output.

But, as a user, are you willing to see some form of reduction or difference in the scope and depth of standard output in order to benefit from a user defined flexible table construction facility?

While it theory it may be desirable to serve the complete cross section of the user community by delivering results from the 2011 Census in all combinations and flavours, there may inevitably be some degree of compromise or trade off required in any balance struck between the degree of fixed core output vs. flexible user defined output.

The questionnaire asked two further questions designed to capture user views of any balance and compromises associated with a fixed vs flexible output model.

Question Some National Statistics Institutes are moving towards providing more flexibility in their Census (and other statistical) outputs by providing users with the facility to generate their own user defined tables and producing less pre-planned output. Which of the following would you prefer ?

Table 11.3: Predefined tables or flexible output

Almost two thirds of respondents said they would prefer to have a smaller set of pre-defined tables supplemented with a facility to create their own tables from data cubes. A quarter would prefer that the status quo is maintained, i.e. a set of 2011 pre-defined table’s equivalent in scope and quantity to those produced from the 2001 Census.

The question allowed for a write in response is users chose the 'other' option. Here, users took the opportunity to either express the view that they wanted both the wide set of predefined tables analogous to 2001 output AND a full functional flexible facility, or to emphasize the need to ensure that any flexible output functionality is as simple to use and easy to understand as possible.

One issue that the questionnaire did not raise, and one that can probably only be addressed through wider discussion, is the impact that a reduction in standard core output will have in making 2011 data as comparable as possible with that from 2001. While all required 2011 tables needed for full comparisons to be made may be derived using flexible functionality, this may not be immediately obvious to all users, and certainly may not be as quick and easy to implement at the user end. If any users have strong opinions in this area then we would like to here them.

Question In general, where would you prefer the balance to be struck between pre-planned tables and flexible user defined output?

Table 11.4: The balance between pre-planned tables and flexible user defined output

In general, when asked where the balance should be struck between pre-planned tables and flexible user outputs, the scales were tipped towards flexibility, with a sizable minority (40 per cent) expressing a preference in neither direction.

Table 11.5: The balance between pre-planned tables and flexible user defined output, by user group

User opinions show little variation when broken down by user group, with all sectors showing only a small deviation from the overall average balance rating of 2.65.

As we noted at the beginning, this is an area undergoing continuing developments. As each statistics organisations begins to disseminate data from the current era of censuses, we have and will continue to see, new functionality and features appear in each implementation.

If you who have already used one or more flexible user driven dissemination tools, we would encourage you to share and discuss your experiences of what is out there - sharing views of best and worst practice, perhaps with attention not only to the range of data and functionality included, but also how the system works in practice - how easy it is to understand, learn and use.

See other posts similar to this one: