HCI

I find that products are getting too 'futuristic' for the average user. This can cause difficulties in user testing. I need to put participants in the appropriate mindset. Do you know of methods being developed to ease the testing of advanced concepts?

What an interesting question!

Let's look at your assertion that "products are getting too 'futuristic' for the average user".

If this is true, then the products are not user centred in the first place.

What you are asking for here - "the appropriate mindset" - is training for people prior to using the products.

Training is fine if the user will be getting training in the real world. But if these are consumer products, the "mindset" will be fairly similar to the mindset your users already have. If users don't like it now (prior to "training"), why will they like it when the product gets to market?

The answer could be that you are designing products to be used 10 years out. By that time, more people will be familiar with more 'futuristic' products. So you want to get users who have that 10 years experience with technology.

I think what we will find is that technology gets simpler, rather than "futuristic". We will find dedicated products doing specific things (ubiquitous computing) as opposed to single products doing everything (current pc / multimedia technology). If my assertion is right, then we already have those people who are 10 years older: they are the people in 1996. Our vision, hearing, sense of touch, need for companionship etc etc etc - these won't change over the next 10 years. So we can design for those needs and capabilities now.

As a specific example, I built a 'futuristic' videoconferencing set-up at BT Laboratories called the Electronic Agora. This won't see the light of day until the next century, but all the design decisions were based on what people can do now, exploiting the social skills and conventions used in regular meetings to support virtual meetings.

I'd be interested in your comments on this and I hope it doesn't dent your enthusiasm for user testing which is precisely the way forward.
DT

Back to top

The Australian university for which I work is starting to look at teleworking. Are there any standards that I can use?

In the UK, employers must follow the UK Health & Safety Executive Guidance. This is written within their interpretation of the European Directive on Display Screen Equipment. The HSE guidance refers out to national and international standards (in particular, ISO 9241). Unfortunately, the standard doesn't refer very specifically to homeworking, but it treats it like any other office environment. But from the point of view of ergonomics, this is probably a good thing. The seminal text on Teleworking is Teleworking Explained Mike Gray, Noel Hodson and Gil Gordon. Published by John WIley and Sons, 1993. ISBN: 0 471 93975 7. However, it doesn't cover ergonomics in much depth.
DT

Back to top

I have been assigned to research the availability of ergonomics software. Specifically any data collection software (respiration, force, etc.) that is currently available. How do consultants do this? Do they collect information on the assembly line or is it done in a laboratory? Is there any anthropometric databases in software form that you are aware of?

Data collection is done at the site and generally does not involve a computer. Weights and pulling forces can be determined with a hand held fish scale device. Respiratory measures would be more complicated, requiring either a Douglas bag or wiring a person up with electrodes to measure their heart rate and try to determine respiratory rate from that. Consultants rarely have the opportunity to perform this, simply because it is rather intrusive to the individual. Other measures we commonly take are light levels with a light meter, or noise measurements with a noise meter.

As far as data analysis, there are some helpful packages. We often use 2D or 3D biomechanical analysis using software from University of Michigan (You may want to look at one of our back issues of our Internet site where we offer a version which estimates the disc compressive force in selected postures). U of M also have a Metabolic Energy Expenditure programme which estimates energy used by modelling the job and drawing upon a database of energy expenditures.

Another useful programme is one to calculate AWL from the NIOSH lifting equation. This can very easily be put together from the NIOSH formulas.

There is a very good anthropometric database in software form, Peoplesize, put out by Friendly systems Ltd, 443 Walton Lane, Barrow on Soar, Loughborough, Leics LE12 8JX (Tel: +44 (0) 1509 218 333). It has data on US, UK, Japanese and German populations of various ages. It is also very easy to use. Cost: about £150 or so.

For usability testing, we recommend The Noldus Observer. We are so impressed with this product that we have recently designed a course on usability testing around it!
CR

Back to top

Is there any information about symmetry of products and their usability? What is the connection between the symmetry of a product and the usability of the product?

This is a fascinating question, but to my knowledge not one that has been formally addressed. Our experience suggests that the three key components of a usable system are:

• Presentation of information / layout



• Consistency



• Task focus

Informal measurements with our clients suggests that the relative contributions of these to a usable system are 15% (presentation) 35% (consistency) and 60% (task focus). So if you see symmetry as mainly an issue of presentation, I would suggest at best it makes only a small contribution to usability (max 15%). However, there are times when symmetry has a task-related function (eg a steering wheel) and for those examples you would expect a more significant impact on usability.

You may also be interested in our article showing that a system's usability comes from task focus.
DT

Back to top

I am designing a magazine-style website. I am also looking into the hierarchical and navigational issues that surround web sites. Have you any suggestions as to where I can specifically look to get research on these topics?

Good sources of guidance include:

• The Yale Manual of Style



• Jakob Nielsen's website

There hasn't really been a lot of empirical work - people are feeling their way.

My suggestion would be to first define the content of the website, and then define a navigational structure to contain it. Remember to allow room for more growth than you expect. Finally, surf the web, find a visual design you like, and copy it as closely as you can without it being illegal.
DT

Back to top

We are looking at introducing a new identifier code for staff, students contractors etc. at Imperial College to replace a variety of different codes. We are unsure of the form this should take - purely numeric or an alphanumeric combination. A purely numeric seven or eight digit number would be most suited to our data needs, having the advantage of easier batch allocation, but people will need to remember and use the number, hence the consideration of some alphanumeric combination that may be easier to recall. Most people would have to recall the number quite infrequently, so we are unsure of their ability to learn it in the same way as a telephone number. We would start printing and encoding the number on all new ID cards, but all members of the college would not necessarily have that available immediately, and many may forget to carry their ID with them. Do you have any suggestions on a suitable choice in light of any human problems connected with issuing a long numeric identifier or could you point me to any relevant research papers?

The most memorable codes are those which have meaning or for which meaning can be contrived eg using rhymes or relating them to something which is already meaningful. The problem is that such meanings are usually individual. I assume you wish to allocate codes and not let people change them, so it will be difficult to make them meaningful.

If the allocation of codes to individuals is arbitrary, then simply combining alphas and numerics will not help a great deal. It will actually make entering the codes, eg using a keyboard, much more complex (as UK and Canadian Post Offices have found with their post codes).

One 'trick' which may work, is to link the numeric code to the way letters are shown on a telephone keypad eg '1' means 'a,b or c'. Users can then try to turn their number into a meaningful word. There are two smallish problems with this. One is that letter/number allocation on phones differs in the US and the UK and most people will need to see a labelled keyboard to remember the allocation.

I assume you have already considered using names! or even email names?
TS

People are not good at remembering numbers, and going beyond seven digits is likely to defeat even those people that can recall telephone numbers with ease!

Many years ago, BT commissioned an expert in human memory to address this very issue. At the time, BT had an alphanumeric system where the first three letters were the exchange (eg WES for Westminster) and the next four numbers defined the subscriber. BT wanted to know the impact of moving to an all-digit system. The expert told them that people would find the all-digit system much harder to remember.

As history has shown, BT went ahead with the change. There are two morals here:

• If people need a system enough they will struggle through;



• Alphanumeric combinations work better but only when the alpha-part is especially memorable (such as an abbreviation).

My recommendation would be to go through the following checklist:

• Do you really need a number? For example, although the system may need a number as an index to the database, could you put a user-friendly front end on it, so that the user could just give their name, or e-mail, as Tom suggests? (It's not clear if there are any security implications here).



• If you do need to use a computer-generated identifier, what is the very shortest number you could use and still future-proof the system? (Do you really need 7 or 8?). If you can get down to 4, you will be able to capture the memories of most of your users.



• If the identifier must be long, make it memorable (for example, concatenate the users initials and date of birth).

If you are serious about tuning this to the needs of end users, you may want to try out a few alternatives with just a small number of potential end users.
DT

Back to top

I have recently been working on the design of a keypad for a wall unit that incorporates many thermostat functions with the ability to create schedules (when lights, air conditioning equipment, etc.) go on and off. We had decided on two navigational keys to use on the keypad, an up arrow and a down arrow. We also used a plus and minus sign to change values (temperature, names, etc.) Now they want to change the keys - four navigational keys, up, down, left, and right arrows. The reason they want to do this is to accommodate for moving a highlighted character left to right in the display (for instance, you have an area named Classroom 1 and you want to change it to read Classroom 3; you would not want to change the first word and move to the number to change it. I guess I am confused on whether this is a good design. I would like to test both types of keys, but that means two different program interfaces. Do you have any suggestions?

You didn't mention who the users of this system are or how often they will use the system. From the context ("Classroom") I'm guessing that this is for use in a school or college and the users will be teachers or students. I'm also guessing that the usage will be intermittent (that is, the users will not be experts).

If these assumptions are correct, then you want to make the interface as simple as you possibly can: the minimum of buttons and the minimum of options. Allow users to change only the major things, such as temperature. Things such as "air flow" (which I've seen on other thermostatic controls) are only of interest to experts; remove these types of options or hide them in an area for expert users only.

To find out the key things users will want to do, you will have to interview likely users.

If my assumptions are wrong (and even if they are right) I can't urge you strongly enough to do some user trials. In order for you to change opinions within your company, you need to show how end-users feel about the two designs. Remember that there is always the possibility that users may want to change the labels often, and so you need to base your decision on data rather than supposition. I'm not sure that a testimonial from me supporting your analysis will help you (and it would be unprofessional of me to give one without understanding the context of use).

If you use paper mock-ups of the two alternative systems and devise realistic task descriptions of what users are likely to do, you could get all the answers in under a day. You don't need prototype systems to do these trials. It will also provide you with a bullet-proof defence of your final design.
DT

Back to top

GUI was a major breakthrough in HCI. What will be the next big breakthrough?

Good question but if I knew the answer I doubt I would be giving it away for free! But if pushed, I'd guess mixed media interfaces using combinations of hardware/software etc acting interchangeably and including gesture, voice etc.
TS