Visitors' Use of Computer Exhibits: Findings
from 5 Grueling Years of Watching Visitors Getting It Wrong
Ben Gammon
Introduction
The following points summarize what we at London's Science Museum have learned over the last five years testing computer interactive exhibits, from mostly bitter experience. This report is a status statement and will be expanded and modified as we conduct additional studies.
General Comments
Computer screen-based interactives can be robust and very effective exhibits. They are extremely popular with visitors, especially children, although not to the exclusion of other forms of exhibitory.
We have detected little or no evidence of techno-phobia among visitors. Indeed the usage of computers among our visitors - at work, home or school - is well above the national average. Almost three-quarters of the visitors questioned had at least some experience of using networked computers. Visitors are more comfortable with Qwerty than with alphabetical keyboards, again indicating familiarity with computers. We tend to find techno-fatigue rather than techno-phobia among some adult visitors.
These days museum visitors are not amazed simply by a touch-screen full of words. They expect something exciting, colorful, challenging, with graphics, sound, and movement. Thus, wherever possible use sound in the exhibit. Not only does this attract visitors but also it helps to indicate when a button has been activated, an answer is right or wrong, or a mistake has been made. Sound also can add humor to an exhibit.
There is no 'average' time that visitors spend at computer exhibits. Visitors will spend anywhere from 30 seconds to 40 minutes or more on computer exhibits. The time they spend is determined mainly by the quality of the content and screen layout, whether or not seating is provided, and how crowded and uncomfortable the gallery is. Providing seating can dramatically increase the amount of time someone will spend at an exhibit.
Restart buttons always cause problems. Visitors often mistake them as 'next' or 'previous page' buttons. If a person has spent 5 minutes getting into the activity and then accidentally gets returned to the start they are unlikely to try again. On other occasions other visitors (e.g. siblings waiting to have a turn on the exhibit) reach across and press the restart button to stop the other person from getting into the game or simply to see what will happen. Rather than a restart button, it is far better to have a time-out option where the computer returns to the screensaver if nobody touches the interface for a few minutes. There should always be a warning screen which appears before the computer returns to the start asking visitors whether they really want to finish the activity.
There tend to be three modes of behavior at computer exhibits:
- purposeful - people carefully and thoughtfully searching though the software looking for something specific;
- exploratory - people flicking through pages looking carefully to find out what is there;
- playing - people (usually children) rapidly moving through the different screens at random to see what happens.
Visitors often misinterpret the function of help buttons - believing that they will provide very specific help about a particular question in an activity or very general information about how to use the computer. Many people have had bad experiences with help buttons from word processing and other commercial software packages. When they see a help button they expect to find something equally useless. If there are vital instructions to be conveyed they must appear on other screens as well.
Text on Screen
As a rule we try to restrict the amount of text to around 30 to 60 words per screen. Additional text should be put on a following screen with a 'next' or 'more' button.
Avoid using scroll-bars as visitors often miss or ignore them and fail to see the hidden parts of the screen. If it all cannot fit on one screen then put it a following page.
We've found that visitors often will race past text or spoken instructions and go straight into the activity (this is especially true of children). Of course this means that they then cannot figure out what to do. Therefore, if there is a particular set of instructions that visitors must read or hear, design the interface so they have to stop and read or listen. For example, if you have spoken instructions don't allow visitors to move on to the next section until the instructions have finished. If instructions are written, don't activate the 'next' button for a minute or so, to 'force' visitors to search through the text to find out what to do next.
It is unlikely that visitors will read reams of text before starting an activity or selecting an answer. We've found that visitors are much more likely to read text while they are engaged in a task or after they have answered a question, to find out how well they did or why they got the answer wrong.
One trick is to present visitors with what appears to be an easy question, which they get wrong. There then is a powerful incentive to read the text to find out why. Another trick is to make explanatory text look like instructions. Visitors are slightly more likely to read instructions than explanatory text as there is a stronger imperative and a clearer pay-off for them.
We've also found that visitors ignore any text that appears in the top third of the screen and sometimes things that appear at the sides of the screen as well. This is found with both horizontal and angled screens. Therefore, important instructions must appear in the central part of the screen and ideally should clearly stand out from the background. Subtle changes of text on the screen probably will be missed so make changes blatant.
It is difficult, although not impossible, to get visitors to see external instructions or directions (i.e. label text) because they tend to focus immediately on the computer screen. The screen is much more interesting than any printed text around it.
Target Audiences for Your Exhibit
Children are attracted to any computer
exhibit. In the past exhibit developers naively have assumed that they
can target computer exhibits at an adult audience through the design and
content of the screens. Children still use these exhibits, seemly figuring
that if they press the screen enough times they will eventually find something
they can do. This means that the computers are still being used by children,
who gain little or nothing from the interaction. It is better to accept
that whatever you do, children will form part of your target audience,
and therefore layer the information/complexity of the activities so that
children can gain something from the exhibit while more interested or
knowledgeable adult visitors can delve deeper.
Computer exhibits often are used by small groups of visitors rather than
individuals--not surprising since only 18% of Science Museum visitors
are lone visitors. Thus, design the station so that three or so people
can group comfortably around the computer screen. There should be more
than one seat or a bench of some sorts.
Touch-Screens
Touch-screens still are the most common form of computer interface at the Science Museum and are likely to remain so for the foreseeable future. They have proved to be highly effective, robust and easy to use. In fact, visitors seem to assume that any monitor (or even surfaces on which an image is projected) is touch sensitive. This can be rather problematic because things may be changing on the screen as they touch it, leading visitors to assume that they are causing the change to occur.
Main problems with touch-screens are:
- visitors not being able to identify active areas,
- the active areas being too small to be easily activated, and
- the computer responding too slowly.
The active areas on a touch-screen should be brightly colored and clearly stand out from the background. But beware of screen design that uses red buttons on green or vice versa; 8% of the male population and 0.1% of the female population are red-green colorblind and may not be able to see the difference. Smaller numbers of people are blue-yellow colorblind.
Since visitors often assume that any picture, or anything that moves, is an active area, it is best to make these the active areas. Unfortunately, visitors assume that touch-screens are pressure sensitive and when they touch an inactive area of the screen and nothing happens they simply conclude that they haven't hit it hard or often enough.
The active areas should be at least 1 cm
square. A mistake I've seen a few times is where software designed for
a keyboard and mouse interface is accessed via a touch-screen. This is
always a disaster (and I mean always) since the active areas are far too
small for even a child's finger, although they are fine with a mouse and
cursor. If you are using Web-Sites or other software design for conventional
interfaces you will need a track-ball or mouse.
Unlike conventional buttons, touch-screens provide no tactile feedback
so visitors often fail to realize when they have activated the computer.
The touch-screen should, therefore be set on 'mouse-down,' i.e. active
as soon as touched, and not require the user to remove their finger from
the screen. The computer should respond to touch within one second. Failure
to do this will result in visitors ramming their fingers with increasing
force into the screen assuming that the touch-screen is pressure-sensitive
and that the button has not activated because they have not pressed it
hard enough. When touched buttons appear to depress, a color change and
clearly audible sound are important.
Overlapping active areas on adjacent screens causes another common problem. Visitors can accidentally double touch an active area, inadvertently activating the button on the next screen, stacking commands and so jumping on to the third screen. We ensure that once a button has been activated the screen becomes inactive for a few seconds. This gives the computer time to react and means that even if the visitor does touch the screen again by mistake they don't shoot pass this next screen.
Buttons need to be clearly and unambiguously
labeled so that visitors can work out what they are suppose to do. This
requires careful formative evaluation. Visitors assume that the function
of any button is to start an action or sequence of actions; i.e. they
press it and expect something then to happen automatically. Problems arise
when the active areas actually are tools that the visitor must select
and then move the cursor/tool over the screen e.g. to paint or color an
area. Visitors are often observed to click on the tool and then wait …
expecting something to happen.
There are several problems with touch-screens. Only only one person can
use them at a time. Touching the screen in more than one place at the
same time results in no computer response or a button that lies in between
the two points of contact is activated. This means that if two people
try to use the touch screen simultaneously they will have serious problems
getting the computer to respond properly.
Touch screens are not very good for 'click and drag' functions. Finally,
very young children (under 7 years) have problems operating touch-screens
when the active areas are small and close together. Young children have
considerable difficulty touching the screen with just one finger. Often
they use their whole hand. Therefore touch-screens designed specifically
for young children must have active areas that are large (3-4 cm sq.)
and well spaced.
Other Forms of Interface
We now are looking at using forms of interface other than touch-screens. In addition, there are an increasing number of computer based exhibits that do not even use a screen to display the output. So far we have only evaluated some of these but have some initial thoughts. Most of the principles of good touch-screen design apply to other interfaces:
- the computer should respond to an input within 1 second,
- active areas on screen should be large and prominent,
- buttons should be clearly and unambiguously labeled, and
- non-tactile buttons need to respond on mouse-down.
Track-Ball and Push-Buttons
These seem to be easy for visitors to use and are reasonably robust. They also allow using smaller active areas.
They can present maintenance problems with dirt getting trapped under and around the ball. The trackball and button should not be the same color as the surface they are mounted on. While it may look slick, it makes the ball difficult to find. Also, many trackballs come in the same color as the control unit they are mounted in, but it is much easier for visitors to locate them if the colors of the ball and control unit contrast. Black-on-black trackball/control units are especially well camouflaged, as computers are often located in indirectly lit spaces to prevent screen glare.
The movement of the track-ball should match the movement of the cursor on the screen i.e. moving the ball forward should move the cursor up; moving it left should move the cursor leftward and so on. Incredibly some track-balls work the other way around--basically these are computer mice that have been turned upside down. And do they confuse visitors--YES THEY DO!
The cursor needs to be prominent and neither get lost in the background screen design nor disappear off the edge of the screen.
A track-ball control is not the same as a computer mouse. Track-balls are not good interfaces for click and drag functions where the user has to hold the button down while they move the ball. Due to the design of most tracker-balls this proves to be a very awkward maneuver. Also moving the ball is not the same as moving the whole mouse - there is not the same mapping of the action onto the outcome. Ideally the track-ball simply moves the cursor to select active areas or applies a tool to an area of the screen.
Light Beams
We've found these to work well provided visitors can clearly see where to break the light beam and can see the image of the beam on their hands. This interface allows multiple users at the same time although this then causes serious problems where visitors don't know who is causing what to happen. So perhaps it is not actually a good idea after all.
Infrared Beams
These are disastrous because visitors cannot see where the beams are, nor can they see when they have broken the beams. This means that they fail to active the computer when they are supposed to and activate it by mistake when they shouldn't.
Touch Sensors Fixed to the Inside of Glass Display Cases
These seem to work surprisingly well as long as there is a clear physical link between the sensors and the computer screen (e.g. a big cable) so visitors realize what these are and to which exhibit they are linked.
The sensors need to be set so that they
are continuously 'polled' to see if someone is touching them. Visitors
often place their hands over the sensor and leave it there until something
happens on the screen. The problem is that the sensor may be inactive
when the visitor first places their hand over it. If the sensor is set
to look for a change then it will not respond until the visitor removes
their hand and replaces it again.
The sensor technology must be robust as visitors often press the sensors
repeatedly and forcefully. The sensors should be set to activate on mouse-down.
Touch sensors should be labeled so that visual reference can be made to
them on the screen. Sensors should be grouped close together so that they
are in-line-of-sight for the user and can be easily reached by children
or wheel-chair users. Ideally the array of sensors should cover an area
of no more than about 1.5 meters.
Sensors should be within the visual field
of someone looking at the screen while standing within easy reach of the
sensors.
It is vital that there is high quality sound out-put so that visitors
can follow the instructions.
Keyboards
We have found that both Qwerty and alphabetical keyboards work well with visitors. Qwerty keyboards are easier for visitors who are familiar with computers (by now most visitors) while alphabetical keyboards work better with young children. The difference is slight.
Mechanical keyboards (as opposed to keyboards
on touch-screens) have robustness problems and need to be reinforced and
protected from liquid spills.
We have tried a wide variety of keyboard designs. It is best to find a
design that can be easily customized so that redundant keys can be removed
and special keys added. Keyboards that do not provide any tactile feedback
- i.e. where the keys do not move - need to generate a sound when each
key is pressed. These keyboards should be on mouse-down setting. The keyboard
also needs to have arrow buttons (or similar functions) so that visitors
can move the cursor within the text and delete mistakes they have previously
made. Visitors find it very frustrating when they have to delete whole
lines of text just to remove one small typo in the first line.
Electronic Pens
Electronic pens with moving parts have proved not to be robust enough to survive even the briefest of exposures to visitors. Electronic pens with no moving parts are much more reliable. Electronic pen systems that involve writing directly on the screen have proved to be much more effective than systems that involve writing on a separate tablet.
Visitors can sometimes mistake electronic pen systems as some form of children's drawing activity rather than a serious exhibit aimed at adults and children. The quality of people's handwriting when using the electronic pen can also lead adults to assume that only children have previously used the exhibit.
Fingerprint Scanners
These can be highly effective methods of logging on to a network of computer exhibits so that each exhibit can recognize the visitor and store information about them. However, specific problems have been identified as listed below.
Children often forget which finger they
have used previously, or do not realize that they need to use the same
finger each time. The also tend to press just the tip of their finger
onto the scanner rather than laying their hand flat and placing the whole
finger tip across the sensor. The sensor itself can have problems logging
on and/or recognizing small fingerprints from young children.
One approach is to use an outline handprint with the sensor at the tip
of the first finger. This acts as an effective prompt to visitors to position
their hand correctly and to use the same finger each time. The handprint
should be child-sized, as children tend to align their hands with the
bottom edge of the handprint rather than the fingertips.
Showing visitors the image of their fingerprint from the scanner on the computer screen is enormously helpful. Visitors can then see how to adjust the position of their finger to get a better image and can judge when they are pressing on the scanner too lightly or too hard.
Visitors often assume that they need to keep their finger on the scanner throughout their interaction with the exhibit and need to be told when to remove it.
Video Cameras
Visitors often find it difficult to position
themselves in front of a video camera because what is shown on the screen
is not a mirror image. Visitors often move in the wrong direction or move
too far one way or the other.
One possible solution is to present visitors with a mirror image rather
than a conventional video image.
Another problem we have encountered is
when visitors are asked to position themselves in front of a video camera
and then have to lean forward to press a button to capture their image.
This means that they are not in the correct position when the camera takes
their picture.
One solution is to design the exhibit to be more like a commercial photo-booth
where visitors press a button which starts a countdown at the end of which
the camera takes their picture. This means that visitors have a few seconds
to position themselves and can be stationary when their image is captured.
Visitors do weird things
Visitors do odd things that they would
not do outside a museum setting.
For example, we once developed a series of linked computer exhibits which
included a videophone. Visitors were supposed to play a game and when
someone arrived at another terminal phone them and talk. The telephone
was made to look and operate like a conventional phone because, we reasoned,
people would know how to use it. However, we forgot that the meaning visitors
attach to a phone in a museum setting is very different to the meaning
they attach to a phone in a work or home environment. What actually happened
was that as soon as a visitor approached a terminal they picked up the
telephone handset and jammed it under their chin. Even though they could
clearly see the loud-speaker in front of them and surely can hear that
the audio output is coming from the speaker and not the telephone, they
still held on to the handset. What this meant was that the terminal is
engaged all the time and no one at another terminal can phone them.
The underlying reason seems to be that when a visitor sees a telephone
handset in a museum environment they assume that by picking it up they
will hear a commentary. Obviously they would not go into someone's office
or home and pick-up the telephone but then they would be in a different
environment with a different set of rules and expectations.
Although we discussed this exhibit at great length we never tested a proper prototype version--which just shows the value of evaluation.
Acknowledgments
A big thank you for their hard work and dedication to all the people who have contributed to the findings outlined in this document: Eleanor Bridgman, Peter Bailes, Emma Birch, Lindsay Cook, Jane Croucher, Karen Davies, Jo Graham, Keith Greaves, Catherine Halcrow, Yvonne Harris, Barbara Keating, Rachel Kingston, Ros Mist, Theano Moussouri, Amanda Parkes, Dave Patten, Carmine Ruggiero and Nick Smith.
Ben Gammon is Head of Visitor Research at the Science Museum, London.