Speed-Accuracy Tradeoff
Mikko Allen D. San Miguel
Pamantasan ng Lungsod ng Maynila

Abstract
In Speed-Accuracy Tradeoff, a low speed means higher accuracy and a high speed means lower accuracy. This principle is applied in terms of reading. It is predicted that when people read faster, they tend to be less accurate about what they read. This is tested by compelling subjects, to read faster. The subjects were asked to read faster than average, by increasing the target number of lines they were required to read. Also, they were asked to cross-out all the letter e’s that they see as they go through a reading material. Most of the results are consistent with the prediction. When subjects read faster, they committed more mistakes and when they read slower, they were able to commit less. However, other findings are inconsistent with this prediction. For example, even when the subjects were not required to read faster than average, they were still unable to gain a higher score in correctly crossing-out the letter e’s that they came across.
Introduction
Is a job done fast, a job done inaccurately or can a job be done both fast and accurate? Individuals attempt to perform well in both factors (Zimmerman, 2011). Everyone wants a job well done in a minimum amount of time. For example, it would be amazing if a repairman can fix a broken phone properly in less than the average time it usually takes; this is beneficial to both parties. The client saves more time than usual. Similarly, it gives the repairman more time to fix other things which means, more income. Most of the time, this is not possible. People tend to increase their response time (speed) at the cost of reducing accuracy, or decrease response time in order to increase accuracy (Proctor & Vu, 2003). In speed-accuracy tradeoff, a low speed means higher accuracy and a high speed means lower accuracy. For example, when doing printed assignments, reports, etc., students try to type faster, usually, it is impossible for them not to accidentally hit a key that they were not supposed to hit. This would then require them to go back to the part where they committed the typographical error. Because of this, they would now type carefully to avoid mistakes and this takes time. This usually is not a problem unless a task requires a limited amount of time to respond (Fairbrother, 2010). If a task or job requires a person to respond in a limited time, a person has no choice but to speed-up. This then leads to the decrease in accuracy. Sometimes, tasks require more accuracy than speed. For example, in reading, it is more important that the reader understood the book or text than, finishing it fast. Unless of course, let us say that a teacher requires his/her students to finish reading a text in a certain amount of time. It is possible for some students to actually understand the text even if they read it fast but most probably, students would sacrifice accuracy in order to finish the book or text in the limited amount of time. If the teacher now, decides to give a short test about the text s/he assigned the students to read, it is probable that the students would commit more error than if they were given more time. Still, this demonstrates that many tasks require individuals to choose between speed and accuracy. How much a person actually understands when reading a book or text may be dependent by how fast s/he reads it; the way a person reads a book or text may affect how much s/he understands about it. As said earlier, a person increases speed at the cost of accuracy. With that it could be said that, if a person reads a text faster, it is expected that s/he will be less accurate (or commit more mistakes) about it. However, this is still a hypothesis; it is still to be tested. An average person can read 250 words per minute or can read an average page in a book or a document in 1to 2 minutes (“Mindtools”, n.d.). There are some ways where a person can be required to read faster. An example to this is the previously stated situation where a student is required by a teacher to finish a book or text in a limited time. From this, we can again say that, a person can be compelled to read faster by setting a target and a time limit. As mentioned, a person can only read 250 words on average. If a person is then asked to read 250 words for only 30 seconds (two times faster than average), will s/he be accurate about what s/he read? This can be seen throughout the study.
Method
Participants
Six undergraduate subjects participated in this study. Subjects are second year psychology students from Pamantasan ng Lungsod ng Maynila. They belong to the same class, taking Experimental Psychology Lecture and Laboratory (PSY 221 & PSY 221.1, respectively) and consist of 2 males and 4 females; ages of 17-18 (17.7 mean age).
Procedures and Materials
The subjects are first given the instruction to cross out all the letter e’s that they come across while reading a material that is to be given to them; they are given 30 seconds to do so. Subjects are seated in between an experimenter and a timer. The reading material which is a cut-out article from a newspaper is given to the subjects by the experimenter. The subjects are free to start whenever they are ready; the timer is attentive to start timing upon the moment which the subjects started crossing-out letter e’s. The experimenter counted the lines covered by the subjects. That is, lines where the subjects have crossed out letter e’s. The first run of this procedure determined the baseline.
When the baseline was determined, a target was set for the second run; a new reading material was then given to the subjects. An additional 10% of the baseline was added to itself as the target. For example, if a subject was able to cover 14 lines at first, the target set for that subject to cover for the second run was 16; about 10% was added to the actual baseline. This was done to compel the subjects to read faster. In the experiment, two pens which are different in colour were used. Pen 1 was used by the subjects to cross out the letter e’s while pen 2 was used by experimenter to underline the target.
After every run, a new reading material is provided for the subjects; these materials also have different targets. The subjects are still to cross out all the letter e’s that they come across. For the third run, the target was lowered by about 10% of the baseline; this is contrary to the second run. A target of 12 was set for a baseline of 14. However for the fourth run, a greater target was set by adding 15% to the baseline. For example, a target of 17 was set for a subject whose baseline was 14. Similar to the third run, the last run of this procedure has a lesser target. 15% was taken off the baseline; a target of 11 was set for a baseline of 14. The experimenter recorded the number of lines each subject was able to accomplish. The timer also, made sure that the subjects were only able to cross out letter e’s for 30 seconds.
The materials are then gathered after the experiment. From the start of the article to the target line, the letter e’s were encircled. A crossed-out and encircled letter e means that the subject was able to cross it out correctly while a simply encircled letter e means that the subject missed it. These were separately counted. If a subject was able to cross-out a letter which is not e, it is counted as missed. This process, determined information such as: (1) the number of e’s correctly crossed-out, (2) the number of e’s missed, (3) the total number of e’s, and (4) the percentage error.
Results
The baseline or the number of lines the subjects were able to accomplish at first, as well as the target number of lines is presented in Figure 1.

Figure 1. Baseline and Targets
When the target is increased by either 10% or 15%, the number of lines that a subject should accomplish, also increases. It means that the subjects are compelled to read the material given to them faster. This is shown in Figure 1. Almost all the subjects’ baseline scores are roughly the same, except for subject 1 which score’s higher. The number of e’s crossed correctly by the subject can be seen in Figure 2.

Figure 2. Number of E's Crossed-Out Correctly
As can be seen in Figure 2, in most of the subjects, the number of e’s crossed-out correctly increased in both target where the baseline was increased by 10% and the other decreased by 10% (B+10% and B-10%). The increase in number of e’s crossed-out correctly in the baseline decreased by 10% (B-10%), was expected but not in the other (B+10%). This may be because of the increased number of lines. More lines to cover means, more e’s to cross-out. As expected, most of the subjects got lower number of e’s crossed-out correctly when the target was increased by 15% ( B+15%). Inexplicably, when the target was decreased by 15% (B-15%), the number of e’s crossed-out correctly also decreased in most of the subjects. This means that even the target was lower, still some of the subjects were unsuccessful at correctly crossing out the letter e’s. The number of e’s missed by the subjects is shown in Figure 3.

Figure 3. Number of E's Missed
Figure 3 shows that in most of the subjects, the number of e’s missed increased when the target is increased by 10% (B+10%); this result was expected. It is also expected that if the target is decreased by 10% (B-10%), the number of e’s missed will also decrease. In most subjects, when the target was increased by 15% (B+15%) the number of e’s missed also increases. In contrast, when the target was decreased by 15% (B-15%), the number of e’s missed also decreased. Overall, results presented by Figure 3 were as expected. The percentage error is presented in Figure 4.

Figure 4. Percentage Error
The percentage error was determined by dividing the number of e’s missed by the sum of the number of e’s crossed-out correctly and the number of e’s missed or, the total number of e’s. A lower percentage error of course, is better because this means lower number of mistakes. As seen in Figure 4, when the target was increased by 10% (B+10%), the percentage error also increased; as the target was decreased by 10% (B-10%), most of subject’s percentage error also decreased, an expected result. When the target was increased by 15% (B+15%), the percentage error also increased compared to the baseline. Also, when the target was decreased by 15% (B-15%), the percentage error also decreased as predicted.
Discussion
Speed-Accuracy Tradeoff has long been studied. Prior studies suggest that when people favour accuracy, they perform a certain task slower and if they favour speed, they perform certain tasks faster hence, making them less accurate about it. Although, in some cases, people can be faster and still be accurate about the task which they are performing. Predictions then were derived from these studies. One of these is that when a person does a task faster, s/he is most likely to be less accurate (inaccurate) about it.
The aforesaid statement was applied in reading; which led to the prediction that if an individual reads faster, then s/he may be less accurate about it. This may also mean that when an individual reads slower, s/he will be more accurate about it; however, the some parts of the results are somehow inconsistent with this statement. When the subjects were required to read more lines faster (the target number of lines to be read, were increased by 10%), most of them were still able to correctly cross-out the letter e’s. This may be because of the number of lines. Moreover, when the subjects were less compelled to read faster (the target number of lines to be read were decreased by 15%), most of the subjects got lower number of correctly crossed-out letter e’s, compared to the baseline. Nevertheless, the rest of the results are consistent with the prediction. As predicted, when the subjects were required to read more lines faster (the target number of lines to be read, were increased by either 10%or 15%), most of them, committed more mistakes. In addition, when the subjects were less required to read faster (the target number of lines to be read were decreased by either 10 or 15%), most of the subjects were able to commit less mistakes. This further concludes that a person increases his/her reaction at a cost of reducing his/her accuracy.

References
Fairbrother, J.T. (2010). Fundamentals of Motor Behavior. Retrieved from http://www.humankinetics.com/excerpts/excerpts/the-speed-accuracy-trade-off.
Karsilar, H., Simen, P., Papadakis, S., & Balci, F. (2014). Speed Accuracy Trade-Off Under Response Deadlines. Front. Neurosci. 8:248. Doi: 10.3389/fnins.2014.00248
Speed-Accuracy Tradeoff. (n.d.). Psychology Dictionary. Retrieved from http://psychologydictionary.org/speed-accuracy-tradeoff/.
Speed Reading. (n.d.). Mind Tools. Retrieved from http://www.mindtools.com/speedrd.html.
Toward a unified view of the speed-accuracy trade-off: behaviour, neurophysiology and modelling. (n.d.). Frontiers. Retrieved from http://journal.frontiersin.org/ResearchTopic/1647.
Zimmerman, M.E. (2011). Speed-Accuracy Tradeoff. Springer Link. Retrieved from http://link.springer.com/referenceworkentry/10.1007%2F978-0-387-79948-3_1247.