Evaluation of computer-assisted instruction in histology: Effect of interaction on learning outcome


  • Lih-Wei Lei,

    Corresponding author
    • Chihlee Institute of Technology, No. 313, Sec. 1, Wen-hua Rd., Baneiao, Taipei 220, Taiwan (ROC)
    Search for more papers by this author
    • Fax: +886-2-2253-7527

    • Dr. Lei is the principal investigator of this study. After the study was completed at the University of Washington, he became an assistant professor at Chihlee Institute of Technology. His research and teachings are focused on the applications of computers, multimedia, and the Internet in higher education.

  • William Winn,

    Search for more papers by this author
    • Drs. Winn, Scott, and Farr are professors in curriculum instruction, medical education and biomedical informatics, and biological structure, respectively, at UW as well. Dr. Winn's research interests include instructional theory and design of computer-based instruction while Dr. Scott's research and teachings are focused on the informatics competencies of medical and graduate students. Dr. Farr teaches in the areas of histology and immunology.

  • Craig Scott,

    Search for more papers by this author
    • Drs. Winn, Scott, and Farr are professors in curriculum instruction, medical education and biomedical informatics, and biological structure, respectively, at UW as well. Dr. Winn's research interests include instructional theory and design of computer-based instruction while Dr. Scott's research and teachings are focused on the informatics competencies of medical and graduate students. Dr. Farr teaches in the areas of histology and immunology.

  • Andrew Farr

    Search for more papers by this author
    • Drs. Winn, Scott, and Farr are professors in curriculum instruction, medical education and biomedical informatics, and biological structure, respectively, at UW as well. Dr. Winn's research interests include instructional theory and design of computer-based instruction while Dr. Scott's research and teachings are focused on the informatics competencies of medical and graduate students. Dr. Farr teaches in the areas of histology and immunology.


The purpose of this study was to determine better strategies for the design and use of computer-assisted instruction (CAI) in health science subjects that require visual learning. Evaluation of current use of CAI was focused on three CD-based modules developed to teach histological images to beginning medical students at multiple sites. For internal control, students' learning outcomes and perceived effectiveness were analyzed with their demographic characteristics, computer attitude, computer experience, and learning behaviors being considered. Results indicated that students who used at least two different CAI programs scored significantly higher on the final examination than those who used only the CAI tool designed by their site's instructor. Further investigation indicated that students might have benefited from the interactive features of a specific CAI tool. Such scaffolds could have successfully supported encoding processes while students were restructuring their mental models. In addition, students perceived the CAI programs to be more effective when the tools were fully integrated into the curriculum. Perceived module effectiveness was significantly correlated with examination performance, suggesting a well-designed and appropriately used CAI tool may help students achieve not only learning efficiency, but also better learning outcome. Anat Rec (Part B: New Anat) 284B:28–34, 2005. © 2005 Wiley-Liss, Inc.


There are a number of health science subjects that require visual learning, such as anatomy, histology, radiology, hematology, and urinalysis. Because of the visual complexity of medical images and the intensive tutorial experience necessary to develop image recognition expertise, professional training programs in medical schools have only been able to educate limited numbers of students at certain places during restricted periods of time (Jaffe,1989).

Nowadays, new technologies are expanding opportunities by casting a wider net in health science learning. Interactive CD-ROMs and the Internet with multimedia have made it possible for a personal computer to provide a learning environment similar to that of conventional time-consuming one-on-one tutorial methods. At the University of Washington School of Medicine (UW-SOM), for example, the Department of Biological Structure has developed a series of online and CD-based instructional resources. With easier and more flexible access to high-quality medical images, faculty and students now have the potential to benefit from the efficiency and convenience brought by the digital information era.

Since the Web-based and CD-based learning tools became available in the histology course at UW-Seattle, students' feedback on the Web-based/electronic resources was almost universally positive when asked in the online end-of-quarter survey, “What aspects of the course did you find most valuable/useful?” Histology course evaluation records also showed that student ratings of the Web-based/electronic resources were the highest of eight course features for 2 successive years (with average ratings of 4.8 and 4.0 on a five-point scale in autumn quarter 2001 and 2002, respectively). However, it was unclear if students who highly valued the electronic learning resources truly learned from those tools.

Accordingly, medical education researchers at the UW-SOM began a study that would evaluate the use of three existing computer-assisted instruction (CAI) programs designed for medical students in the histology course at UW-SOM's multiple WWAMI program sites (the WWAMI program provides community-based medical education for five states, including Washington, Wyoming, Alaska, Montana, and Idaho; the five sites were labeled as UW-Seattle, UWyo, UA, MSU, and UI, respectively, in this study). The relationship between perceived effectiveness and learning outcome was of special interest. For years, other researchers have tried to compare the effects of medical CAI with traditional media, such as lectures, laboratory exercises, or textbooks; the studies were usually difficult to do well because variables such as differences in pedagogical techniques, differences in informational content, and the novelty factor have usually been inadequately controlled (Clark and Solomon,1986; Clark,1992). Therefore, this study mainly focused on the main effects of different CAI tools while avoiding comparison between the effects of CAI and microscopic tools on learning. Especially, the pedagogical strategies used by each CAI were analyzed and how students might better benefit from the CAI tool was investigated. Particular attention was paid to control variables that may influence the main effect caused by the CAI itself.


This study involved three CD-based programs (labeled as HCD1, HCD2, and HCD3) developed by three different histology course faculty members, each located at a different course site in the UW-SOM's five-state WWAMI region. The characteristics of the three programs are summarized in Table 1. First developed in 1998, the HCD1 contains over 400 images and allows students to study, magnify, or review an image through different presentation modes. The HCD2, developed in 2002, contains over 450 images. It is designed to teach histology to students using a simple format that provides interactive learning in several modes. The HCD3, containing nearly 500 images, is designed to include all course-related information for students at the UW-Seattle site. It was developed in 2003. Before this study began, each CD-based program had been used only at the site where it was developed (Table 2).

Table 1. Summary of characteristics of three CD-based learning programs that were labeled as HCD1, HCD2, and HCD3*
  • *

    It appeared that the HCD1 and HCD2 were more interactive.

Overall interface/program design   
 Quick access to different parts of the program+++
 User manual is available++ 
 Hyperlinks to online resources  +
 Program can be opened with a browserRequires installation++
 Illustrations of histophysiology  +
Presentation of images   
 Key features are highlighted separate from images++ 
 Key features are prelabeled on images  +
 True microscopic zoom in/out + 
 Digital zoom in/out+ +
Learning interaction   
 Interactive (clickable) images + 
 Interactive (clickable) key words+  
 Instant feedback in quiz mode++ 
Table 2. Participant grouping and the use of instructional CDs at different school sites before and after this study began
Study GroupSubject SourceaNumber of ParticipantsParticipation Rate of Total ClassPrimary CAIAdditional CAI
  • a

    UW-Seattle, Washington; WSU, Washington State University; UWyo, Wyoming; UA, Alaska; MSU, Montana; and UI, Idaho.

1UW-Seattle4951%HCD3HCD1 and HCD2
2WSU/UI1335%HCD1HCD2 and HCD3
3MSU2051%HCD2HCD1 and HCD3
 UA/UWyo  Not availableHCD1, HCD2, and HCD3

In order to provide equal access to the three histology modules to students at each site, copies of four histology CDs (one HCD1 disk, one HCD2 disk, and two HCD3 disks) were distributed to all first-year medical students throughout the five WWAMI school sites at the beginning of course in the 2003–2004 academic year. Students were encouraged but not required to use any or all the additional CDs. Students also had access to the identical online learning resources, including course schedule, links to references, and a Web-based quiz bank.

Three major study groups, with reasonable number of subjects in each, were formed for comparative study. Group 1 comprised of 96 students who were taking the histology course at the UW's main campus in Seattle, Washington; group 2 comprised of 37 students at a combined Washington State University (WSU)/University of Idaho (UI) site; group 3 had a total of 39 students with 10 at the University of Alaska (UA), 21 at Montana State University (MSU), and 8 at the University of Wyoming (UWyo). Due to the fact that subjects of this study were all volunteers, 82 participants were successfully recruited from the total of 172 students enrolled in the histology class for a combined participation rate of 48% (Table 2).

At the beginning of this study, the 82 volunteer participants were invited to complete an online pretest, the computer attitude and experience survey, which consisted of 25 (mainly Likert-scale type) questions asking their attitude toward and experience with computers and the Internet. The histology CDs were then integrated into the curriculum by different instructors at each WWAMI site (Table 3): faculty at the UW-Seattle and MSU sites used their own CDs to teach in the laboratory section and left it up to the students as to whether to use the other two CD programs; faculty at other WWAMI sites allowed students to make use of the three CD programs as they desired. Close to the end of the course, all students were invited to fill out a course evaluation survey (a survey generally used for all courses in the WWAMI program) plus a histology CD evaluation survey (a survey specifically designed to gather data regarding whether students used each CAI program, how much they used the CDs, and their satisfaction). Subgroups of students from the different sites were formed for analysis once the number of respondents from each site became clear in order to improve statistical power.

Table 3. How each WWAMI site integrated histology CDs into the curriculum in autumn 2003
CDs used by the instructor to teach in laboratory session+ +  
Actual microscope available for use/demonstration in laboratory ++++
Students are required to use real microscopes to identify images +  +
CDs used by students to study on their own++++ 


In general, neither age nor sex was found to be correlated with students' perceived effectiveness of the CD-based and the Web-based study tools. Students' computer attitude was significantly correlated with their computer experience, but not correlated with their perceived effectiveness of the CD-based and Web-based resources or with their learning outcome (examination performance). Based on the data gathered from the UW-Seattle site (group 1), students' general computer experience, such as their experience with Internet browser, E-mail software, and word processing, was correlated with time spent learning histology online (Pearson correlation coefficient = 0.323; P < 0.05). However, students who spent more time learning online did not score better on the final examination even though they tended to have reviewed online practice quizzes and reference Web sites more often. Students' MCAT scores (Medical College Admission Test composed of verbal reasoning, physical sciences, writing sample, and biological sciences) were significantly correlated with performance on the histology examination (Pearson correlation coefficient = 0.345; P < 0.05). This was the only demographic variable significantly correlated with final examination scores. Such finding implied that conclusions directly drawn from the comparison of mean examination scores between study subgroups could be problematic and all variables that might potentially affect learning outcome had to be considered during analysis.

In terms of how the CDs have been used in each study group, students in group 1 had HCD3 as the primary CD; of these, 50% used HCD2 and 12.5% used HCD1. Students in groups 2 and 3 tended to choose either the HCD1 or the HCD2 to supplement their studies. More than half of them indicated that they felt both HCD1 and HCD2 had useful image-based virtual laboratory and interactive practice quizzes. Two-thirds of the students in group 2 not only used their primary CD, but also found the HCD2 to be a useful supplemental learning tool. Students in group 3 tended to prefer the HCD2 (about 60% of them rated HCD2 as somewhat or very useful). One-third of them, however, reported that the HCD1 was a useful supplementary learning tool.

Students Might Achieve Better Learning by Using CDs Not Designed by Their Own Instructor

Results from the histology CD evaluation survey indicated that students in group 1 tended to use different learning strategies for the CDs and the online resources. Those who used at least two different CAI programs (and also spent some more time learning through the CDs) scored significantly higher on the final examination than did those who used only the CAI designed by their own instructor (Table 4). When the mean scores were compared using ANCOVA to control for the concomitant variables (e.g., MCAT score), the difference was still significant, although the effect size was relatively small [F(1, 28) = 4.45; P < 0.05; partial η2 = 0.137]. The former group, however, did not spend significantly more time using all the available electronic learning tools. In other words, students who used at least two of the available CD-based programs tended to spend less time accessing the online quiz bank and the reference Web sites.

Table 4. t-test of average final examination scores of two study subgroups at UW-Seattle
 Used HCD3 OnlyUsed HCD2 and HCD3
t-value 2.484
P (two-tail) 0.013a

In fact, nearly half of the students in group 1 reported that they felt the CDs or the Web-based resources were of most value and they were one of the most useful aspects of the course. Eight of them specifically cited HCD2, the one that was not developed by their site's instructor. However, they did not expect that this learning program would help them achieve better examination scores. For example, a student wrote, “The HCD2 was particularly helpful, although what it covered wasn't really tested directly.” Another student also found the HCD2 as the most useful tool “except that the slides on the CD were not representative of the slides on the exam.” In theory, how the students were tested in the examination should determine what was the best way to prepare for the examination. As some students pointed out, “The online question bank was helpful in test preparation. Unfortunately, I was not aware of it until the third exam” and “the question banks were great. Wish I'd started using them earlier.” So ideally, the best way for them to have prepared for the examination was to review the online quiz bank. The time students spent learning online should have been correlated with their examination performance. That, however, was not the case. Therefore, the CD-based programs have proved to be effective instructional and learning tools since students who spent more time learning from the CDs and spent less time reviewing the online practice quizzes tended to perform better on the examination. It also appeared that making use of that additional CAI program was an efficient way to achieve better learning outcome because students did not have to spend much additional time learning through the tool.

Interaction Could Be a Key Element That Made an Image-Based CAI Successful

In this study, students who reported using both HCD3 and HCD2 in group 1 scored slightly but significantly higher on the final examination than those who reported using HCD3 only. The HCD2 did not include more images than HCD3, nor did it include more textual explanations. As some students pointed out, the HCD2 may have been too simple to give users enough in-depth information. However, it allowed for interactive learning opportunities with every image, which the HCD3 did not provide (Fig. 1): students using HCD2 first learned the names of tissue components and were then guided by the program to identify these components in images. With this CD, students could use a mouse to move the cursor to any spot on the image. Then, with a mouse click at any given point on the image, the structure at that point would be identified by a label. Students could also click on a zoom-in or zoom-out button to learn the image at a different magnification. For each image, the program included test questions that related tissue structure to function, diseases, and cell biology. When students clicked on an answer, immediate feedback was provided to help them understand why their answers were correct or wrong. On the other hand, students using HCD3 for each course session were presented with all example images of small size and image descriptions on the same page, a layout similar to that of a textbook (Fig. 2). Students could then click on an image to view a larger copy, which already had the key features labeled with arrows. Thus, students were not required to look for those features by themselves.

Figure 1.

A sample screen of the HCD2 program. Students can click anywhere on the image and receive immediate textual feedback. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com].

Figure 2.

A sample screen of the HCD3 program. The enlarged image, with key features labeled already, is displayed in a separate window. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com].

For instructional electronic images, the purpose of designing interaction strategies is to support the encoding process, through which students must progress in order to experience meaningful learning. Encoding support will be effective to the extent that it helps students select, organize, and integrate the learning experience within mental models that have been clearly formulated in memory (Bliss,1994). When students were learning histological images with the CAI tools, they were exposed to an unfamiliar domain and presented with complex information (i.e., digital images with different morphological features and their components with different patterns). The lack of familiarity and perceived complexity caused competition for limited cognitive resources in their mental models. In such unfamiliar domain, complexity could be managed by initially inducing simple, discrete interactions, such as a few mouse clicks on different parts of the images. Then, the immediate feedback, which was also part of the interaction, further helped students recognize inadequacies in their mental models and stimulated deeper understanding (Govora and Hannafin,1995; Hannafin et al.,1996). Through such repeated cognitive restructuring by “click-and-feedback,” students have become more and more familiar with the features and patterns of each cell and tissue category. This might be the reason that the HCD2 has effectively promoted students' learning through interaction.

Students Perceived Electronic Resources as More Effective When Integrated Into Curriculum

In this study, students perceived the CAI programs to be more effective when instructors actually used them to teach in their classes instead of merely providing the CDs as references for self-study. The UW-Seattle and MSU were the only two sites where instructors used the CAI to teach in laboratory sessions. Then in the course evaluation survey, students from the two sites also rated the Web-based/electronic resources relatively higher (mean = 4.3 and 4.4, respectively, on a five-point scale vs. 3.8 in WSU/UI; 4.4 in UA; 3.4 in UWyo).

Due to the fact that examination scores of students at non-Seattle-based WWAMI sites were not included in the analysis (because students at each site were tested with different questions), it was not clear if different ways of integrating the CAI into the curriculum led to different examination performance. However, two variables measured in the course evaluation survey have been identified to be significantly correlated with examination scores (Fig. 3). They are rating of the amount learned in the course and rating of Web-based/electronic resources.

Figure 3.

Correlations between students' responses to three course survey items and their final examination performance (not a formal path analysis). Numbers indicated Pearson correlation coefficients: asterisk, P < 0.05; double asterisk, P < 0.01.

Students Learned More If Required to Use a Microscope

It was found that both WSU/UI and UWyo students gave a higher rating on the item asking their perceived amount learned in the course (Table 5). Coincidently, these were also the only two sites that required students to use real microscopes to identify histological images by themselves (Table 3). Further, rating of the amount learned in the course was found to be highly correlated with overall satisfaction of the course (Fig. 3; Pearson correlation coefficient = 0.78; P < 0.01). Accordingly, allowing students to identify images by themselves appeared to be the best strategy, which not only could help students achieve better learning outcome but also could improve their overall course satisfaction. This was the strategy uniquely used by the HCD2.

Table 5. Students' satisfaction of the Web-based/electronic resources and the amount learned in the course*
  • *

    Survey conducted in autumn 2003. Rating score: 1, poor; 2, fair; 3, good; 4, very good; 5, excellent.

Average rating of Web-based or electronic resources4.
Average rating of the amount learned in the course3.
Number of responses81302175


Histology is a morphologic science in which the structure of the cells, tissues, and organs of the body are examined with a microscope (Cotter,2001). During the process of learning in histology, students first need to develop the ability to classify patterns (or morphological features) as examples of a category (e.g., simple columnar epithelium) based on stored memory representations of perceived patterns (Marks and Dulaney,1998). Once they master the basic patterns of each tissue category, they are expected to apply that knowledge to recognize separate examples in each of these categories (e.g., different tissues) as well as to detect novel examples in categories (e.g., X type of cell/tissue in Y organ) and transformations of examples (e.g., X type of tissue cut from different angle or seen under different magnification). Therefore, students use the microscope to detect and discriminate elements (or symbols) that they expect to appear at a particular location within visual display so that they can identify the pattern with active participation of their prior knowledge.

However, experts in a field usually notice different features of an array, represent their knowledge differently, and use different retrieval plans from novices (Chi et al.,1988). When a student is initially exposed to something new (especially if it only exists in a microworld), identifying it requires considerable attention and effort; but as the student becomes more familiar with the features and patterns, identification requires less attention and effort; when the student constantly practices to detect the same features and patterns and such features and patterns are always presented in the same way, recognition and identification can become automatic, and the student can become an expert (Schneider and Shiffrin,1977; Shiffrin and Schneider,1977). Because of the differences of experience between experts and novices in a domain, it is necessary for the teacher (i.e., the expert) to create “scaffolds” that help students (i.e., the novices) notice appropriate features and dimensions of the content and thereby assist students in acquiring a knowledge base that subsequently will support effective observation, representation, and retrieval (Goldman,1999).

Harris et al. (2001) carried out a formative evaluation of a virtual microscope laboratory Web site, which consisted of a syllabus, a table of content, virtual slides, gross images for gross microscopic correlation, and links to additional histology Web sites. They found that students gave significantly higher ratings to the accessibility and efficiency of the virtual microscope laboratory vs. the regular microscope laboratory, except for two weakness of the virtual laboratory: no “arrow” was available to help inexperienced students locate structures when the instructor was absent, and no mechanism was available to help students assess their progress. Obviously, there was a lack of effective scaffolds in this CAI tool. Cotter (2001) integrated computer applications into a histology laboratory during the first half of the course. It was concluded that by replacing some microscope exercises with more efficient computer applications, the histology course accommodated curricular change by both reducing contact time and continuing to offer valuable microscope laboratories for most of the organ systems of the body. A study done by Heidger et al. (2002) indicated that faculty teaching time has been reduced by almost half with the computer-integrated approach. In their study, students also commented that the virtual microscope was very time-efficient, though some of them indicated it would be better if they could zoom in on the images. Brisbourne et al. (2002) experimented with the use of animations to teach histology. It was found in an informal survey that first-year medical students strongly preferred the Web-based course featuring animation to the traditional laboratory-lecture format. As their program made extensive use of animations to help students master basic histological principles and integrate histological structure with physiological function, they found the types of questions students asked about histological content were increasing in sophistication. Thus, they concluded that such an increased level of sophistication may indicate increased student understanding of key principles and integration of information, as well as the development of better mental models. Similar to some of the cases above, a major reason for UW-Seattle to experiment with replacing traditional microscopic laboratory sessions with CAI was to achieve efficiency. Every year, nearly 100 students were expected to enroll in the histology class. A few students would decide to “challenge out” of the class by taking an examination before the school started. Even so, there would still be about 90 students in the class, and only 3 instructors were responsible for taking care of the students. As a result, students may have to wait for quite a long time when they needed assistance with the microscopic images in the real laboratory. This problem could only be solved if each student (or a small group of students) was assigned a tutor. If it was not feasible to do so, then a tutorial-type CAI program could be used to supplement lectures or laboratory sections and also enable students to work at their own pace. In this study, the goal of efficiency should have been met given the fact that nearly 90% of the UW-Seattle students used at least one CAI program for their own study and more than 80% of the UW-Seattle students rated the electronic learning resources (including CDs and Web sites) very good or excellent. As a student put it, “I found it a much better use of my time to just view the lab figures on my own.” Another student said, “I found the CDs to be great for learning as well as for quick review. I really appreciated having the information available to me at home rather than locked away in a slide box somewhere.”

Though preferred by many students at different schools because of the efficiency electronic learning tools can provide, most histology CAI tools are limited to static images that do not functionally resemble the microscope; for the most part, they are not interactive. They do not allow students to explore characteristics of the images by moving the tissues and changing magnification, to identify structures independently, and to discover relationships (Harris et al.,2001). Therefore, the perception that the best way to read histological slides was using a microscope usually made it difficult for faculty to entertain seriously replacing microscope exercises with computer applications. Not being able to explore, identify, and discover fully and independently can be a serious limitation. The HCD2 did not have this weakness. Students who used this CAI tool, though unaware of its advantage, may have in fact benefited from the scaffolds designed to help them achieve better learning effectiveness. Further, in contrast to the conclusion of Elizondo-Omaña et al. (2004), there was no evidence in this study that the number of hours students spent studying via the electronic material would influence their performance on the final examination. The finding of this study implied that a CAI tool designed with sound pedagogy would not only help students learn efficiently, but also help faculty overcome the difficulty of not teaching effectively due to reduced time allocation and the frequent lack of personnel support in the laboratory session. Accordingly, designers of image-based medical CAI should avoid putting students in the role of passive information receivers. A CAI tool, as well as other Internet technology, should take advantage of these tools' ability to provide for active participation in the learning process. Doing so could truly move instruction forward by enabling students to experience more meaningful learning.

Finally, in this study, students were assessed only through their performance in the multiple-choice examination. Yet as Oswald (1999) put it, the advantage of healthcare outcomes as a measure of effect should be the major reason for undertaking medical education. Therefore, it would be advantageous to develop a longitudinal study to measure students' performance in subsequent courses and their behavior that relates more directly to healthcare outcomes. In this way, the relationships between findings of this study and variables measured subsequently can be evaluated and better understood.