Abstract
Objective:
The aim of this study was to determine if interruptions affect the quality of work.
Background:
Interruptions are commonplace at home and in the office. Previous research in this area has traditionally involved time and errors as the primary measures of disruption. Little is known about the effect interruptions have on quality of work.
Method:
Fifty-four students outlined and wrote three essays using a within-subjects design. During Condition 1, interruptions occurred while participants were outlining. During Condition 2, interruptions occurred while they were writing. No interruptions occurred in Condition 3.
Results:
Quality of work was significantly reduced in both interruption conditions when compared to the non-interruption condition. The number of words produced was significantly reduced when participants were interrupted while writing the essay but not when outlining the essay.
Conclusion:
This research represents a crucial first step in understanding the effect interruptions have on quality of work. Our research suggests that interruptions negatively impact quality of work during a complex, creative writing task. Since interruptions are such a prevalent part of daily life, more research needs to be conducted to determine what other tasks are negatively impacted. Moreover, the underlying mechanism(s) causing these decrements needs to be identified. Finally, strategies and systems need to be designed and put in place to help counteract the decline in quality of work caused by interruptions.
Introduction
Imagine that you need to finish writing a report for a presentation tomorrow morning. You sit down and begin typing. Your BlackBerry rings, indicating you have received a new e-mail. You read it and respond. You attempt to collect your thoughts and begin working on your report when your boss comes in and asks you a question. Once again, you have been interrupted and before you know it, 30 min have passed and you have gained no ground on completing your report. In office environments, employees are interrupted up to six times per hour (PitneyBowes, 1998) and may shift tasks every 3 min (Gonzalez & Mark, 2004).
With such an impact on daily life, it is not surprising that there is an abundance of research focusing on interruptions. Interruptions have been well documented in office environments (Czerwinski, Horvitz, & Wilhite, 2004; Gonzalez & Mark, 2004), in aviation (Dismukes, Young, & Sumwalt, 1998; Loukopoulos, Dismukes, & Barshi, 2001), while driving (Strayer & Johnston, 2001), and in the health care industry (Chisholm, Collison, Nelson, & Cordell, 2000; McGillis, Pedersen, & Fairley, 2010; Tucker & Spear, 2006; Westbrook, Coiera, et al., 2010; Westbrook, Woods, Rob, Dunsmuir, & Day, 2010).
A majority of traditional interruption research involves time and errors as measures of disruption, as they are often the most appropriate metrics for determining the effects that interruptions have on task performance. Research focusing on time has shown that interruptions increase completion time on tasks (Eyrolle & Cellier, 2000; Gillie & Broadbent, 1989; Hodgetts & Jones, 2006b; Monk, Boehm-Davis, & Trafton, 2004; Trafton, Altmann, Brock, & Mintz, 2003; Ziljstra, Roe, Leonora, & Krediet, 1999). In some cases, the delay or time lost as the result of an interruption can lead to negative consequences. For example, interruptions have been shown to increase failures to detect traffic signals and slow reaction times while driving (Strayer & Johnston, 2001). Research focusing on errors has shown interruptions reduce overall accuracy in task performance (Eyrolle & Cellier, 2000; Gillie & Broadbent, 1989) and increase postcompletion errors (Ratwani & Trafton, 2008). They have also been shown to disrupt overall surgical flow in the operating room (Wiegmann, ElBardissi, Dearani, Daly, & Sundt, 2007), where a common error is repeating or skipping a step during a procedure.
However, time and errors may not be appropriate measures of disruption for all domains. For example, qualitative domains, such as reading comprehension and writing, may be better served by analyzing the quality of the final work product as the primary measure of disruption. If an interruption causes only a loss of time when writing an essay and time is not essential, then the interruption has not negatively affected the quality of the work. However, if an interruption has resulted in the lessening of the quality of the essay, then the interruption has had a real impact on the work product. We have been unable to identify any research that attempts to measure the effect of interruptions on the quality of the final product. Since research has shown that interruptions negatively affect performance on tasks as measured by time and errors, we hypothesize that interruptions will also negatively affect the overall quality of the work product.
The memory-for-goals model (Altmann & Trafton, 2002) has been used in several experiments to successfully predict interrupted task performance (Cades, Boehm-Davis, Trafton, & Monk, 2011; Hodgetts & Jones, 2006b). This is an activation-based model instantiated in the ACT-R, a computational modeling framework (Anderson, 1993; Anderson et al., 2004; Anderson & Leibere, 1998). The model describes tasks as being represented as goals in memory, with each goal having a certain level of activation associated with it. Working on a specific task (the active task) generates activation of that task goal. In this framework, the most “active” goal drives behavior, and the goal associated with the current task will always be the most highly activated goal. When an interruption occurs, the current goal (the primary task a person is working on) must be suspended, and the goal associated with the interrupting task becomes the most highly activated. When this suspension occurs, the activation associated with the primary task goal decays as a function of time; thus the longer one is away from the task, the more decay that occurs and the lower the activation level becomes for that task. Therefore, the longer one is away from a goal, the more difficult it will be to retrieve.
Due to the time factor, there are instances in which a task goal can decay beyond the ability to be retrieved. Retrieval of this task would require an outside mechanism to boost activation. The memory-for-goals model does provide such mechanisms for boosting the activation level of goals. The first mechanism is strengthening, which suggests that the frequency and/or recency of a goal can boost activation. The second mechanism is priming, which suggests that environmental and mental cues or contexts can boost activation. Strengthening can be explained as the ability to rehearse during an interruption, which would increase the activation of that goal and thus lead to a higher likelihood that the goal would be retrieved. For example, an alert before an interruption, such as a knock at the door, may allow someone to rehearse what he or she is doing before answering the door, thus boosting the activation of the primary task. Similarly, a simple interruption task may allow one to think back to the primary task while still performing the interruption task, boosting the activation of the primary task goal upon each rehearsal. Moreover, the priming constraint allows for boosts in activation through environmental context or cues. For example, if a lengthy interruption task has led to a decrease of activation of the primary task below a threshold at which it is able to be retrieved, a cue in the environment, such as a highlighted last sentence or a cursor on the computer screen, can help boost that activation level back above threshold, allowing for retrieval (Cades, Ratwani, Boehm-Davis, & Trafton, 2008; Hodgetts & Jones, 2006a; Ratwani & Trafton, 2008).
This model has typically been used to explain resumption performance directly following an interruption as a person attempts to resume the primary task. Broadly speaking, the memory-for-goals model makes robust predictions for resumption performance in terms of time to resume, with longer interruptions leading to longer resumption periods while accounting for the constraints discussed previously. Furthermore, the model makes the prediction that longer interruptions would be more likely to result in errors upon resumption of the primary task, as activation of the last action may be too low due to decay to recall. However, given the correct environmental context and enough time, this handicap can be overcome in most instances (Cades et al., 2008; Hodgetts & Jones, 2006a; Ratwani & Trafton, 2008).
These time delays and errors are important to consider and to protect against. Much research has been devoted to understanding these predictions and how to mitigate the effects of decay. We currently have an abundance of research on interrupted task performance on the few steps following the interruption, and the memory-for-goals model is robust in accounting for this performance. However, it is unclear whether this model can be used to explain interruption performance on the overall quality of a task. On the basis of the predictions of this model, some might postulate that given enough time, decrements caused by interruptions could be overcome. However, multiple time delays and sequential errors could potentially lead to goal confusion, interference, and error-laden performance.
Experimental Rationale
We were interested in whether the overall quality of work suffers when interrupted. To study this topic, we needed a task whereby quality could be defined beyond the number of errors made or time to complete the task. We selected a complex, creative thought task that mirrors a common real-world task, outlining and writing an essay. To mirror interruptions and their effects in the real world, the interruptions needed to occur without warning. Ensuring the interruptions occur without warning reduces the time available for transitioning from the primary task to the interruption task to a very small time frame. We were also interested in whether the timing of the interruption mattered. Thus, interruptions were introduced at different phases of the experiment, either when the participants were outlining their essay or when they were writing it.
Experiment 1
Method
Participants
Twenty-seven students from George Mason University participated for course credit. The data from 1 participant were excluded as both graders felt the participant did not take the assignment seriously. The participants (18 females and 9 males) had an average age of 23.6 years and were fluent in English.
Task and materials
The primary task required participants to outline an essay using pen and paper and then to write the essay on a computer using Microsoft Word. The three essay prompts and assignments (see Table 1) came from a stock bank of essay topics created by the College Board (2012b).
Sample Essay Prompt and Assignment
The interruption task consisted of answering a series of unrelated questions using pen and paper. These questions included basic arithmetic, American history, problem solving, and word unscrambling. The answer format was either fill in the blank, multiple choice, or open response. Participants completed as much of the interruption task as they could in the given time. The interruption task was self-paced, and it continued throughout the interruption interval. The researchers ensured participants actively completed the interruption task and scored the accuracy of participant responses to the interruption task.
Design and procedure
The experiment consisted of a planning phase immediately followed by an execution phase. During the 12-min planning phase, participants outlined their essay. During the 12-min execution phase, participants typed their essay. The experiment used a within-subjects design in which two factors were manipulated: presence or absence of an interruption and placement of the interruption (planning or execution phase). Participants served in all three conditions (planning phase interruption, execution phase interruption, and no interruption) and wrote all three essays. Participants were assigned to conditions using a Latin squares design that counterbalanced the essay type (Prompts 1, 2, and 3) and interruption location (planning, execution, and control). During an interruption condition, the participants were interrupted three separate times for 60 s each time. The interruptions occurred at the 3-, 7-, and 11-min marks. The 3 min lost to interruptions were added to the total time available for that phase so that participants would still have a total of 12 min to complete that phase (see Figure 1). The interruptions occurred without warning. A piece of paper with the interruption task questions was placed over the participants’ outlines. They were instructed to answer as many questions as possible. Once the interruption time was complete, the paper was removed and the participants continued working on the primary task. Participants were not informed that this study focused on understanding interruptions. They were informed that they were supposed to complete each task (outline or essay and the interruption) to the best of their ability.
Time course of each phase and condition. Twelve minutes were allotted for each phase.
Measures
Word counts were calculated for both the outlines and final essays. A graded score, as averaged by two independent graders, was used to evaluate the final essays. The interruption task was scored for accuracy.
Two independent graders evaluated all essays using the College Board Essay Scoring Guide (College Board, 2012a; available at https://sat.collegeboard.org/scores/sat-essay-scoring-guide). Scores range from 0 to 6. A 0 represents complete failure and indicates that the participant did not write at all or did not write on the appropriate topic. A 6 represents “clear and consistent mastery” of the essay that “effectively and insightfully develops a point of view on the issue” (College Board, 2012a). The measure of agreement between graders was tested using Cohen’s Kappa. The results of the interrater reliability was Kappa = .705. The graders were blind to the experimental conditions. They received only the final essay text. No other identifying information was attached.
The two independent graders were trained on how to properly evaluate essays before scoring the essays from the participants. Both reviewed the College Board Essay Scoring Guide (College Board, 2012a) and read preevaluated essays from all the score ranges (0 to 6). Each grader practiced scoring on preevaluated essays, comparing their scores to those official scores (College Board, 2012a). Last, the graders independently scored preevaluated essays and then discussed their own evaluations together in an effort to ensure consistency. The training period lasted approximately 7 days. They were paid for their work and had no other affiliation with the experiment.
Results and Discussion
Mauchly’s test of sphericity was used to verify the assumption of homogeneity of variance for all the repeated-measures ANOVAs reported in this experiment. The results showed that the assumption was met for homogeneity of variance in all cases, p > .10.
A repeated-measures ANOVA was conducted to examine whether the overall quality of work was affected by interruptions. Each final essay score was averaged from the two scores of the independent graders. The analysis revealed a significant difference among the scores across the three conditions, F(2, 50) = 21.638, p < .001. Post hoc tests using a Bonferroni correction revealed that the scores in both interruption conditions were significantly lower than the scores in the noninterruption condition (p < .01 for both comparisons) with roughly a 0.5-point decrement on the 6-point scale in the interrupted conditions. There was no difference in quality (p > .10) between the two interruption conditions (see Table 2).
Mean Essay Scores, Outline Word Counts, and Essay Word Counts for Experiment 1
p < .05. **p < .001.
Before doing further analysis, we wanted to ensure that fatigue and type of essay (Essay Prompts 1, 2, and 3) did not influence the lower scores. A repeated-measures ANOVA revealed no significant reduction in quality over time, F(2, 50) = 1.143, p > .10, and no significant difference as a function of type of essay, F(2, 50) = .267, p > .10.
To understand the source of the quality differences, the number of words produced in each condition was examined. This analysis revealed a significant difference in the final word count of each essay across the three conditions, F(2, 50) = 4.86, p = .012. Post hoc tests using a Bonferroni correction revealed that the word count was significantly lower when interruptions occurred during the execution phase (M = 281.6) when compared to the noninterruption (M = 306.8) and the interruption during planning phases (M = 315.3). That is, interruptions during the execution phase led to a reduction in the amount of content produced.
The number of words produced in each outline was also examined. This analysis revealed no significant difference in the final word count of each outline across the three conditions, F(2, 50) = .661, p > .10. Post hoc qualitative analysis of the outlines revealed that participants used many different formats to plan their essays, including a traditional bullet format, a pros-and-cons list, and a web diagram.
A correlation coefficient was computed to assess the relationship between the word counts of the outlines and final essays and the final score. There was a significant correlation between the word count of each essay and the final score of the essay, r(26) = .723, p < .01. Participants who were able to produce more content in their essays scored higher. This result is consistent with real-world results from actual essays scored for the SAT. That is, data analyzed by Dr. Perelman of the Massachusetts Institute of Technology from real SAT essays and scores associated with those essays revealed that word count significantly correlated with final score (Winerip, 2005). Moreover, the Princeton Review highlights essay length as a tip to scoring higher on that portion of the SAT (Princeton Review, 2013). There was not a significant correlation between the word count of each outline and the final score of the essay, r(26) = .304, p > .10.
Accuracy for the interruption task was also measured. The average score was 81%, with scores ranging from 64% to 98%. A correlation coefficient was computed to assess the relationship between the accuracy of the interruption task performance and the final score. There was no significant correlation between interruption task performance and the final score of the essay, r(26) = .07, p > .10.
Experiment 2
The results from Experiment 1 suggested that interruptions during a creative, complex task affect the overall quality of work. However, it is possible that participants did not have enough time to complete the assignment effectively. It is also possible that participants anticipated the interruptions and this played a role in the outcome. In an effort to further investigate these findings, two changes were implemented for Experiment 2. First, the time at which the participants were interrupted was randomized, instead of being presented at regular 3-min intervals, to abate any potential expectancy effects. If the participants were able to anticipate incoming interruptions, they may have been able to use rehearsal or other means to help retain information during the interruptions. The second change extended the overall time that participants had to complete their essay in the execution phase. This change was implemented to ensure that participants had the necessary time to complete their essay.
Method
Participants
Twenty-seven students from George Mason University participated for course credit. The participants (22 females and 5 males) had an average age of 18.7 years and were fluent in English.
Task and materials
The primary task, interruption task, and materials were identical to Experiment 1.
Design and procedure
The overall design and procedure was identical to Experiment 1 with two exceptions. First, instead of interruptions occurring at regular, 3-min intervals evenly spread across the total time (3:00 to 4:00, 7:00 to 8:00, and 11:00 to 12:00), interruptions occurred at random intervals. Therefore, the time in which the interruptions occurred varied within and between subjects. Second, the overall amount of time the participants had to complete each essay was extended. Instead of having 12 base minutes and 3 min of interruptions, participants had up to 20 min to complete their essays in the interrupted conditions. Participants also had up to 20 min in the no-interruption condition.
Measures
The measures were identical to Experiment 1. The measure of agreement between graders was tested using Cohen’s Kappa. The results of the interrater reliability was Kappa = .768. Again, the graders were blind to the experimental condition associated with each essay. They received only the final essay text, and no other identifying information was attached.
Results and Discussion
Mauchly’s test of sphericity was used to verify the assumption of homogeneity of variance for all the repeated-measures ANOVAs reported in this experiment. The results showed that the assumption was met for homogeneity of variance, p > .10.
A repeated-measures ANOVA was conducted to examine whether the overall quality of work was affected by interruptions. Each final essay score was averaged from the two scores of the independent graders. The analysis revealed a significant difference among the scores across the three conditions, F(2, 52) = 7.742, p < .001. Post hoc tests using a Bonferroni correction revealed that the scores in both interruption conditions were significantly lower than the scores in the noninterruption condition (p < .05 for both comparisons) with roughly a 0.5-point decrement on the 6-point scale for the interrupted conditions. There was no difference in quality (p > .10) between the two interruption conditions (see Table 3).
Mean Essay Scores, Outline Word Counts, and Essay Word Counts for Experiment 2
p < .05. **p < .001.
Before doing further analysis, we wanted to ensure that fatigue and type of essay (Essay Prompts 1, 2, and 3) did not influence the lower scores. A repeated-measures ANOVA revealed no significant reduction in quality over time, F(2, 52) = 2.001, p > .10, and no significant difference as a function of type of essay, F(2, 52) = .908, p > .10
To understand the source of the quality differences, we examined the number of words produced in each condition. This analysis revealed a significant difference in the final word count of each essay across the three conditions, F(2, 52) = 9.266, p < .001. Post hoc tests using a Bonferroni correction revealed that the word count was significantly lower when interruptions occurred during the execution phase (M = 269.4) when compared to the noninterruption (M = 317.3) and the interruption during planning phases (M = 304.9). That is, interruptions during the execution phase led to a reduction in the amount of content produced.
The number of words produced in each outline was also examined. This analysis revealed no significant difference in the final word count of each outline across the three conditions, F(2, 52) = 1.724, p > .10. Post hoc qualitative analysis of the outlines revealed that participants again used many different formats to plan their essays.
A correlation coefficient was computed to assess the relationship between the word counts of the outlines and final essays and the final score. There was a significant correlation between the word count of each essay and the final score of the essay, r(27) = .567, p < .01. Again, this result is consistent with real-world results from actual essays scored for the SAT (Winerip, 2005). There was not a significant correlation between the word count of each outline and the final score of the essay, r(27) = .134, p > .10. Once again, participants who were able to produce more content in their essay scored higher overall.
Accuracy for the interruption task was also measured. The average score was 78%, with scores ranging from 61% to 95%. A correlation coefficient was computed to assess the relationship between the accuracy of the interruption task performance and the final score. There was no significant correlation between interruption task performance and the final score of the essay, r(27) = –.02, p > .10.
Experiment 2 allowed participants to take up to 20 min to complete their essays with interruption time factored out. The average completion time was not significantly different across condition, F(2, 52) = 1.232, p > .10.
General Discussion
Previous work on interruptions has focused on the impact interruptions have on the ability to resume the primary task as measured by time and errors. The goal of this study was to investigate the effect that interruptions have on the overall quality of the primary task work product.
Our data suggest that interruptions lead to a reduction in quality for a complex, creative writing task. Quality scores for the essays were lower in both interrupted conditions. In addition, word count was reduced when the execution phase was interrupted. This reduction may have been due to lost time in resuming the primary task—that is, delays suffered by the participants while collecting their thoughts to resume after an interruption (Altmann & Trafton, 2004). Specific resumption lag times would have been difficult to measure because resumption occurred within the participants’ thought process rather than with a physical process, such as clicking or typing. However, if the reduction of quality was caused solely by resumption lags, the increased time given to participants in the second experiment should have resulted in higher quality scores. Since quality did not increase in the second experiment, we believe the interruptions are causing more than just a delay in the thought process. It may be that the interruptions are causing a complete disruption in the participant’s train of thought and that this disruption is causing a reduction in the overall amount of content produced, resulting in a lower-quality final product.
Attempting to apply the memory-for-goals model (Altmann & Trafton, 2002) to these results is difficult, as the model does not speak directly to an impact on quality of the work product. However, goal activation is an important component of the model that may be relevant in understanding these results. The primary goal of this experiment is to complete the essay. In turn, each argument the participant develops represents a subgoal that needs to be completed in service of completing the larger goal of completing the essay. When participants are interrupted, both the higher-level goal (writing the essay) and the subgoal (developing a specific argument) are interrupted.
Memory for goals posits that the activation of a goal must be greater than the activation of the interference level to drive the action upon resumption from an interruption. The level of interference will depend on the amount of “mental clutter” to which a participant has recently been exposed. In our experiment, interference arises from two primary sources, the interruptions themselves and, less obviously, the number of previously sampled subgoals. That is, all previously sampled subgoals up to that point represent mental clutter that will increase the interference level, which in turn reduces the likelihood of resumption of the proper subgoal being sampled preinterruption. To successfully resume the subgoal being sampled preinterruption, the participant would need activation of that goal to be greater than the interference level. Keeping the activation level of the subgoal being sampled preinterruption above the activation of the interference level becomes increasingly difficult in our experiment over time, as the interference level is continuously growing due to repeated interruptions and the creation of new subgoals across the experiment. Therefore, we believe that the memory-for-goals model can account for the lack of proper activation of the most recently sampled goal, or subgoal, postinterruption. However, we believe it is too far of a stretch to say that activation directly speaks to changes in quality of work.
In line with the aforementioned argument, participants may have difficulty reactivating the specific subgoal. Instead, participants may reactivate another subgoal or create a new subgoal. That is, when an interruption occurs in the planning phase, the participant may be interrupted before fully developing the idea he or she was mentally planning or currently writing. When the interruption comes to an end and participants resume the task, they may not necessarily return to the last argument (subgoal) on which they were working. Anecdotally, the proctors noticed that some participants appeared to return to the beginning of their outline or essay after an interruption occurred. In either case, it is likely that the subgoal being sampled preinterruption was not fully developed postinterruption. As such, the quality of that argument (subgoal) would be lower. This return to the highest level is consistent with modeling that was done to examine the impact of interruptions on completion of checklists (Diez, Boehm-Davis, & Holt, 2002). This finding is also consistent with anecdotal evidence in the real world. For instance, when people are interrupted when telling a story, they often do not resume where they left off, perhaps replaying some of the story in their head or repeating the last few lines. That is, “Where was I?” followed by “Oh, that is right, this happened and then this happened” is often seen in real-world settings.
Conclusion
This research represents a crucial first step in understanding how interruptions affect the quality of work. Adults today are fond of multitasking, and they feel that they can do so with no impact on the individual tasks in which they are engaged (Aratani, 2007). This work suggests that this is not the case.
Future research in this area is needed to further explore these results. The types of tasks that are negatively impacted by interruptions needs to be further examined. Our results have shown that interruptions negatively impact the overall quality of work in a complex, creative writing task. However, it is unclear if the same decrements would occur in other types of tasks. The relationship between expertise on a task and quality of work could also be explored. In procedural tasks, overall disruption reduces as skill increases (Altmann & Trafton, 2007). It is unclear if this would translate to quality of work.
Once a better understanding of the types of tasks is understood, the mechanisms behind the decrement should be explored more fully. This research might be done through an evaluation of physiological measures or performance tests, which can be used to assess changes in the body and brain. For example, research has shown a strong correlation between task performance, as measured by resumption times, and working memory capacity, as measured by the operation span (Werner et al., 2008). Therefore, assessing working memory capacity may provide insight into individual performance while interrupted. Other research has shown that anodal stimulation of the dorsolateral prefrontal cortex using transcranial direct current stimulation can increase neural firing in that area, leading to increased performance on a host of tasks (Nitsche et al., 2008). Therefore, using anodal stimulation over the dorsolateral prefrontal cortex to increase neural firing may improve task performance while interrupted.
Once the mechanisms by which interruptions affect performance are better understood, researchers can begin to focus on creating and testing procedures and techniques that can overcome reductions in quality of work. For example, can an alert of an incoming interruption be effective at overcoming the reductions in quality of work? Or would encouraging an approach that has been effective at reducing errors, such as rehearsal, be better? As technology continues to increase the number of interruptions we experience daily, answers to these questions will become increasingly important.
Key Points
Interruptions are commonplace in the home and office. Little research has focused on the impact of interruptions on overall quality of work.
Our research suggests that quality is reduced when interrupted during outlining and writing of essays.
Traditional memory-based models of task performance provide only partial explanation of our findings and may not be suited to understanding how interruptions affect the overall quality of work in a complex, creative task such as essay writing.
Footnotes
Acknowledgements
The research described here was supported in part by a grant (FA9550-10-1-0385) to the Center of Excellence in Neuroergonomics, Technology, and Cognition (CENTEC) from the Air Force. The views expressed in this submission do not reflect the official views of the sponsor. We would also like to thank Stacey Fernandes, Anuj Sharma, and Christina Lau for their ongoing aid and support in completing this research. Additionally, we would like to thank Daniela Barragán for editing the final manuscript.
Cyrus K. Foroughi is a PhD candidate in human factors and applied cognition in the Department of Psychology at George Mason University. He received his bachelor of science in psychology in 2012 from George Mason University.
Nicole E. Werner is a PhD candidate in human factors and applied cognition in the Department of Psychology at George Mason University. She received her bachelor of science in psychology in 2009 from George Mason University.
Erik T. Nelson is a research scientist for Honeywell. He received his PhD in human factors and applied cognition from George Mason University in 2013.
Deborah A. Boehm-Davis is dean for the College of Humanities and Social Sciences and university professor for the Department of Psychology at George Mason University. She received her PhD from the University of California, Berkeley.