Tutor Messaging and Its Effectiveness in Encouraging Student Participation on Computer Conferences

Allan C. Tagg, Julie A. Dickinson

VOL. 10, No. 2, 33-55

Abstract

Following student requests for more frequent tutor intervention in academic computer conferences, research was conducted to discover whether different patterns of tutor intervention did in fact result in greater student activity, measured both by the number of student contributions and by instances of dialogue between tutor and student. The conclusion reached was that certain tutor behaviour could encourage participation where it might otherwise not take place, but that it was not a necessary precursor to student activity in all circumstances.

Résumé

En réponse aux demandes des étudiants souhaitant une intervention plus fréquente des tuteurs aux téléconférences informatisées, une recherche a été effectuée afin de déterminer si divers modèles d'interventions des tuteurs encourageaient effectivement la participation étudiante. Le niveau de participation a été évalué en fonction du nombre de contributions des étudiants et d'échanges entre tuteur et étudiants. Les conclusions ont révélé que certains comportements de la part des tuteurs pouvaient en effet encourager la participation des étudiants, qui autrement ne prendraient pas part au dialogue, mais ce changement de comportement de la part du tuteur n'était toutefois pas l'unique facteur.

Introduction

Computer Conferencing (CC)-a computer messaging system that allows geographically dispersed members of a group to communicate with one another by way of text-based messages-has been used as part of a distance learning program within the Department of Organizational Psychology, Birkbeck College, University of London, since 1989. Network Birkbeck serves students engaged in two-year, part-time MSc in Occupational Psychology and Organizational Behaviour and, more recently, a Diploma in Career Counselling and Consultancy. Students have generally expressed a high level of satisfaction with CC as a mode of tuition (Hartley, Dickinson, Noakes, & Tagg, 1994), although an area of concern that has repeatedly been raised involves tutor input and feedback. When the 1991/93 MSc cohort were asked at the end of their first year to list the three most negative aspects of computer conferencing, 36% of the 48 students who responded included some form of tutor-related problem within their list. These mostly concerned lack of input (68% of tutor-related responses specifically cited this), slow responses, or lack of encouragement from tutors. The following year 37% of the 27 students from the 1992/94 cohort who responded also specifically cited a tutor-related problem amongst the three most negative aspects of their computer conferencing experience.

Such concerns principally arise as a result of the “communication anxiety” (Feenberg, 1989, p. 23) produced by the asynchronous nature of the medium, which can mean varying delays (sometimes of a few days duration) between a message being sent to the system and a reply to that message being received. This difficulty of assessing how one’s contribution has been received-in a medium where immediate feedback by way of facial expression, verbal response, or physical reaction is entirely absent- has been remarked on by a number of writers on computer-mediated communication (e.g., Hesse, Werner, & Altman, 1988; McGrath, 1990), whilst Hiltz and Turoff (1993) cite lack of leadership (in other words, someone capable of maintaining momentum and coherence within CC discussions) as one of the main reasons computer conference discussions fail. The principal role of tutors within the conferencing program is therefore to support academic discussions by providing both guidance and encouragement to students where necessary, as well as to supply additional theoretical material when the situation demands. The need for fast and regular tutor feedback has also been noted by commentators in the field of traditional distance education (e.g., Holmberg, 1993; Taylor et al., 1993).

Yet if a lack of tutor activity causes anxiety amongst students and can act to the detriment of the discussion, is there any counter evidence that a meticulously adhered-to policy of responding to virtually every student’s contribution brings concomitant rewards in terms of increased messaging activity? Work in this area has previously been conducted by Ahern, Peck, and Laycock (1992), who studied the effects of tutor behaviour on a non-distance based undergraduate group, but within a controlled environment: tutors adopted either a question-only, statement-only, or conversational style towards student groups, each of which were exposed to only one type of tutor behaviour. Results showed that a conversational style was most effective in encouraging student participation and more complex patterns of interaction. The present study differed in a number of respects: firstly, the students were postgraduate distance learners. Secondly, tutors received no instruction on how to conduct their interventions, being free to respond in the style with which they felt most comfortable. Thirdly, the modularized pattern of teaching meant a single group of students were exposed to a number of varying tutor styles throughout the year. Specifically, the research sought to test the following hypotheses:

H1: More tutor messages will result in increased student activity.

H2: Prompt tutor responses will result in increased student activity.

H3: Greater encouragement from tutors will result in increased student activity.

The Structure of the Birkbeck Course

Students were divided into conferencing tutor groups of approximately 12 to 15, plus tutors. As is common in other courses, emphasis was placed on students being able to meet, face-to-face, prior to the course commencing (Gray, 1989; Levinson, 1989). Throughout the year they received printed Course Module booklets, one for each of six modules. Each booklet contained a series of chapters, which provided preliminary reading on the subject under review, plus brief self-test questions. In addition, students also received a Course Handbook, which, along with general information concerning both the course and Department, contained a scheduled list of exercises. These exercises related to the material covered in the Course Modules but required a more substantial response than the self-test questions. In acknowledgement of the deeper level of understanding required, each exercise guided students towards recommended reading in the form of articles and book chapters. The Course Handbook exercises were therefore designed to test students’ ability to consolidate and apply to specific situations the materials contained within the printed modules and recommended readings. The scheduling of the six modules and their related exercises were as follows:

The link between this schedule and computer conferencing was achieved by identifying each exercise in the Course Handbook by a short, unique code number, which indicated the module, number-in-sequence of that exercise, and a one-word prompt as to the subject. For example, exercise number two in Introduction to Occupational Psychology, on the subject of job training courses, was allocated the code op2training. Tutor groups’ conferencing topics would be given corresponding codes, with the result that each exercise in the printed Course Handbook was represented by a computer conferencing topic within each tutor group. Students would proceed by preparing their answer to the set question and then placing that answer into the relevant computer conferencing topic for others to read and respond to.

In fact, the system of adding messages was not quite as ad hoc as the foregoing description suggests. For various reasons (Tagg, 1994), the initial contributor to each topic was chosen beforehand by the group members themselves; once the discussion had been opened, others were free to contribute as they wished, with the tutor providing additional guidance and information as he or she saw fit. Each exercise topic ran for two weeks with a scheduled start date, and although there was no formal assessment procedure for contributions, students were expected to contribute regularly. At the end of the two-week period, another student would summarize the topic and formally close it, but discussion on popular topics would often continue informally beyond the closing date. For most of the year, students were required to deal with two exercises within any one fortnightly period; topics did not therefore run consecutively.

Group tutors (one per group) were present throughout the year and additionally bore responsibility for tutoring their group’s Introduction to Occupational Psychology module. Each of the subsequent five modules was supervised by a different tutor, hence students were taught by six different tutors during the course of the year. This structure does not mean all five tutor groups necessarily shared the same tutor for each of the five modules following OP: module PC could, for instance, be taught by tutor A in three groups and tutor B in the remaining two. In fact, only one module (PR) was taught by the same tutor across all five groups.

Methodology

A combination of quantitative and qualitative measures was adopted in keeping with those of other researchers in the areas of both traditional and CC-mediated distance learning (e.g., Harasim, 1987; Mason, 1991; Morgan, 1995). Quantitative analysis served to identify one measure of student activity: the number of student messages contributed. Since the conference content related to academic exercises, and steps were taken (see below) to eliminate irrelevant material, the quantity of student contributions provided a useful indication of relative levels of activity (Quinn, Mehan, Levin, & Black, 1983; Riedl, 1989). If a module was found where student messaging activity in one conference was higher than in the others, then, since the modules covered the same material and took place at the same time of year, further qualitative investigation might reveal patterns within tutor behaviour that might account for this increased activity. Clearly, any direct line of causality would be hard to establish, nevertheless, one further indication that tutor intervention was effective in promoting student activity would be instances of students referring to a tutor’s message and, more importantly, occasions on which a student and tutor entered into a dialogue comprising more than a simple initiation-response sequence (Ahern et al., 1992).

Although quantitative data can be obtained easily from CC-generated statistics, it was important to scrutinize the message transcripts thoroughly for the following reasons:

The data from five teaching conferences were available for analysis; however, the aim of the research limited the extent to which all data could be used. The comparison of tutors’ styles of intervention required as many other factors as possible to be held constant, and one factor that could have had a significant effect on levels of interaction was a diminishing number of students in the group as the year progressed. Tutor groups that suffered a number of student withdrawals, particularly later in the year when those withdrawing had previously made significant contributions, were therefore deemed unsuitable for intra-group comparison between modules. Elimination of these tutor groups left two: 2Group and 3Group. For simplification, quantitative data from 2Group and 3Group alone have been included in this paper, although from time to time, and in order to emphasize extreme or unusual behaviour in a particular module, reference will be made to its characteristics of behaviour in all five tutor groups.

As the principal aim of the study was to assess tutor behaviour, and a separate tutor was solely responsible for each course module taught to a specific tutor group, it was decided to treat course modules-rather than the exercise topics within them-as the basic units of analysis. However, not all modules contained the same number of topics: whereas modules OS, PR, IO, and PC comprised four topics each, module OP comprised eight topics, and module OC, ten. Clearly, a simple message count would be heavily skewed in favour of modules OP and OC, which had a greater number of topics. For this reason, all the data relating to quantitative messaging activity, except where indicated otherwise, have been adjusted to redress this imbalance: message and commenting totals for module OP have been adjusted by a factor of 0.5 and those for module OC by a factor of 0.4. Where the results deviate from this policy (i.e., when discussing tutor-student interaction), due note will be made in the text. The overall (adjusted) number of student messages totalled 290 for 2Group and 347 for 3Group. Tutor messages totalled 141 for 2Group and 110 for 3Group. Where the data permitted, chi-squared tests were employed to establish statistically significant differences. Results will be shown where appropriate within the text.

Qualitative content analysis was conducted to establish tutors’ styles of messaging. Henri (1991) proposed a content analysis schema for identifying “deep-level” learning processes, although this schema was of course applicable to student rather than tutor messages (which are the subject here). Nor is the substantial work on small groups of much help in a medium in which many defining characteristics of a face-to-face group are simply absent (e.g., Bramley, 1979; Harnack, Fest, & Jones, 1977; Shaw, 1981). As Vertecchi (1993) notes of computer conferencing: “very little of classical educational methodology can be applied in an instructional situation . . . where the spatio-temporal contiguity between teachers and students does not exist” (p. 126). In view of this paucity of research, it is perhaps understandable that McConnell (1994) recently argued for a “grounded” theory approach (Glaser & Strauss, 1967; Strauss & Corbin, 1990), in which analysis categories were developed from the data themselves, and this is indeed the approach that has been adopted here. By developing an initial set of categories, applying them to the data, and modifying them as new circumstances arose, an iterative process of refinement resulted in the set of categories that is presented at the beginning of the section on qualitative analysis below. By applying these categories to each tutor message, it became possible to discover where a definitive style of tutor messaging existed, for example, where messages predominantly consisted of an initial acknowledgement to the student followed by guidance on their contribution. This approach also avoids the problem of redundant or non-validated cases, noted by Howell-Richardson and Mellar (1996) in their work on course design.

A similar exercise was attempted in respect of student messages, with the aim of measuring the quality of discussion, but it was abandoned when it became clear that students’ contributions were so diverse, both in their composition and content, that no meaningful data could be assembled within the time frame available.

Results-Quantitative Data

Table 1
Total (adjusted) Numbers of Student/Tutor Messages

Message totals alone provide a poor indicator of tutor activity. For instance, the tutor responsible for module PR in both 2Group and 3Group appears to have a very low rate of input. His particular messaging style, however, dictated a small number of lengthy messages, each of which incorporated responses to a number of students. In contrast, another tutor might address each individual by a single message; hence, the appearance of less tutor activity in PR is illusory, at least in terms of the number of students who received a response from the tutor.

In order to obtain a clearer picture of individual student activity within modules, it was necessary to ascertain how many students were active. In this context, an active student was defined as being one who contributed at least one message per module. Table 2 shows these numbers, along with the average number of messages generated per student per module:

Table 2
Number of Active Students/Average (adjusted) Messages per Student per Module

A noteworthy point here is the generally low level of student messaging. Bearing in mind that the figures have been adjusted to give an equivalent of four topics per module, almost half the modules fail to achieve an average of one message per student per topic, with only one topic (3Group/PR) managing to achieve an average of two messages per student per topic. Tables 3a and 3b indicate the levels of commenting that took place within the two conferences, commenting being every occasion on which a contributor acknowledged the presence of another participant within the discussion. Whilst the existence of an acknowledgement says nothing about the quality of interaction, the data nevertheless do provide an indication of the relative levels of interaction taking place within the conference modules, a significant factor in a medium whose main benefit is held to be the opportunity it provides for discussion between members of a distance learning course.

Table 3b: Levels of Commenting within Tutor Groups

In addition to the total numbers of tutor messages within each module, tutor messages were analysed on a module-by-module basis according to their average length (measured in terms of words), the average time taken to respond to a student message (expressed in days), and the percentage of student messages that were responded to. The results are shown in Tables 4a and 4b:

Table 4a: Quantitative Analysis of Tutor Messages

Table 4b: Quantitative Analysis of Tutor Messages 3Group

Analysis of the quantitative results made it possible to narrow the focus of the qualitative analysis to four modules: 2Group modules PC and OC and 3Group modules PC and OC (from here on referred to as 2PC, 2OC, 3PC, and 3OC for the sake of brevity). The reason they were chosen was the contrasting behaviour both within and between the two tutor groups with regard to these modules.

2Group’s PC module achieved the highest ratio of tutor to student messages: 24 tutor messages to 27 student messages; moreover, 23 tutor-to-student comments indicate that virtually every message incorporated a response to a student. In fact, 2PC’s tutor succeeded in responding to 78% of the students’ messages, the highest response rate overall in the two tutor groups. And yet student activity in the module was amongst the lowest, both within 2Group and within the five tutor groups as a whole: 27 student messages was the second-lowest figure recorded for module PC within any of the five tutor groups, as was the average of messages-per-student which, at 2.7, represented less than one message per exercise topic. There were only two instances of students commenting on tutor messages within the four exercise topics (one response for every 12 tutor messages).

In contrast, 3Group exhibited a high level of activity in module PC. The 26 tutor messages in this module were accompanied by 66 student messages, the highest student messaging level for module PC of the five tutor groups. At 5.9, the average number of messages contributed by each student was also the highest for module PC. The 3PC tutor achieved the second-highest number of tutor-to-student responses in 3Group, responding to almost half (42%) of all student messages. The number of students’ comments on tutor messages was the highest in any of the 3Group modules. Noteworthy too was the difference in response times taken by the two PC tutors: 3Group’s tutor took only about half the time to respond to students than the tutor in 2Group did (2.9 and 5.6 days respectively).

Moving on to the OC modules, at first sight there seems little to distinguish 2Group and 3Group. Certainly, 2OC was the only OC module that showed an increase in activity over the previous module, but then, as noted above, activity on 2PC was unusually low. It was also true that the number of student messages in 2OC was the highest at 44; however, when the number of active students was taken into account, both 2Group and 3Group achieved a similar average of messages-per-student (3.4 and 3.3 respectively). Where activity differed significantly was in its consistency throughout the module. Messaging activity amongst students in 3Group fell off dramatically during the second half of the module. Tables 5a and 5b show the unadjusted message totals for the 10 topics in 2 and 3Group module OC. It was not, however, the behaviour within 3Group that was exceptional: module OC occupied the first six weeks of the final, Summer term, and activity was generally patchy throughout the five tutor groups. Rather, the exception was in 2Group; here, activity increased in the 8th and 9th topics and was evident in all topics barring the last.

Table 5a: Message Totals (unadjusted) for Module OC - 2Group

Table 5b: Message Totals (unadjusted) for Module OC - 3Group

Tutor behaviour within the two groups was mainly differentiated in terms of the numbers of tutor messages: 2OC’s tutor contributed 28 messages against 44 student messages, a ratio of almost 2:3, whereas the tutor in 3OC contributed only 11 messages, compared with 33 student messages, a ratio of 1:3. Elsewhere, average response times between the two OC tutors were similar: 2.0 days in 2OC and 2.2 days in 3OC, whilst the percentage of student messages responded to, although higher in 2OC (54%, compared with 33% in 3OC), did not represent a statistically significant difference (c 2 = 2.64, >0.10).

Summary of Quantitative Analysis of 2group and 3group Modules PC and OC

The results thus far illustrate the difficulties associated with attempting to identify contributory factors behind differing levels of student activity on the basis of quantitative data alone. While there was virtually no difference in the numbers of messages contributed by the tutors in 2PC and 3PC, 2PC’s tutor responded to a significantly higher percentage of student messages (78% compared with 42%; c 2 = 8.24, <0.01). But this response did not encourage student activity. The total number of student contributions in 2PC was half that for 3PC. 2PC’s tutor did, on the other hand, take twice as long to respond to students. This finding might have suggested that it was the delay in time taken to respond-rather than numbers of either tutor messages or responses to students-that accounted for the variation in activity were it not for the evidence presented by the OC modules.

Here the average time taken by the two tutors to respond to student contributions was almost identical. Had the hypothesized link between response times and student activity been correct, one would have expected to have seen a longer response time in respect to module 3OC. 2OC’s tutor did, however, contribute a greater number of messages than the tutor in 3OC and did respond to a higher (although not significantly higher) number of students’ messages. Both might have suggested a possible explanation of increased activity had it not been for 2PC’s remarkably high proportion of tutor-to-student responses resulting in one of the lowest levels of student messaging, whilst 3PC’s tutor achieved one of the highest, despite contributing only two more messages than the tutor in 2PC.

The quantitative stage of analysis had identified two modules in which student activity differed according to the tutor group. 2Group’s activity in module PC represented the lowest of all five tutor groups, whereas they were the only group to achieve a consistently high level of activity throughout the last module, OC. 3Group’s activity in OC was, on the other hand, typically low for that particular module, whereas their activity in module PC was the highest for that module. Qualitative analysis of tutor messages was therefore undertaken to ascertain whether the tutors in the “successful” modules-2OC and 3PC-exhibited similar patterns of behaviour that were not evident in that of the 2PC and 3OC module tutors.

Results-Qualitative Analysis

Tutor message texts were broken down into discrete “moves,” a move being a fragment of text or “unit of meaning” (Henri, 1991) conforming to one of the following analytical categories:

Organizational Move-Group Directed Only

This concerns questions or suggestions about how the group is to be organized or the task is to be approached.

Socio-Affective Move-Individual or Group Directed

Socio-affective moves are not concerned with the course material but with the general ambience of the conference.

Acknowledging Move-Individual or Group Directed

This acknowledges the contribution(s) of one or more participants, usually through an expression of gratitude, although it may simply involve naming the contributor(s).

Reinforcing Move-Individual or Group Directed

This usually begins a message and is an evaluative acknowledgement of a contribution (i.e., unlike an Acknowledgement it implies a judgemental appraisal).

Guiding Move (With External Reference)-Individual or Group Directed

This frequently follows on directly from either an acknowledgement or reinforcement of input. It builds on points made by the contributor(s), citing one or more academic references in support.

Guiding Move (Without External Reference)-Individual Orgroup Directed

This frequently follows on directly from either an acknowledgement or reinforcement of input. It builds on points made by the contributor(s), but either makes no reference to external sources or, where it does, repeats sources used by the original contributor.

Requesting Move-Individual or Group Directed

This frequently follows on from guidance and is usually expressed in the form of a question. Whereas Guidance only implies that the discussion is not complete, requesting moves explicitly invite the contributor(s) to elaborate either on their previous point or on that made by the tutor.

Summarizing Move-Group Directed Only

This will always appear at or near the end of a topic, and it will most closely resemble a Guiding move. The implication will, however, be that nothing more is expected.

Combining analysis of tutor messages, according to these categories, with factors such as the frequency and response levels attained by tutors made it possible to identify variations in tutoring styles between the modules. In turn, this identification, when combined with an analysis of student activity within modules, both in terms of message numbers and levels of interaction, provided an indication about which styles of tutoring seemed most effective in promoting student activity.

2group Module PC

Style of Tutor Messages

It was difficult to identify a specific pattern of tutor messaging in module 2PC, save that the tutor responded diligently to each contributor. Even in the less active topics, this style was maintained, thereby accounting for the high figure of 78% of student messages being responded to. With the exception of the third topic of four (which alone accounted for half the total number of messages contributed by students in the whole module), activity-whether by the students or tutor-was, nevertheless, sparse. In the first three topics, the tutor took a considerable time to make an initial contribution in response to the student introductory message (23, 16, and 16 days respectively). This lateness contributed to the high average delay in responding to student contributions of 5.6 days (average across the 12 2Group and 3Group modules = 3.0 days).

Tutor messages tended to be fairly short, with 67% comprising two moves or less. Thirty per cent of 2PC’s tutor messages began with an acknowledgement or individual reinforcing move followed immediately by individual guidance, although half of these also featured an additional move, usually in the form of a question addressed to the group as a whole (in total 21% of tutor messages posed questions to the group, with there being no instances of individually addressed questions). Despite the high numbers of student messages that received an individually addressed response from the tutor, such responses tended to be clustered-for example, in one instance the tutor addressed nine people by way of nine consecutive messages-rather than dispersed throughout the topics (cf., 2OC and 3PC). Seventeen per cent of the tutor’s messages contained organizational moves concerned with establishing why so few people were active. Overall, there was little evidence of a consistent pattern emerging in the style of tutor messages. Sixty-eight per cent of all moves were addressed to individuals, lower than 2OC (79%) and 3PC (86%), although the difference only achieved significance in the latter case (2OC: c 2 = 1.85, >0.10; 3PC: c 2 = 4.46, <0.05). The apparent contradiction between the high number of student messages responded to and the relatively low proportion of individually addressed comments can be explained by the tutor’s tendency to incorporate group-directed comments within messages that began by addressing an individual.

Activity Within the Topics

With the exception of the third topic, there was little activity. The fourth topic comprised two messages, one from a student and one from the tutor, while the first and second had four and five students respectively contributing four and seven messages. Topics were characterized by the late initial contribution by the tutor. Nor, in topics one and two, was there much activity amongst the students prior to the tutor’s first intervention: the 23 and 16 days before the tutor intervention produced only four and three student contributions respectively, including the introductory message. There were only two instances of a student acknowledging a contribution by the tutor: the first in topic two, which asked for-and received-clarification on a reference provided earlier by the tutor, represented the only instance within module 2PC of a three-part dialogue. The second occurred in topic three and represented the sole student response to a total of nine consecutive messages contributed by the tutor, despite each of these having individually addressed previous students’ contributions.

2group Module OC

Style of Tutor Messages

The tutor responded to 54% of all student messages (average across the twelve 2 and 3 Group modules = 46%), and the average response time was 2.0 days (average across the twelve modules = 3.0 days). Initial responses to student contributions never took more than three days from the opening contribution, and in four of the ten topics an introductory message by the tutor was the first to appear.

A definitive style of tutor messaging was far more identifiable in 2OC than in 2PC. Messages were overwhelmingly short and tended to adhere to an identifiable pattern: 93% comprised two or fewer moves, with 45% commencing with an acknowledgement or individual reinforcing move followed immediately by individual guidance. Unlike 2PC, however, these acknowledgement/guidance messages contained only those two moves. Combined with the fact that they routinely appeared about two days after the message they were responding to (and achieved a response rate of over 50%), this style created an impression that the tutor was giving constant attention, identifying the most salient point of a student’s contribution, and swiftly offering a brief comment on it. Individual acknowledgement was commonplace, with 66% of all 2OC tutor messages containing such a move. Unusually amongst the modules, eight out of the ten topics contained a tutor summary that commented on and clarified the student summary. Few messages (14%) contained individual reinforcing moves, and equally few (16%) contained either individual or general requests. Seventy-nine per cent of all moves were addressed to individuals.

Activity Within the Topics

Examples of interaction were evident throughout the topics. What distinguished 2OC from 2PC was, however, the level of interaction, not between students, but between students and tutor. In 2PC there were only two instances of any of the 24 tutor messages being acknowledged by a student, and only one of a three-part exchange between tutor and student. In 2OC, by contrast, there were 16 instances of three-part or more sequences between the tutor and a student. In other words, there were 16 instances of dialogue as opposed to a simple, two-part initiation-response sequence (which was, of course, in addition to dialogue between students themselves). Five of these three-part sequences were initiated by the student, and 11 by the tutor. One of the student-initiated sequences was four-part (i.e., student-tutor-student-tutor) and one five-part. Despite there being few examples of direct questions being aimed at students, the style of tutor messaging seemed particularly apposite in engaging individual students in dialogue, perhaps partly because the prompt responses ensured the theme was still fresh in the student’s mind. Students’ contributions were consistently shorter and seemed more focused than those in 2PC, with less tendency to cover a number of points within a single message.

3group Module PC

Style of Tutor Messages

The tutor responded to 42% of messages contributed by students, an average response rate (overall average = 46%), as indeed was the tutor’s response time of 2.9 days (overall average = 3.0 days). Initial responses to student contributions occurred much faster than in 2PC, although a day or two slower than 2OC, taking between four and five days for each of the topics. In terms of message complexity, 3PC’s tutor messages were similar to those of 2PC, with 63% comprising two moves or less. They did not adhere to the consistently short, two-move pattern of the kind favoured by 2OC’s tutor. They did, however, resemble the message style of 2OC in that 44% commenced with either an acknowledgement or individual reinforcing move followed immediately by individual guidance, but they were dispersed regularly throughout the discussion rather than being clustered together. Atypically, almost one-third (29%) contained socio-affective comments, with 22% consisting exclusively of such material. Although 56% of all the messages contained an instance of individual guidance, very few (8%) contained individual guidance with reference to supporting literature. Eighty-six per cent of all moves were addressed to individuals.

Activity Within the Topics

Activity throughout the topics was consistently high with, like 2OC, a number of instances of dialogue as opposed to simple initiation-response exchanges. There were nine instances of three-part or more exchanges (a figure more remarkable for the fact it occurred within four topics, whereas the 16 instances within 2OC occurred over 10 topics). Five of these were initiated by a student, and four by the tutor. Three of the student-initiated exchanges were four-part (i.e., student-tutor-student-tutor), and one was five-part. In general, the exchanges were more “chatty” than those for the other modules, partly as a result of the high socio-affective content of the tutor’s messages, which seemed particularly effective in engaging the students: module 3PC had the highest number of student-to-tutor comments. 3Group Module OC

Style of Tutor Messages

The tutor responded to 33% of students’ messages, below the overall average of 46%, but not significantly lower than the response rate of the tutor in 3PC. The response time was, at 2.2 days, below the overall average of 3.0 days. Moreover, in only one case did the initial tutor response appear more than four days after the first student message. This latter behaviour bore distinct similarity with 3PC’s tutor.

Although the tutor’s messages were not long (the average length was less than those in 3PC), they were more complex: almost one-third (29%) contained four moves or more, compared with only 14% in 3PC. It was therefore difficult to identify a consistent style of messaging (only 14% of 3OC’s messages adhered to the pattern of an individual acknowledgement or reinforcement followed immediately by individual guidance, in contrast to 3PC [44%] [c 2 = 4.68,

Activity Within the Topics

In terms of activity, the 10 topics could be divided evenly: the first five topics showed a relatively high level of activity, containing between 10 and 22 student messages; the last five a far lower level, with between nought and six messages per topic. Examples of dialogue only occurred during the first five topics: there were five instances of three-part or more exchanges, four initiated by the tutor and one by a student. Two of these occurred in the first topic. One each of the student- and tutor-initiated exchanges was four-part. Instances of dialogue were therefore not as commonplace as they were in either 2OC or 3PC, even in the first, most active half of the module, although student activity was relatively high there.

Summary of Results

The results of the qualitative tutor message analysis is summarized in Table 6 below:

Table 6: Summary of Tutor Message Characteristics

In terms of student messaging, modules 2OC and 3PC could be considered “successful” for various reasons: 2OC was the only OC module within five tutor groups that did not suffer a decline in activity during its course, coming as it did in the final, Summer term. The total number of student messages was the highest of any of the five OC modules, and there was considerable evidence of dialogue taking place between tutor and students. The highest number of student messages within the PC modules was found in 3PC, and it again contained a high number of instances of tutor-student dialogue. By contrast, student activity in module 2PC was the lowest of the PC modules, whilst module 3OC showed a decline in activity from midpoint. Reading these results in conjunction with the qualitative analysis of tutor messages and the quantitative data from the uniformly successful PR modules (the latter taught by the same tutor), the following findings emerged:

H1: More tutor messages will result in increased student activity.

There was no evidence that more tutor messages (with the attendant implication of a greater response rate to students) were either a necessary or a sufficient precondition to increased student activity. The PR modules illustrated they were not necessary: the tutor not only contributed the lowest number of messages within 2Group and 3Group but also responded to the lowest percentages of student messages; yet student activity within those modules was amongst the highest. Module 2PC illustrated they were not sufficient: 2PC’s tutor contributed only two fewer messages than the tutor responsible for the successful module 3PC, responding to a remarkably high 78% of student contributions (compared with 3PC’s 42%). Despite this rate, student activity in 2PC was amongst the lowest of any module.

H2: Prompt tutor responses will result in increased student activity.

Evidence did not support this hypothesis as either a necessary or a sufficient precondition to increased student activity. Once again, the PR modules showed it was not a necessary condition: the tutor delayed responding to student messages by an above-average 3.6 days (Tables 4a and 4b), yet the topics still exhibited high levels of student activity. Module 3OC, on the other hand, illustrated it was not sufficient: the tutor achieved a below-average response time of only 2.2 days, and yet activity dropped off half way through the module.

H3: Greater encouragement from tutors will result in increased student activity.

This hypothesis received support as a sufficient condition, although not as a necessary one, and, in any case, “encouragement” in this sense needs some elucidation. Tutor behaviour within the two most successful modules, 2OC and 3PC, shared the following six characteristics: a reasonably prompt response to the initial student introduction; rapid subsequent responses to student contributions, conversational moves directed predominantly towards individuals rather than the group; tutor responses that were dispersed throughout the discussions, rather than being clustered together; responses to about half the total number of messages contributed by students; and a pattern of individually addressed messages that followed an acknowledgement/guidance format. When situated within the con-ferencing discussions, this combination of factors created an impression of constant attention being paid to individual contributions. The tutors in 2OC and 3PC seemed to be continually “popping in and out” of topics, attending to the most recent contributions on an individual basis whilst offering guidance on specific aspects of a student’s contribution. Note that this impression was maintained even though both tutors did in fact only respond to about half the students’ contributions. In contrast, although the 2PC tutor responded to a high proportion of student messages, the responses were initially late and subsequently infrequent, which meant that the messages, although relatively numerous, were clustered together. In 3OC, on the other hand, although the tutor responded promptly both to the initial student introduction and subsequent messages, there was a tendency to address comments to the group rather than individuals along with messages that did not conform to the simple acknowledgement/guidance pattern followed by 2OC and 3PC.

Such “ideal-type” behaviour as that exhibited by the 2OC and 3PC tutors did not however constitute a necessary precondition for student activity: 2Group’s PC tutor took as long to make an initial contribution in the third topic of that module as in the preceding two topics-behaviour that could hardly be described as encouraging-and yet by that time students had already contributed eight messages, half the eventual total for topic three. Nor could the tutor messages that did appear be used to explain subsequent activity: 12 messages prompted only one student acknowledge-ment. Further evidence was provided by the PR modules, where the tutor contributed a few lengthy, complex messages, and yet student activity was consistently high.

Conclusion

The results suggest that student concerns over tutor behaviour were, to an extent, justified and that in certain circumstances tutor behaviour can indeed encourage more student activity. There is, however, no single, simple recipe for an “ideal” tutor style of messaging: quantity alone is no answer; nor is promptness. Encouragement seems to work but not in the most basic sense of an occasional “Well done!” Encouragement consists of students perceiving a continual tutor presence, evidenced both by the promptness and the frequency of responses. Encouragement consists of messages that address overwhelmingly individuals rather than the group in general. Finally, encouragement seems to come through students being able to rely on a pattern of short, succinct tutor messages that acknowledge an individual’s contribution and immediately follow with guidance. Typical examples, one each from 2OC and 3PC are:

Good to hear from you Tracie. I would add to your list:

5) Recognition of new ways and rewards of some sort for them 6) Role modelling and demonstrating superiority of the new ways. On your first comments, Lianne, about how some politics are formal, I think it is useful to distinguish between politics and control. I would call paperwork, rules, “how things are done around here” control mechanisms, but some of the decisions taken to change or install control mechanisms may be political-in terms of serving the interests of particular groups or individuals within the organization.

A pattern of frequent, prompt tutor responses that address individuals and offer guidance in a succinct and predictable manner seems, therefore, to be most effective in encouraging student activity. This finding is given sup-port by the work of Ahern et al. (1992), who found a conversational style to be most effective in promoting student activity in CC, and Hiltz (1988), who suggested tutors should respond explicitly to contributions by students.

One final note of caution with regard to any research into interaction on computer-mediated distance learning courses is that there are so many influences beyond the group itself (McConnell, 1994) that one will never know conclusively why activity-whether on the part of an individual or the group-increases or decreases at a particular time. The high levels of activity recorded in the PR module, for instance, could have resulted from a general interest in the subject matter or simply the fact that by the time it took place, students would have sorted out the acclimatization problems inherent within the first term of any course and, being refreshed by the Christmas break, were returning to the course with renewed enthusiasm. In contrast, module PC came when students were winding down at the end of the Spring term, whilst module OC occupied the weeks leading up to the summer examination (the “ebb and flow” pattern of conferencing behaviour [Levin, Kim, & Riel, 1990]). Taking the timing into consideration in conjunction with the findings, perhaps the overall conclusion should be that, whilst the ideal type tutor behaviour is not a necessary precursor to greater activity on the part of students, it may well be a sufficient one where a group of students might otherwise lack motivation for one reason or another.

Acknowledgments

The authors wish to acknowledge the assistance provided by the Birkbeck College Research Committee, who provided funding for this research.

Correspondence To:

Allan C. Tagg and Julie A. Dickinson
Department of Organizational Psychology
Birkbeck College
University of London
Malet St.
London WC1E 7HX
Tel: (+44) 171-631 6748
Fax: (+44) 171-631 6750
E-mail: a.tagg@org-psych.bbk.ac.uk

References

Ahern, T. C., Peck, K., & Laycock, M. (1992). The effects of teacher discourse in computer-mediated discussion. Journal of Educational Computing Research, 8(3), 291-309.

Black, S. D., Levin, J. A., Mehan, H., & Quinn, C. N. (1983). Real and non-real time interaction: Unraveling multiple threads of discourse. Discourse Processes, 6(1), 59-75.

Bramley, W. (1979). Group tutoring: Concepts and case studies. London: Kogan Page.

Feenberg, A. (1989). The written world: On the theory and practice of computer conferencing. In R. Mason & A. Kaye (Eds.), Mindweave: Communication, computers, and distance education (pp. 22-39). Oxford: Pergamon Press.

Glaser, B. G., & Strauss, A. L. (1967). The discovery of grounded theory: Strategies for qualitative research. New York: Aldine De Gruyter.

Gray, R. (1989). CMC for in-service training. In R. Mason & A. Kaye (Eds.), Mindweave: Communication, computers, and distance education (pp. 185-189). Oxford: Pergamon Press.

Harasim, L. (1987). Teaching and learning on-line: Issues in computer mediated graduate courses. Canadian Journal of Educational Communication, 16(2), 117-135.

Harnack, R. V., Fest, T. B., & Jones, B. S. (1977). Group discussion: Theory and practice (2nd ed.). New Jersey: Prentice-Hall.

Hartley, J., Dickinson, J., Noakes, J., & Tagg, A. (1994). The comparative evaluation of computer conferencing with other methods of teaching and learning (Final report to the Learning Methods Branch of the Employment Department ). London: Department of Organizational Psychology, Birkbeck College, University of London.

Henri, F. (1991). Computer conferencing and content analysis. In A. R. Kaye (Ed.), Collaborative learning through computer conferencing: The Najaden papers (pp. 117-136). Berlin: Springer-Verlag.

Hesse, B. W., Werner, C. M., & Altman, I. (1988). Temporal aspects of computer-mediated communication. Computers in Human Behaviour, 4, 147-165.

Hiltz, S. R. (1988). Teaching in a virtual classroom: Volume 2 of A virtual classroom on EIES (Final Evaluation Report No. 26). New Jersey: Institute of Technology.

Hiltz, S. R., & Turoff, M. (1993). The network nation: Human communication via computer. Cambridge, MA: MIT Press.

Holmberg, B. (1993). Key issues in distance education: An academic viewpoint. In K. Harry, M. John, & D. Keegan (Eds.), Distance education: New perspectives (pp. 330-341). London: Routledge.

Howell-Richardson, C., & Mellar, H. (1996). A methodology for the analysis of patterns of participation within computer mediated communication courses. Instructional Science, 24(1), 47-69.

Levin, J. A., Kim, H., & Riel, M. M. (1990). Analyzing instructional interactions on electronic message networks. In L. M. Harasim (Ed.), Online education: Perspectives on a new environment (pp. 185-213). New York: Praeger.

Levinson, P. (1989). Media relations: Integrating computer telecommunications with educational media. In R. Mason & A. Kaye (Eds.), Mindweave: Communication, computers, and distance education (pp. 40-49). Oxford: Pergamon Press.

Mason, R. (1991). Evaluation methodologies for computer conferencing applications. In A. R. Kaye (Ed.), Collaborative learning through computer conferencing: The Najaden papers (pp. 105-116). Berlin: Springer-Verlag.

McConnell, D. (1994). Implementing computer supported cooperative learning. London: Kogan Page.

McGrath, J. E. (1990). Time matters in groups. In J. Galegher, R. E. Kraut, & C. Egido (Eds.), Intellectual teamwork: Social and technological foundations of cooperative work (pp. 23-61). Hillsdale, NJ: Lawrence Erlbaum.

Morgan, A. R. (1995). Student learning and students’ experience. In F. Lockwood (Ed.), Open and distance learning today (pp. 55-66). London: Routledge.

Quinn, C. N., Mehan, H., Levin, J. A., & Black, S. D. (1983). Real education in nonreal time: The use of electronic message systems for instruction. Instructional Science, 11, 313-327.

Riedl, R. (1989). Patterns in computer-mediated discussions. In R. Mason & A. Kaye (Eds.), Mindweave: Communication, computers, and distance education (pp. 215-220). Oxford: Pergamon Press.

Shaw, M. E. (1981). Group dynamics: The psychology of small group behaviour (3rd ed.). New York: McGraw-Hill.

Strauss, A., & Corbin, J. (1990). Basics of qualitative research: Grounded theory procedures and techniques. Newbury Park, CA: Sage

Tagg, A. C. (1994). Leadership from within: Student moderation of computer conferences. The American Journal of Distance Education, 8(3), 40-50. Taylor, J. C., Barker, L. J., White, V. J., Gillard, G., Kaufman, D., Khan, A. M., &

Mezger, R. (1993). Student persistence in distance education: A cross-cultural, multi-national perspective. In K. Harry, M. John, & D. Keegan (Eds.), Distance education: New perspectives (pp. 77-93). London: Routledge.

Vertecchi, B. (1993 ). A two-level strategy for mastery of learning in distance education. In K. Harry, M. John, & D. Keegan (Eds.), Distance education: New perspectives (pp. 126-136). London: Routledge.


Allan Tagg has worked in the Department of Organizational Psychology, Birkbeck College, since Network Birkbeck (the computer-mediated distance learning programme) began in 1989. From assisting in the software development of the system, he has moved to researching social aspects of learning technologies, with particular regard to how use is made of computer-mediated learning.

Julie Dickinson has been a lecturer in the Department of Organizational Psychology at Birkbeck College since 1990. Her main area of research is distributive justice with particular application to pay differentials. She is also interested in social influences on knowledge development.