Case Study #6: Information Processing and ATM Design. Barcelona Consulting Inc.

Transcription

1 Case Study #6: Information Processing and ATM Design By Niklas Blymiller, Phillip Greenwood, Stephanie Lamont, and Katie Tice of Barcelona Consulting Inc. IE 327, Section 002 October 14, 2009

2 Executive Summary In today s fast paced world, easy and rapid withdraw from personal banking accounts is a necessity for most debit card holders. The solution for this demand comes in the form of an Automatic Teller Machine, more commonly known as an ATM. The key reason for the success of ATMs across the world is the ability to quickly withdraw money from any ATM location. However the issue of balancing safety with speed becomes an issue- how to keep things quick while still providing optimal account security? In order to do so, ATM machines have keypads where the cardholder enters their four-digit security code. Although most of the efficiency of the ATM transaction process relies on the format and dialogue of the screen, some of it relies on the layout of the keypad itself. The Barcelona Consulting team analyzed two keypad layouts- square and in-line layouts- and found that by calculating the Index of Difficulty (ID) for each layout, it could be determined that the square layout is a better design with an ID of 1.38 compared to that of 2.76 for the in-line design. Also, to evaluate the effectiveness of key size, a Fitts Tapping Task test was completed. It was found that it took an average of seconds to complete one task on two buttons measuring 1.5L x 1.5W and having a center-to-center distance of two inches. This test was conducted on both a computer and on paper in order to determine whether the keypad should be on a touch screen or be physical. The results indicate that the faster time is associated with the on-paper test, therefore indicating that a physical keypad would be the better option. Furthermore, to determine reaction time, two additional tests (Simple Reaction Time and Choice Reaction Time) were conducted and the results were plotted on a Hick-Hyman Plot. According to the R 2 value displayed the graph, the linear relationship between the x and y values is , an indication that the relationship is very linear and therefore reliable. Finally, when comparing the response times for the specific task (T 1 ), the redesigned layout (T 2 ), and the actual ATM (T 3 ), it can be seen that the values for T 1 and T 2 are somewhat less than that of T 3. This is an indication that the specific ATM that the test subject used required slightly more time to complete the specific task. With respect to the values of T 1 and T 2, there is not much variance. This is an indication that the current design and the proposed design have similar response times. It can be seen, however, that the redesign is better is better when the ID is compared for both designs. The original design has an ID value of 1.38 while the redesign has an ID of only , indicating that the proposed design is in fact easier to use. We hope that this information helps you in your decision making process. If you have any questions or concerns, please feel free to contact our firm. -Barcelona Consulting Incorporated

3 Introduction The main purpose for an ATM is to provide money for customers in a secure, fast and efficient manner. In order to fulfill this objective, an ATM keypad must be used. There are two different types of keypads that may be chosen from: an inline layout like that found on a computer keyboard or a 3 x 3 square layout with either [1 2 3] or [7 8 9] at the top. In order to effectively choose an appropriate layout for the most efficient keypad, there are certain tests that must be conducted and analyzed. The Difficulty Index and Fitts Tapping Test provide information about reaction times and ultimately for selecting the most efficient keypad layout. From the Fitts Tapping test, it is seen what the average reaction time is due to the average time it takes to tap between two squares on a monitor with a computer mouse. By experimenting with different sizes of buttons and different spaces between the buttons, it was possible to verify the optimal size and spacing for the buttons. Expected benefits of observing and improving this process will have positive effects on customers who regularly utilize ATMs. The new layout will be user friendly, allowing for comfort and familiarity, and these improvements will collaborate with speed to provide ATM users with a secure, speedy, and worry free transaction. Objectives Observe and analyze the current layout of a typical ATM keypad layout. Determine the best keypad layout (square or in-line) based on the index of difficulty of each. Redesign the keypad layout for optimum speed of cash withdrawal Verify that the proposed design is better than the original design. Methods The Barcelona Consulting team first observed the two ATM keypad layouts and calculated the Index of Difficulty (ID) for both layouts. With this information, the appropriate keyboard layout was determined to be the square layout due to its lower ID. With this layout in mind, observations regarding trials within Fitts tapping task, Simple reaction time, and Choice reaction time tests were made. Using the obtained values, a linear regression plot was made by plotting the data into excel. These observed values were then entered into an equation that was able to relay information about reaction times. In order to ultimately evaluate the redesign, the entry process of four equally likely numbers into the original ATM keypad and the team s proposed keypad was timed.

4 Results Several different values were calculated in order to determine the efficiency of the current and proposed designs. Table 1 shows the calculated values for the index of difficulty (ID) of the square keypad design and the in-line keypad design. The results from these calculations are significant and indicate that one layout is better than the other. Table 2 displays the average tapping times from the Fitts Tapping Tests conducted in DesignTools. Typically, an ATM would either have the keypad on a touch screen or physically laid out somewhere near the display screen. Due to the fact that the Barcelona Consulting team did not have an actual touch screen to test, they used a computer simulation using the DesignTools program instead. In addition to this, they created a drawing of the same test sample in order to find out what the average tapping time would be for a physical keypad. This drawing can be found in Appendix A. The results were and seconds for the two computer simulations and seconds for the drawn keys. Table 3 shows the average reaction times from the Simple Reaction Time (SRT) test and the three Choice Reaction Time (CRT) tests. These results illustrate the test subject s reaction times to simple tests involving sight perception and finger-eye coordination and speed. For the simple reaction test, the test subject took an average of seconds to respond. With respect to the tests involving finger-eye coordination and speed for different sets of numbers, the test subject required an average of , , and seconds to respond for each test respectively. To determine the reliability of these values, they were then plotted in a Hick- Hyman Plot. Figure 1 shows this plot. The plot then gave the equation of the line, y = 0.102x , which enabled the Barcelona Consulting team to later calculate the channel capacity (bandwidth), and the response times for 10 equally likely alternatives, 4 successive digits, and a specific task (T 1 ) by adjusting the equation to read RT (Response Time) = a + bh. These calculations are shown in Figure 2 and can be found in Appendix B. To verify improvement from the current design to the proposed design, the ID was calculated for the proposed design. These values are shown in Table 4. In order to compare additional improvement, the overall response times for the specific task (T 1 ), the redesigned layout (T 2 ), and the actual ATM (T 3 ) had to be compared. Table 5 shows this data. It can be seen that the values for T 1 and T 2 are somewhat less than that of T 3. It can also be observed that the response time for the original ATM keypad layout is slightly less than that of the proposed design. Furthermore, a life-size drawing of the redesigned keypad layout was created in order to have the original ATM test subject mock the specific task of entering four digits, ENTER, and FAST CASH. This drawing is represented by Figure 3 and is located in Appendix C.

5 Results (cont d) Table 1. Calculated Values of Index of Difficulty Index of Difficulty Square design 1.38 In line design 2.76 Table 2. Average Tapping Times from Fitts Tapping Tests Average Time (sec) Touch Screen Touch Screen Hand Table 3. Average Reaction Times from Simple Reaction Time (SRT) and Choice Reaction Time (CRT) Tests Average Time (sec) SRT CRT CRT CRT Table 4. Calculated Values of Index of Difficulty- Original vs. Proposed Index of Difficulty Original Square Design 1.38 Proposed Square Design

6 Results (cont d) Table 5. Comparison of Reaction Times Average Time (sec) Specific Task Redesigned Layout Actual ATM Reaction Time (msec) Hick Hyman Data Information (bits) y = 0.102x R² = Series1 Figure 1. Hick-Hyman Data from ATM User

7 Discussion After analyzing the average distance and width between each button of a typical ATM keypad, an ID value of 1.38 was determined for the square layout, and an ID value of 2.76 was determined for the in-line layout. According to the original calculated Index of Difficulty for both the square and the in-line layout, the optimum layout is a square design, however there remains room for improvement. Accounting for the majority of the ID value was the distance between the buttons. By eliminating the gaps within each of the buttons, the distance value was slightly decreased. To further improve the ID, adjusting the size by both width and length of each individual button is needed. The original dimension of a button was 13/16 W x 5/8 L. The proposed dimension of a button is now 1 W x ½ L. After modifying the size of the buttons and eliminating the gaps, a new Index of Difficulty value of.8998 was determined. In order to further evaluate the effectiveness of the redesign, a time to complete a specific task of entering 4 random digits, pressing enter, and then fast cash was needed. After completing the task, a time value of seconds was found. This time is slightly greater than the time calculated by using the equation given by the Hick-Hyman Plot to estimate the time for the original keypad. The time difference, however, is justified since an equation does not necessarily account for human error. Therefore, the redesigned keypad would allow the user a better Index of Difficulty as well as a fast time. Conclusion With respect to the original Index of Difficulty for the original square and in-line keypad layout, the values of 1.38 and 2.76 are not sufficient. Utilizing data obtained from tests such as Fitts Tapping Tests, Simple Reaction Time (SRT) and Choice Reaction Time (CRT) Tests and inputting them into specified equations, it was observed where the design needed improvements and how to accommodate for these niggling areas. Furthermore, comparing the time it takes to input equally likely numbers into the original design and the proposed design indicates how the new design maximizes efficiency. By adjusting the size of the buttons, and eliminating the gap between them, the value of the Index of Difficulty was able to drop to a value of.8998.

8 Appendix A Figure 1. Drawing from Fitts Tapping Test

9 Appendix B Figure 2. Calculations

10 Appendix C Figure 3. Drawing of Proposed ATM Keypad Layout