Abstract
Social networking sites like Twitter and Facebook have become very popular among college students. These sites enable people to be in constant contact and communication. Their value in meeting educational objectives is less clear. We use Twitter “tweets” to remind students of psychology topics while they are outside of class. The intervention itself was straightforward: Students received an informative tweet about once per day. Students remembered these topics significantly better in a test situation. The Twitter intervention appears to be an effective way to increase memory for important class concepts.
Many people have recently discussed how they used social networking as a means to enhance instruction (e.g., Galagan, 2009; Manzo, 2009; Young, 2009). College educators have used sites such as Facebook and Twitter as a platform for students to ask questions of the instructor or to have students carry on conversations with each other outside of class. For example, Young (2009) discussed how he used these sites during class time so that students can ask or respond to questions anonymously. While the instructor leads the class discussion, students converse and ask questions on a course website. A teaching assistant monitors the site, bringing interesting points and questions to the instructor’s attention. Although this provides a forum for students to ask questions they might not have asked otherwise, it unfortunately also provides a way for students to be distracted and ignore or miss important course content.
Manzo (2009) discussed how she used Twitter to have students discuss course content outside of the classroom setting. In this method, students shared insights and other online content they found useful and asked questions regarding points of which they were unsure. This mimics the class discussion boards that are common to most learning management systems (e.g., Blackboard or WebCT) but has an immediacy not allowed by such systems because most students have real-time updating abilities to social networking sites via a mobile device.
Although instructors have considered how social networking can increase the social presence of students in a course, and how those interactions can subsequently enhance the learning experience (Dunlap & Lowenthal, 2009), most empirical work has centered more on usability issues (e.g., Greenhow & Reifman, 2009), rather than on learning effectiveness. Those studies that have considered learning effectiveness have been anecdotal in nature lacking empirical investigation (e.g., Miners, 2010). In the current research study, we provide empirical data supporting the use of social networking in an educational setting.
We investigated how we could use social networking to enhance the learning of course concepts. Twitter allows the quick transmittal of short messages (tweets) to anyone who subscribes to a user’s message feed. These messages keep the receivers apprised of the sender’s thoughts and actions on an ad hoc basis. Users receive these messages not only through traditional computers but also through most mobile devices. Through these means, people can receive twitter messages whenever and wherever they are. Some Twitter users have begun using the site to tweet as historical figures, either humorously (e.g., http://historicaltweets.com) or in a more educational way (e.g., http://twhistory.org). We took that idea and implemented an educational intervention in which we used Twitter to send messages to students, approximately once a day, outside of class concerning a concept that was central to the class discussion that day.
This intervention is a novel way to signal to students what concepts are important just as boldface text does in a textbook. Many textbooks have marginal inserts that provide more information about a concept. The tweets that we developed for this study serve this role in a humorous and time-delayed manner. Research has shown a small learning advantage for students using a text with such signaling (Nevid & Lampmann, 2003).
Cognitive and developmental literature consistently reports facilitative effects of reminders on long-term memory tasks. Laboratory studies with adults have indicated that intermittent test sessions reminding students of previously learned information boost subsequent memory performance for a variety of stimuli (Barber, Rajaram, & Marsh, 2008; Karpicke & Roediger, 2008). Furthermore, when utilized appropriately, humor can benefit student learning, interest, and performance by reducing anxiety, boredom, and stress while also improving student attitudes toward the course subject (Berk, 2003; Berk & Nanda, 1998; Goodman, 1995). Kaplan and Pascoe (1977) reported higher recall of course-related information from students after the use of humorous examples. We believe that using technology to highlight and remind students of important knowledge after the class and in a humorous manner is consistent with past memory research and will provide an advantage for recall of those concepts in subsequent assessment.
Method
Participants
Sixty-three undergraduate students (42 female, 21 male) at the University of Tampa participated in the study. All students attended one of two introductory psychology sections taught by two of the authors. The majority of students in these classes were 18–20 years old, Caucasian, and from a middle to upper socioeconomic status. We excluded data from four additional students who dropped the course, and one other student had incomplete data. Otherwise, all students enrolled in those sections participated in the intervention.
Materials
Both course sections used the full edition of Weiten’s (2010) textbook Psychology: Themes and Variations. We wrote two sets of 84 tweets. One set related to the course concepts, six per chapter. The tweets covered the main aspects of a chapter, not the minutiae of it. We wrote them to be both informative and humorous, pertaining to a topic recently covered in class. The tweets were written in such a way that they appear to come from different individuals by prefixing each tweet with a person’s name (to mimic the “@” convention within tweets to indicate a receiver, we developed a “!” convention to indicate a sender). We adopted as much as possible other conventions and abbreviations popular among tweeters. One of the tweets for the memory chapter was, “!Sperling: I want to order a Snuggie, but the phone # flashed too quickly. Does anyone have it?” We refer to the participants who received these tweets as the course concept group. We wrote a second set of 84 tweets that contained simple jokes that were not psychology related, such as, “Why did the basketball player bring a duck to the game? She wanted to shoot a fowl shot!” (For a complete set of tweets in both conditions, see http://utweb.ut.edu/sblessing/psychology_resources/psychtweets.html). We refer to the participants who received these tweets as the humor-only group. This second group served as a placebo control group, in that they received all the mechanics of the treatment (receiving the tweets) but not the treatment itself (the psychology-related content).
We used two instruments to assess the impact of the tweets on student memory. One assessment was a cued recall task that we referred to as brain dump to the students. In this activity, students listed up to five items that came to mind first when thinking about each particular chapter. Students also had to list the source of that memory (e.g., read it in the textbook, heard it in a lecture, saw it in a tweet, etc.). We primed students with only the name of the chapter and gave five blank lines below the prime to complete their work. This instrument also contained questions concerning how often and in what manner the students accessed the tweets. The second, more critical, assessment was a set of multiple-choice items that corresponded to particular tweet content. These items appeared on the regular course exams given throughout the semester and, from the student’s perspective, were indistinguishable from the other items (e.g., a question pertaining to the Sperling tweet above would ask a general question about iconic memory). We used two such questions for each chapter covered during the course.
Procedures
We randomly assigned students to the two different twitter feeds within the two classes at the beginning of the semester. On the first day of class, we gave students an instruction sheet that contained general information about Twitter, how to subscribe to a Twitter feed, and to which Twitter feed they should subscribe. Based on prior experience (Blessing, Blessing, & Fleck, 2010), we knew that few of the students (less than 20%) would have past personal experience with Twitter, but essentially all used Facebook. With that in mind, we also created two Facebook groups that mirrored the two Twitter feeds. Items posted to the Twitter feeds automatically and immediately appeared as status updates to these groups. Only the items that the instructor posted to the Twitter feed appeared in these Facebook groups. Students could choose whether to subscribe to their assigned feed via Twitter itself or the linked Facebook group. This allowed students to use which technology they preferred while functionally receiving the same information as the other students in their group. Sixty-three percent chose to subscribe to the Twitter-like Facebook group. Regardless of which they chose, all students had the same experience: a short item appearing almost daily in a news feed of a social media site to which they subscribed. We ascertained that all students registered for their assigned set of tweets, and only their assigned set of tweets, with all students participating. We informed the students that we were interested in how humor affects the learning process, that the tweets would contain humorous statements and jokes, and that there were multiple feeds so they should not talk about their tweets to their classmates.
Students received the tweets once per day on average. Instructors used a Twitter utility that allowed them to queue tweets in advance and to set the specific transmission time for each tweet. (Information concerning both this queuing utility and how to create a Facebook Twitter mirror is located at the web address given previously.) At four points during the semester (the class period before an exam), students completed the cued recall task based on the previous three to four chapters of material. In addition, students completed the multiple-choice items that we had selected a priori as matching particular tweet content during each regularly scheduled exam. These questions allowed us to compare student performance on these items across the two conditions. We gave five participation points for the cued recall task, plus a bonus point if they listed a particular tweet (regardless if it was a course concept or a humor-only condition tweet), which accounted for less than 3% of the grade.
Results
We used the information provided on the cued recall activity to assess if the people who received the psychology-related tweets were more likely to think about the concepts contained within those tweets than the people who received just the joke tweets. A research assistant blind to condition scored the tweets. She categorized each item that the students wrote as either pertaining to a tweet item or not. She adopted a liberal criterion for counting an item as a tweet item. For example, a tweet from Chapter 1 was, “!Skinner: @Watson good call on just the observable behavior. I’ll tell you more about operant conditioning and free will at poker Friday.” She gave credit as a tweet item if the student listed anything to do with Skinner, Watson, behaviorism, operant conditioning, or free will. Although this casts a wide net, it makes the task of deciding if an item was a tweet item or not more objective.
Across chapters, participants (n = 30) in the course concept condition listed a tweet item 33% (SD = 0.10) of the time and those (n = 33) in the humor-only condition 29% (SD = 0.08) of the time. While the trend is in the predicted direction, the difference between the groups is not significant, t(61) = 1.80, p = .08, d = 0.45. In examining the source information that students listed, in only 19 instances (of the 2,788) did a student explicitly state that their remembrance came from a tweet. The vast majority of students reported their remembrances from either class discussion (1,471 times) or from the textbook (1,195 times). The other 103 instances came from film clips, a CD-ROM resource, or an unnamed source.
Although the results of the first assessment were not significant with its open-ended nature, we believed that students who received the psychology-related tweets would perform better on the targeted exam items related to the contents of the tweets. One instructor covered 12 chapters and the other instructor covered 13 chapters, and both instructors gave four exams throughout the semester, with the final exam being noncumulative. In the end, all students received either 24 or 26 of these multiple-choice questions. The instructors intermingled these items with the other multiple-choice items. When writing or selecting the other items for the exam (the exams had 30–40 multiple-choice items, along with some short answer and essay questions), we used no special criteria. In this way, six or eight of the items on each exam were the special tweet-based items, with the other items being representative of the material as a whole. For these analyses, we applied the arcsine transformation to test score percentages. Significance tests statistics present analyses using the transformed variable as the dependent measure. Descriptive statistics present raw percentages to retain interpretability. The humor-only group (M = 0.67, SD = 0.16) performed significantly worse than the course concept group (M = 0.74, SD = 0.12), t(61) = 2.02, p = .048, d = 0.52, on the target multiple-choice items. Analyzing exam performance as a whole, there was no difference between the humor-only group (M = 0.78, SD = 0.10) and the course concept group (M = 0.78, SD = 0.12).
We asked students to report how they attended to the Twitter feed each time they did the cued recall activity. Participants could choose to review the set of tweets at any time. Examining usage across all four of these activities, 90% (57 of the 63) of the participants self-reported attending to the feed at least once during three of the four time periods between the cued recall tasks; 83% (52 of the 63) did so all 4 times. A sizable majority attended to the intervention at various intervals throughout the entire semester. We examined just those participants who self-reported looking at the feeds on at least a weekly basis. Eighteen of the 30 participants in the course concept condition met that criterion (60%), and 23 of the 33 (70%) in the humor-only condition. This subset of the humor-only group (M = 0.67, SD = 0.17) had the same level of performance as the whole humor-only group, and the subset of the course concept group (M = 0.78, SD = 0.11) scored slightly higher than the course concept group as a whole did. Comparing these two subsets (i.e., the weekly-or-better users in both conditions) to one another results in a significant difference, t(39) = 2.43, p = .02, d = 0.82.
Discussion
When asked specifically about tweeted content on exam items, students who received the tweets significantly outperformed students who did not receive the psychology-related tweets. For items that instructors deem most important, this simple intervention appears to be effective. Although not significant, students recalled tweeted content slightly more often than nontweeted content. Students rarely reported the tweets as a memory source. Given the nature of the intervention, that of getting students to reflect on class material for a few additional moments outside of class, the fact that the trend exists is interesting. A previous study (Blessing et al., 2010) contained a significant effect using the same cued recall task.
Previously, we noted the similarity of our intervention to signaling in textbooks, a tactic that many books employ. In their study of text signaling, Nevid and Lampmann (2003) also found similar results, with students earning 83% on test items related to signaled text and 75% on nonsignaled text test items. This 8% increase for signaled text in their study mirrors the 7% increase (74% vs. 67%) we had for the tweeted context. Furthermore, for the heavier users of the intervention, those who monitored the tweets at least weekly, that difference grows to 11% (78% vs. 67%). Disparities exist between the two procedures, but there appears to be a small, consistent effect on memory for when a student notes an important concept. For those concepts that an instructor wants to ensure that students grasp, adopting a technique like this appears to be effective.
As typically occurs in classroom research, several limitations exist within this study. Using the placebo-like control group allowed everyone access to using the technology, thereby controlling for novelty, but using a control group that did not use Twitter at all would allow another point of differentiation between groups. We used a design in which we randomly assigned students in the same class to the different groups. This controls for various course variables, such as instructor and content covered in class, but meant that we could not discuss the tweets in class nor carry on extensive online discussion. We only looked at the impact of the tweets themselves, which might have served to attenuate the effects one might see if used in a more substantive way in the class. In addition, students in the course content group might have discussed their tweets with the humor-only group, but this again would serve only to attenuate the effect. Although students in the course content group engaged the tweets in a way that positively affected memory for the material, as evidenced by the results, class and online discussion of the tweet concepts should result in deeper learning. We were also unable to closely monitor student attention to the tweets. Although we attempted to check these as best we could, we were not able to obtain specific data related to these variables. This might have affected students’ motivation and interest in the material, as might the content of the tweets themselves, impacting their memory for course-related information. The analysis that examined frequency indicated that students who more often saw the tweets remembered the material better. A future study could examine the frequency of the tweets more explicitly and strongly. It would be of benefit to replicate these findings with a larger and more diverse population, where one could more substantively look at any differences between ethnicities, ages, or genders. Most notably, existing comfort levels of technology use for both instructor and student would impact success at using this intervention. One possibility would be to examine Twitter versus other technology, such as e-mail or a website. Twitter has flourished in the age of smart phones, but other ways for presenting reminders may yield similar results in a manner that better fits a particular course structure.
With the seeming ubiquity of social networking site usage among college students, integrating a means of learning within that information stream seems beneficial. This study, one of the first to provide an empirical account of using Twitter in a course, presented one way to inject psychology-related content into that stream in a manner that increased student’s retention of that content. This intervention is domain general, so that other instructors can model similar assignments within their own courses.
Footnotes
Acknowledgment
We thank Jaclyn Maass for assisting with data entry and analysis.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.