Abstract
BACKGROUND:
Lever-operated taps have become more popular and are commonly used in operating theatres, food preparation areas and where users have poor strength; however, there is very little data available for user expectations on tap operation. Thus, an experiment on dual lever-operated water tap (faucets) was conducted with the aim of for providing information for improved design.
OBJECTIVE:
This study aims to compare different lever-tap designs and their stereotypes adopted by the end-user to operate them also to verify the stereotypes for increasing or decreasing the water flow.
METHODS:
240 participants were requested to rotate the lever tap to indicate direction for increasing and decreasing water flow with simulated hardware, using actual taps placed at the top of a simulated washbasin. Nine initial positions of the lever were used for increasing and decreasing flows, ranging from the ends of both levers facing outward from the bowl center to the ends of both levers facing inward. All levers operated in the horizontal plane.
RESULTS:
Strong stereotypes (greater than 80%) for several initial lever orientations were found for increasing water flow, especially when the initial lever end positions were facing outwards. However, for different initial positions at which participants were told that the water was flowing and the flow was to be decreased, no strong stereotypes existed.
CONCLUSIONS:
The stereotypes for increasing water flow of dual-lever taps were strong, whereas those for decreasing water flow were weak and hence the stereotype reversibility was also weak. In terms of user expectations, lever taps do not show any great advantage over cross-taps in terms of operator expectations for increasing and decreasing water flow.
Nomenclature
Anticlockwise
clockwise
dual taps both levers rotated anticlockwise
greaterthan dual taps, left lever rotated clockwise; right anticlockwise
Index of Reversibility
Angular location of left lever, measured from 12 o’clock as reference position
Angular location of right lever, measured from 12 o’clock as reference position
Angles are measured relative to 12 o’clock (zero degree), with positive rotation measured clockwise from the 12 o’clock position.
Introduction
In every home, bathroom and many public facilities, there are taps controlling the supply of water. Some types of taps have been studied in detail. For example, Campos and Paschoarelli [1] reported a study of preferences for different geometric forms of taps, including standard cross-tap design, lever-operated taps and others with differently shaped control knobs that allowed improved grip on the control. In the Campos and Paschoarelli tests, the lever-operated tap was preferred by their group of participants in terms of ease of control.
Ease of control is, however, only one of the possible criteria for tap design. Other criteria may be related to space requirements, purpose (such as where free-hands operation is required), plumbing standards in the country of use and use by special population groups, such as persons with some form of disability. Boger et al. [2] studied the effect of familiarity with the form of water tap on the ability of persons with varying levels of dementia to correctly use the tap. They reported only the ability to turn the flow on, and not that for turning off the flow. In the Boger et al. experiments, of the five types of tap tested (crosshead, dual lever, single lever, electronic, and plastic wand), the dual-lever tap was found to bemost usable.
Another important aspect of design is the expectation or stereotype that people have for operation of the tap. Cross-taps have been studied by Evans and Kwok [3], Hoffmann et al. [4] and Hoffmann and Whitfield [5]. Also important is the reversibility of the stereotype. The stereotype or expectation for decreasing the water flow should be opposite to that for increasing the water flow. A good design would have 100% of users rotating the control for decreasing the flow in the opposite sense to that for increasing the flow.
The correct direction of movement of levers for an increase or decrease of water flow is dependent on the plumbing used in the installation. In practice, it is found that there is a mixture of valve arrangements that have been used by installers so that there is no established expectation within the user population. For example, Fig. 1 illustrates a counter with two basins, each having hot and cold water taps. In this Figure, the tap levers are shown in the ‘on’ and ‘off’ positions. For the right side basin, the tap levers have to be turned clockwise in order to turn on hot water. The left side basin requires an anticlockwise rotation for the hot water tap and a clockwise rotation for the cold water tap. Such designs are common but not ideal, and arise from a mixture of the valving arrangements that are used with standard cross taps and lever-operated taps. Many such cases have been observed by the authors in USA, Hong Kong and Taiwan.
If we are to design lever tap arrangements in which users have a high expectation of how they operate, firstly it would be required to determine such expectations and, secondly, ensure that local government standards are set that require the correct valving arrangements to be used.
There is very little research available for user expectations for lever-operated taps, apart from that of Hoffmann et al. [4] and Yu et al. [6]. The work of Yu et al. was for a single initial lever angle and used paper/pencil tests, which, as shown by Hoffmann and Whitfield [5] may not give a true measure of population expectation.
Due to little reported research about stereotypes for lever-tap operation, the study presented here aims to investigate arrangements that have strong stereotypes for tap control for increasing and decreasing water flow.
Methods
Participants
A randomly selected group participated in the experiment. These were people who were asked in various locations in and around City University if they would be able to spend about 10 to 15 minutes to participate in a study. Two different groups of participants were used. This study has volunteer participation of 240 people. One group of 120 volunteers indicated the turning direction of the levers to increase water flow and the second group of 120 participants indicated turning direction to decrease water flow. Participants were selected so that there was a wide range of age and were balanced over gender. The participants’ age ranged from 11 to 50 years old (50% male and 50% female). The majority were in the age range of 20 to 40 years old (67%). The median age of participants was 34.2. Participants were from a variety of occupations, including students, managers, industrial workers and housewives. All were fully verbally informed of the purpose of the tests and took part under the ethical guidelines of City University. The tests were performed in the University surrounds and in a large adjacent shopping center (Festival Walk). As the participants were randomly approached in the University and shopping center, the participants in the increasing flow were not available for testing on the second part of the experiment involving decreasingflow.
Equipment
All tests were done with hardware only, as illustrated in Fig. 2, with the taps situated at the top of the basin. Feedback by visual flow of water was not given for either the increasing or decreasing water flow conditions. The angular positions of the levers in the following tables and figures are relative to the 12 o’clock position, with positive angles as shown in Fig. 3. This basin and tap arrangement was shown to participants with the levers in different positions (Fig. 4). Participants placed their hands on the levers and moved the levers to indicate the direction that the levers should be moved in order to increase the water flow (Group 1) or decrease the water flow (Group 2). The experimenter noted the direction of rotation of the levers on the left and right side of thebasin.
Instructions to participants
The authors of the study indicated the instructions to the participants using a written questionnaire (presented in Cantonese).
Group 1. “The two taps on the top of this wash basin are at the moment closed, so that there is no flow of water. In what direction would you turn the levers to increase the flow of water?” This was followed by a second instruction: “On this second arrangement, in what direction would you turn the levers to increase the flow of water?” This was repeated for the remaining seven arrangements of initial lever positions, with a different random order of presentation of the lever initial positions for each participant.
Group 2. “The two taps on the top of this wash basin are at the moment open, so that water is flowing. In what direction would you turn the levers to turn off the flow of water?” A second instruction followed: “On this second arrangement, in what direction would you turn the levers to decrease the flow of water?” This was repeated for the remaining seven arrangements of initial lever positions, with a different random order of lever initial positions for each participant.
Experimental design
There were 9 initial positions of tap levers from which the ‘on’ request was made. There were also 9 conditions from which the ‘off’ request was made. Thus, there were a total of 18 test conditions. A different subject group was used for the ‘on’ and ‘off’ requests.
For the ‘increase flow’ group, the initial lever positions were measured clockwise relative to the 12 o’clock position (Fig. 4): 270/90 degrees (left/right); 270/270 degrees (left/left); 90/90 (right/right); 90/270 degrees (right/left); 135/225 degrees; 225/135 degrees; 135/135 degrees, and 225/225 degrees and 180/180 deg (forward/forward).
For the ‘decrease flow’ group, we assumed a turn angle of 30 degrees from the off position to the position where the water was flowing). This was the initial angle given to participants in Group 2. These turning directions were assumed to be that of greatest affordance as defined by Norman [7], so that there was as little interference as possible between operation of the two levers. These directions generally corresponded to a forward movement of the ends of the lever. Thus, the corresponding initial angles for the decreasing flow were: 225/135; 225/225; 135/135; 135/225; 180/180 (same for cases 5, 6, 7 and 8) and 225/135. The sets of initial angles are given in Tables 1 and 2.
Procedure
Group 1. For each initial lever position, participants were asked to indicate the direction the levers should be turned in order to increase the water flow. Data was recorded by the responses AA, AC, CA and CC (see Nomenclature), for the left and right side levers. For analysis, these responses were converted to lever movements in the forward or backward directions.
Group 2. This group were asked to indicate the direction of turn for reducing the flow. They were instructed that the taps were ‘on’ and had to be turned ‘off’.
Participants were not informed of the angle that the levers were displaced from the closed position and hence it was necessary for them to make their own assumptions about the open positions relative to tap closure. However, this is not an artificial situation as such a situation may occur in practice, where the person opening the tap may be different to the one turning off the flow.
Results
Statistical significance of stereotype strength and reversibility
The Binomial test of proportions has been used to test the statistical significance of the stereotype strength and reversibility of the stereotypes. Responses are analysed in terms of forward or backward movements of the levers. There are four combinations of forward/backward that may be used for specifying the movements made by participants. However, in all cases the dominant movements were symmetrical, that is, both levers were moved forward or both were moved backward. (The proportion of non-symmetrical responses was small in all cases Tables 1 and 2). The purely random response of forward or backward for the dual-lever movements would have a probability of 0.5. With a one-sided test, the proportions for α= 0.05, 0.01 and 0.001 are 0.58, 0.61 and 0.64, respectively.
An index of reversibility of lever movements
As noted earlier, the current study consisted of two parts (increasing and decreasing flow) and the 120 participants from the first part were not available for the second part of the experiment. Due to this experimental design, the usual definition of an Index of Reversibility (IR) cannot be used. The Index of Reversibility (IR), which is a measure of the probability of a given movement of a lever for increasing the flow being in the reverse direction of movement for the decrease of flow can be expressed as (using forward and backward movements of the lever), as follows:
In the above expression, ‘F’ and ‘B’ refer to forward and backward movement of the end of the lever and, for example, p(F/F)inc is the probability that the left and right-hand side tap levers will be moved forward for an increase in water flow. The vertical bar (|) indicates that the probability is conditional. For example, in the first term of the above expression, the p(B/B)dec refers only to those participants who responded (F/F)inc for an increase inwater flow.
Because separate participant groups were used for the increasing and decreasing flow conditions, it is not possible to evaluate the probabilities for tap closure for each individual participant. Thus, a definitive measure of reversibility of responses cannot be obtained. The best option that might be used is that of Equation 2 below, being aware of the following limitations:
When participants use a single direction for increase and the opposite direction for decrease, the IR has a value of unity. This is a condition that is not approached in the experimental data. When there is a completely random input for both increase and decrease of flow, the value of IR should be 0.5. When the same direction is used for both increase and decrease, the value of IR must be zero.
The Appendix gives an analysis of the likely errors in computation of IR using a different group of participants for the increasing and decreasing water flow cases.
This IR has a possible maximum error of ± Δp[p(F/F) inc – p(B/B) inc , where Δp is the change in proportion of responses made in the decreasing flow direction of lever movement, when there are separate participant groups, compared to the case of the calculation being made with individual participants (see Appendix).
Thus, tests of statistical significance are made relative to the random value of 0.5, with deviations away from this value indicating a response that is either partly reversible (>0.5) or has similar control movement for both increasing and decreasing flow (<0.5).
Data for the percentage of lever movements in the four possible combinations for the left and right side taps are given in Table 1 for increasing the flow and in Table 2 for decreasing the flow. Comparison of Tables 1 and 2 shows that there is very poor reversibility of the lever rotation directions for increase and decrease of water flow. The major direction of lever movement for decreasing flow was the same as that for increasing flow. If there was good reversibility of responses for increasing and decreasing flow, we would expect the decreasing flow responses to be opposite to those of increasing the water flow. This was not always the case. Values for IR are given in Table 3. All values were in the range of zero to 0.5, indicating that the dominant response for decreasing flow was the same as that for increasing the flow. This is a strong indication that participants did not have a strong expectation of the way in which lever taps operate.
Initial angles of 180/180 degrees
The cases of initial angles of 180/180 degrees (both levers facing the participant) were treated separately as they do not allow analysis in terms of the movement of levers in the forward or backward directions. Data are presented in Table 4 in the form of clockwise and anticlockwise responses for increase and decrease of water flow. This Table shows that there was poor stereotype strength, which would be accompanied by poor reversibility. These lever arrangements cannot be recommended for practical use.
Discussion
According to Campos and Paschoarelli [1], lever-operated taps have advantages in use where hands are not to be used and levers may be operated by other parts of the arm or elbow, as in operating theatres and food preparation areas. There are also obvious advantages for persons lacking muscle strength, where the mechanical advantage and limb use with levers allows ready opening and closing of the water flow. Also, as found by Boger et al. [2], the dual-lever taps were most usable by groups of participants having various levels of dementia. Because of the low levels of reversibility, it is necessary for users to learn the appropriate responses.
Stereotype strength
The stereotype strength for flow increase was strong for many of the lever initial positions. Note however that, for statistical significance (difference to a random lever movement direction), the probability of a given direction was fairly low (0.58). These probabilities of a dominant direction are very low compared to the strength of stereotype required for a good design, where higher levels would be desired. Hence, particularly when using large participant groups, levels of statistical significance may not be of great importance.
Tables 1 and 2 show that, for flow increase, there is a strong effect of the initial lever positions on the strength of the stereotype. When both levers face inwards or outwards (cases 1 to 4), the average forward lever movement is 86.3% and backward movement is 6.7%; when the levers are initially at other angles, the average forward response is 59.8% and the backward response 34%. Note that, as there are other possible responses (Tables 1 and 2), the above percentages do not add to 100%. As might be expected, there is a strong effect of affordance [as defined by Norman, 7] as when the levers are in the horizontal position, there is little space for the levers to be moved backwards. This situation does not occur when the levers are at different angles as in cases 5 to 8. In all cases where the water flow was to be reduced, the levers were at angles other than horizontal and hence space constraints did not affect the choice of lever movements. In these cases, the dominant stereotype was forward for both increase and decrease. The average forward movement was 65.3% for decreasing flow.
The outstanding characteristic of the responses for the ‘on’ request was that there was strong symmetry in lever movements with both levers moved in the forward direction. This form existed even though the request was made individually for the left and right-hand side taps. Such symmetry of response has been found in a number of aspects of human movement, but these are generally movements that are dynamic, rather than the static movements of the present study (for example, Mechsner et al. [8]). One exception to this is the study of Kunde and Weigelt [9], in which bimanual placement of objects was made to differently orientated positions. It was found that reaction times were shortest when the required movements were symmetrical, as in the lever end movements of this study. Symmetry of limb movement, both dynamically and statically, appears to be a preferred mode of humanmovement.
Responses for the ‘off’ condition also showed a very high level of symmetry, but these were not in the same direction – rather, they were divided between the F/F and B/B responses, with the greater part being F/F and hence giving a low level of reversibility (Table 3). The use of symmetrical motions of the levers for both increasing and decreasing the flow was the dominant effect in participants’ responses.
Stereotype reversibility
From aspects of stereotype strength and reversibility of responses, this study showed no advantage of using lever-operated taps over standard cross-taps. Compared with the study of Hoffmann and Whitfield [5], the strength of the stereotypes was generally lower as were the levels of reversibility. This may be due to the more wide-spread use of cross-taps compared to lever taps. Data of the present experiment showed that the level of reversibility was poor and consequently use of these taps would require learning, rather than have any strong innate expectation, as has been found for such principles as ‘clockwise-to-the-right’ in display-controldesign [10].
Because of the experimental design, using different participant groups for the increasing and decreasing flow conditions, there is the possibility of error in the measures of reversibility. This likely error was calculated to be ±0.53 Δp (see the Appendix section), where Δp is the difference in proportion of ‘decrease’ responses between the use of individual and group response data. If we assume a difference of 10% for this proportion, the likely error in the measure of reversibility is about 0.05. Table 3 data then show that no experimental condition reaches a significant level of reversibility with these possible corrections. The values in all cases are still such that the dominant responses for decrease of flow are still in the same direction of lever movement as for increase of water flow.
Practical implications of the results
Possible reasons for the poor performance of lever taps in terms of operator expectation may lie in the design of the plumbing used in such systems. A brief survey of such tap arrangements in several regions and countries (Hong Kong, Taiwan and USA) has shown plumbing systems that are inconsistent in the direction of lever movement required for increasing and decreasing water flow (example Fig. 1).
Two aspects of design are therefore important if lever taps are to become more commonly used: The plumbing arrangements must be so that an anticlockwise rotation of the left-hand valve increases flow, while a clockwise rotation of the right-hand valve increases water flow. This has, from observation in several countries, been found to be a major problem, where the standard cross-tap valve arrangement has been used with levers for operation of the taps. Standardisation of plumbing designs may be necessary. Designs that initially have good levels of stereotype strength and reversibility need to be chosen. Data of these experiments suggest that the design needs to be based on stereotype strength for increasing the flow and, as reversibility was generally poor, users would need to learn the appropriate direction of lever movement for reducing the flow. Using this criterion, Table 1 indicates that the 270/90 degree arrangement is best (both levers facing outwards), which is the commonly used lever arrangement. No doubt this stereotype is strongly influenced by the ‘affordance’ of the levers in this initial position, with possible problems associated with rotation of the levers backwards.
Conclusions
Strong stereotypes for increase of water flow were found for lever taps, especially when the initial lever ends were both facing inwards or outwards. In the case where participants were not aware of the ‘off’ position of the levers, the stereotypes for decreasing water flow were weak and hence reversibility of the stereotypes was also weak. Lever taps did not show any great advantage over cross taps in terms of operator performance as far as operator expectancies were concerned.
Appendix
When an Index of Reversibility is calculated on the basis of individual responses, an appropriate form of the expression is as follows.
This expression takes into account the probabilities that an individual, making a forward response for increasing the flow, will make a backward response of the levers for decreasing the water flow. In the current experiment, separate participant groups were used for the cases of increasing and decreasing water flow. The problem is then to find the relationship between group responses when the groups are different for each case.
It is assumed that the ‘decrease flow’ response is changed by an amount Δp from the value that the overall proportion would have been if individual responses had been made, so that:
Expansion of the expression indicates that the group response may be in error by an amount:
Taking as an example the data of Experiment 2, the mean value of p(F/F) inc is 0.73 and that for p(B/B) inc is 0.20 and hence the error in IR is approximately ±.53 Δp. If we assume, for example, that the proportions for the reverse response of the group data is different to that of individual data by 0.1, the error in IR is approximately 0.05 when using the standard equation for Index of Reversibility. This gives an indication of the level of likely error of the reversibility values when different participant groupsare used.
Footnotes
Acknowledgments
The authors thank anonymous reviewers and editor for their comments, which have improved the paper.