Tacit to explicit knowledge conversion: knowledge exchange protocols Richard T. Herschel Hamid Nemati and David Steiger The authors Richard T. Herschel is an Associate Professor in the Erivan K. Haub School of Business, St Joseph's University, Philadelphia, PA, USA. Hamid Nemati is an Assistant Professor at the University of North Carolina at Greensboro, NC, USA. David Steiger is an Associate Professor in the Maine Business School, University of Maine, Orono, Maine, USA. Keywords Knowledge management, Explicit knowledge, Tacit knowledge, Knowledge transfer Abstract In the knowledge management domain, the conversion of tacit knowledge to explicit knowledge is critical because it is a prerequisite to the knowledge amplification process wherein knowledge becomes part of an organization's knowledge network. In this article, knowledge exchange protocols are examined as a vehicle for improving the tacit to explicit knowledge conversion process. In an experiment testing the use of knowledge exchange protocols, it is learned that while structure may significantly improve the tacit to explicit knowledge conversion process, it also may matter how the structure is employed in this process. Electronic access The current issue and full text archive of this journal is available at http://www.emerald-library.com/ft Volume 5. Number 1. 2001. pp. 107±116 # MCB University Press. ISSN 1367-3270 107 Many practitioners and researchers in the knowledge management arena are aware that there are two forms of knowledge: explicit knowledge and tacit (implicit) knowledge. While both forms of knowledge are important, usually one is managed at the expense of the other (Earl and Scott, 1998; Hansen et al., 1999). However, while emphasizing one strategy over another may alleviate several operational knowledge management issues, it can also compromise knowledge creation activities. Indeed, to ignore the interaction of explicit and tacit knowledge is to potentially inhibit innovation vis-aá-vis the generation of new capabilities, products and services (Choo, 1998; Nonaka and Takeuchi, 1995). Converting tacit knowledge to explicit knowledge is often time consuming and problematic (Earl and Scott, 1998; Davenport and Prusak, 1998), especially to chief knowledge officers (CKOs) who are eager to show results from their knowledge management efforts. Many CKOs are advised that the best way to accomplish tacit to explicit knowledge conversion is via the use of ``convincing'' narratives, but little advice is given to them as to how to accomplish this process in an efficient yet effective manner. We believe that structuring mechanisms might be useful in improving the tacit to explicit knowledge conversion process. Specifically we examine the potential for knowledge exchange protocols to make this activity more productive. Knowledge exchange protocols A knowledge exchange protocol is a process that structures information exchange in such a way that the provider of the information and/or the recipient of the information can systematically present/recall information in a focused manner. One of the best examples of a knowledge exchange protocol is the SOAP protocol that is used in the medical community (see Table I). The SOAP protocol is used to structure and document situation-oriented, physician/patient clinical encounters. SOAP provides a consistent framework for:. structuring clinician-patient narratives;. understanding the clinician's thinking about perceived problems and issues;
Table I A sample SOAP knowledge exchange protocol Jim, Slim 111-2222 3/22/77 Subjective: (a brief narrative of the patient's expressed complaints) Objective: (a description of the specific activities used to better learn the true nature of the patient's situation) Assessment: (A decision about what is wrong with the patient) Plan: (a prescribed course of action for the patient to alleviate the problem(s) Recent onset of fever, diffuse aches, sore throat, clear rhinorrhea, and non-productive cough GEN: I11 appearing HEENT: Inflamed pharynx, TM's normal, PERLA, conjunctiva WNL, nasal congestion NECK: Tender anterior cervical adenopathy LUNGS: Clear to auscultation, equal BS HEART: RR & R, no murmur, gallop, or rub, not enlarged ABD: Soft BJE: Mild diffuse muscle tenderness NEURO: Alert, no meningeal signs SKIN: No rashes or lesions Influenza #487.1. Tylenol, bedrest, and fluids until fever free and feeling improvement in symptoms for 24-48 hours (Typically persons with influenza are ill for one week). learning about techniques and tests employed by the clinician in the knowledge creation process; and. sharing the clinician's reasons for actions taken to address patient issues. In other words, the SOAP process provides a consistent mechanism for documenting:. what the physician understands about the patient's situation (sense making activities);. how the physician closes gaps in his/her understanding about the patient's situation (knowledge creation); and. what actions the physician takes relative to treatments (decision making). Use of the SOAP protocol allows clinicians to accumulate knowledge about their patients over time. Moreover, by structuring the documentation of the patient-clinician dialogue and the clinician's thinking and actions, tacit knowledge is being externalized and can be shared with another clinician. Patel et al. (1999) have studied how experienced physicians have evolved specific heuristics that can be conveyed to medical trainees through effective clinical teaching. The medical students' learning of the SOAP protocol is one important component of this clinical teaching process. SOAP is used to 108 structure not only the clinician-patient encounter, but the resultant documentation in the patient's medical record (where the clinician's tacit knowledge is made explicit). This record can then be used to assess how clinician's tacit knowledge is deployed. For example, analyses of SOAPs by agencies such as the Joint Commission on the Accreditation of Hospitals (JCAH) have been used to gain an overall impression of the nature and quality of patient care at various medical institutions. Put in a knowledge management perspective, the SOAPs can be analyzed to gain knowledge about how clinician/patient perceptions are created and justified. Formal medical review processes use SOAP documentation to evaluate clinician assumptions underlying action, resulting in tacit knowledge coming into focal awareness and scrutiny. We believe that the use of protocols to structure the tacit to explicit knowledge conversion process is important in order to make the process productive yet efficient. Therefore, we suspect that these protocols, which we term knowledge exchange protocols (e.g. SOAP), are transferable structural mechanisms that can similarly be used to functionally facilitate other knowledge management efforts.
Exploring knowledge exchange protocol effects: an experiment Understanding is said to be able to occur if the information presented is relevant and somewhat familiar to the listener. While there is no standard way of measuring understanding, evidence of understanding is said to exist if the listener is able to effectively recall (make explicit) what he/she has learned from the teacher (Burns et al., 1990). Based on this premise, we asked the following question: If we created a rich narrative that was educational and instructive, would listeners' ability to articulate what they had learned depend on whether the narrative was highly structured (e.g. via a knowledge exchange protocol), on whether the recall process was highly structured (e.g. via a knowledge exchange protocol), or both? Theoretically, if a rich narrative consists of a main learning objective and some number of specific methods for achieving it, then the structuration literature (e.g. Bousfield et al., 1958; Cole et al., 1971; Horowitz, 1969) suggests that:. structuring knowledge presentation should improve the understanding of the main point of the narrative by the respondents;. structuring knowledge recall should improve the understanding of the main point by the respondents;. structuring knowledge presentation should increase the number of methods recalled by the respondents; and. structuring knowledge recall should increase the number of methods recalled by the respondents. Literature concerned with information systems development (e.g. Dennis et al., 1996; DeSanctis and Gallupe, 1987; DeSanctis et al., 1989; Nunamaker et al., 1991) also suggests that structuration should increase the verbalization of those queried. Hence, we are curious whether:. structuring knowledge presentation increases the number of comments that respondents made; and whether. structuring the knowledge recall format increases the number of comments that respondents made. Finally, we were curious to simply explore whether the impact of structuring an information exchange via a narrative would be verifiable by the unobtrusive assessment of an uninterested third party. That is, we were curious whether a subjective assessment of explicit understanding could tell us whether:. structuring knowledge presentation improves the apparent richness of the articulation of the knowledge exchanged; and whether. structuring knowledge recall improves the apparent richness of the understanding of the knowledge exchanged. These questions served as the basis for our research hypotheses, which are summarized in Table II. Research methodology To test our hypotheses about the impact of structuration (vis-aá-vis knowledge management protocols) on tacit to explicit knowledge conversion, we conducted an experiment at the University of North Carolina at Greensboro employing a total of 238 students enrolled in 13 sections of an ``Introduction to business computing''. Two videos were created where an expert in student advising provides advice to students about how to improve the efficiency and effectiveness of their class registration experience at the University's Bryan School of Business and Economics. The two films differed in how information was structured. The first contained a rich free form narrative, while the second film contained an equivalently rich narrative, except that it Table II Summary of hypotheses Hypotheses H1a H1b H2a H2b H3a H3b H4a H4b Structuring knowledge presentation improves the understanding of the main point by the respondents Structuring knowledge recall improves the understanding of the main point by the respondents Structuring knowledge presentation increases the number of methods recalled by the respondents Structuring knowledge recall increases the number of methods recalled by the respondents Structuring knowledge presentation increases the number of comments that respondents made Structuring knowledge recall increases the number of comments that respondents made Structuring knowledge presentation improves the richness of the understanding of the knowledge exchanged Structuring knowledge recall improves the richness of the understanding of the knowledge exchanged 109
adhered to a SOAP structure made explicit in the film. The expert delivering the rich narrative on both films was the Director of Student Services who has been directly involved with the Bryan School student registration process for over ten years. What is important to note is that the content of the films is identical. Each film includes:. the director's overview of student complaints voiced about the registration process over past years;. her review of tangible, documented evidence of student problems with the registration process;. her critical assessment of what students were doing to impede their own performance in the registration process, based upon their complaints and the evidence discussed above; and. the director's specific recommendations (including methods) that students should follow to improve their performance in the registration process. However, in the SOAP film, students were told that the presentation was divided into four structured parts: (1) subjective: student complaints; (2) objective: documentation of problems; (3) assessment: director's assessment of the situation; (4) plan: the director's advice on how to solve the problems. These headings appeared in the SOAP video at the appropriate place. Meanwhile, the free form version contained the same information, without any headings, though in a different, but logically equivalent, order. The films used in this experiment represent the sharing of tacit knowledge via a rich narrative by the Director of Student Services. She is sharing her own experience and judgement ± she is not reciting information that can readily be found in any procedures manual or student catalogue. Again, equivalent in content, the films differ only in the structure of the presentation ± SOAP versus free form. While the exchange of information in verbal discussion is never perfect (Dennis, 1996), every effort was made to ensure that the content expressed in the two films was equivalent. When the films were being created, the Director of Student Services had visual access to a list of items that had to be covered in each film. Scripting of dialogue 110 was not used, however, so as not to diminish the richness of the information content or the ``naturalness'' of the narrative. The films were professionally made and edited, involving numerous ``takes''. We were also cautious not to present too much information. As Alavi and Leidner (1999, p. 6) note:. ``because knowledge is personalized, in order for one person's knowledge to be useful to another individual, it must be communicated in such a manner as to be interpretable and accessible to the other individual; and. hoards of information are of little value: only that information which is actively processed in the mind of an individual through a process of reflection, enlightenment, and learning can be useful.'' Both films were designed to meet Weick's (1995, p. 127) requirements for a convincing narrative. That is, it the films and their contents were designed to be:...something that preserves plausibility and coherence, something that is reasonable and memorable, something that resonates with other people, something that can be constructed retrospectively but also can be used prospectively, something that captures both feeling and thought, something that allows for embellishment to fit current oddities... Each film, both approximately five minutes long, was shown in a classroom equipped with personal computers for each student. Once the film was over, students were immediately asked to recall what they had learned from the film by independently typing their comments at a Web site (where they could only see their own comments) designed specifically for this purpose. However, in this process, students were (unbeknown to them) divided into two groups. The first group of students (including both those who had seen the free form film and those who had seen the SOAP version) were asked a simple question ± what did you learn from the film? We call this the ``free form recall'' form. The second group of students (also including both those who had seen the free form film and those who had seen the SOAP version) were provided a list of questions that specifically asked for feedback based on the structured SOAP protocol discussed earlier. The number of participants is noted for each group in Table III.
Table III How students were assigned to the experimental groups Recall Video SOAP Free form Total SOAP 76 52 128 Free form 60 50 110 Total 136 102 238 Note: To eliminate potential bias, students were randomly assigned to each group Again we wanted students to explain what they had learned from the film ± to see whether they could make the director's shared tacit knowledge explicit. What we initially suspected was that those students who had seen the SOAP film and who were presented a SOAP recall form would outperform all of the students who had experienced any other combinations of film and recall forms. We believed that structure matters... and the more the better. We note here that the methodology employed in this experimental process is consistent with prior studies examining tacit knowledge (Reber, 1989, 1993; Seger, 1994). We also acknowledge that some researchers have questioned the generalizability of results based on the use of student subjects (Gordon et al., 1986). However, the validity of using students as surrogates for more experienced decision-makers has also been defended by researchers (Greenberg, 1987). Results A total of 238 freshman (n = 95, 40 percent), sophomore (n = 63, 26 percent), junior (n = 54, 23 percent) and senior (n = 26, 11 percent) students drawn from a core business course participated in this experiment. There were 91 males (38 percent) and 145 female (61 percent). Two respondents did not specify gender. The eight hypotheses discussed earlier were tested using different statistical methods. H1a and H1b were tested using logistic regression method and H2a, H2b, H3a, H3b, H4a, and H4b were tested using ANOVA. In presenting the results, significance levels of 0.05 ( = 0.05) or better will be considered to be ``statistically significant''. For each model, two control variables (gender and class level) were included in a stepwise fashion before the independent variables were entered. Using this method allowed us to exclude variables that were not relevant to each model. In all cases, gender and class level were found not to be statistically significant in predicting the dependent variable. All statistical analyses were conducted using JMP Ver 3.2.3. Table IV summarizes the conclusions of the tests of hypotheses. In the following section we will discuss the details test of each hypothesis and the reasons for our conclusions. Effects of knowledge exchange protocols on the recall of the main point To study the effects of knowledge exchange protocols on the recall of the main point two hypotheses were analyzed: H1a (Structuring knowledge presentation improves the recall of the main point by the respondents) and H1b: (structuring knowledge recall improves the recall of the main point by the respondents). The two hypotheses were tested using the logistic regression method. Logistic regression is useful for situations where the presence or absence of an outcome based on the values of a set of predictor variables needs to be predicted. It is best suited for situations where the dependent variable is dichotomous. In order to examine the effects of gender and class level (freshmen through seniors) on the ability of the students to recall the main point, we conducted separate step-wise logistic regressions. The results of these tests indicate that gender (Wald ChiSquare = 0.726, p- value = 0.6954) and class level (Wald ChiSquare = 5.66, p value = 0.1291) are not Table IV Summary of hypotheses findings Hypotheses H1a H1b H2a H2b H3a H3b H4a H4b Structuring knowledge presentation improves the understanding of the main point by the respondents Structuring knowledge recall improves the understanding of the main point by the respondents Structuring knowledge presentation increases the number of methods recalled by the respondents Structuring knowledge recall increases the number of methods recalled by the respondents Structuring knowledge presentation increases the number of comments that respondents made Structuring knowledge recall increases the number of comments that respondents made Structuring knowledge presentation improves the richness of the understanding of the knowledge exchanged Structuring knowledge recall improves the richness of the understanding of the knowledge exchanged Result Rejected Supported Rejected Supported Rejected Supported Rejected Supported 111
significant in predicting the ability to recall the main point. When controlling for year and gender, we found that the video treatment (whether the respondent viewed a SOAP video or not) was not statistically significant in predicting the recall the main point. H1a was rejected. However, we found that structuring the recall method (whether the respondent used a SOAP recall procedure or not) was statistically significant in predicting the recall of the main point. H1b was supported. Whole-model test and parameter estimates results for the effects of knowledge exchange protocols on the ability of the respondent to recall the main point are summarized in Tables V and VI respectively. To further confirm our conclusion for the support of the first hypothesis and the statistical rejection of the second hypothesis consider the following observations. Out of the total of 238 respondents, 131 (55 per cent) were able to recall the main point and 107 (45 percent) were not. Sixty-one percent (84 out of 136) of those who used the SOAP recall method, were able to recall the main point, while only 54 percent (69 out of 128) of those who saw the SOAP video were able to recall the main point. Additionally, of 131 respondents that were able to recall the main point, 69 (52 percent) had seen the SOAP video and 62 (48 percent) had not. In comparison, out of the 107 respondents that were not able to recall the main point, 59 (55 percent) had seen the SOAP video and 48 (45 percent) had not. Clearly, viewing the SOAP video had no effect on the ability to recall the main point of the film-based narrative. Table V Whole-model test results for the effect of knowledge exchange protocols on whether the respondent was able to recall the main point Model ±LogLikelihood DF ChiSquare Prob>ChiSq Difference 96.77739 2 193.5548 <0.0001 Full 103.52657 Reduced 200.30396 Effects of knowledge exchange protocols on the recall of methods Five specific methods were discussed to achieve the main point of the film (``to plan ahead''). According to the literature, five specific methods consisting of limited information content is a reasonable number for an individual to be able to remember for a short period of time (Dennis et al., 1998). The results of the experiment show that the number of methods that the students were able to recall ranged from 0 to 5 with an average of 1.27 and median of 1. To study the effect of knowledge exchange protocols on the recall of the number of methods recalled, two hypotheses were analyzed: H2a (Structuring knowledge presentation increases the number of points recalled by the respondents) and H2b (Structuring knowledge recall increases the number of points recalled by the respondents). ANOVA analyses were run to test the two hypotheses. Additional ANOVA models were run to test the significant effect of gender and class level on the number of methods recalled by the respondents. The results indicates that gender (Sum of squares = 0.4585, F Ratio = 0.3907, Prob > F = 0.5325) and class level (gender (Sum of Squares = 2.1057252, F Ratio = 0.5965, Prob > F = 0.6179) are not significant in predicting the number of methods that the respondents can recall. When controlling for year and gender, we found that the video treatment (whether the respondent viewed a SOAP video or not) was not statistically significant in predicting the number of methods recalled. H2a was rejected. However, we found that the video treatment (whether the respondent viewed a SOAP video or not) was not statistically significant in predicting the number of methods recalled. H2b was supported. Tables VII and VIII show the results of the analysis of variance and the parameter estimates respectively. Table VI Parameter estimates for the effect of knowledge exchange protocols on whether the respondent was able to recall the main point Term Estimate Std Error ChiSquare Prob > ChiSq Intercept 1.76876932 0.5105275 12.00 0.0005 Knowledge presentation (SOAP vs free form) 0.57318254 0.5316584 1.16 0.2810 Knowledge recall (SOAP vs free form) ±4.1196176 0.3803726 117.30 <0.0001 112
Effects of knowledge exchange protocols on the number of comments made The results of the experiment show that the number of comments made by the respondents ranged from 0 to 16 with an average of 5 and median of 6. To study the effect of knowledge exchange protocols on the number of comments made by the respondents, two hypotheses were analyzed: H3a (Structuring knowledge presentation increases the number of comments that respondents made) and H3b (Structuring knowledge recall increases the number of comments that respondents made). ANOVA analysis were run to test the two hypotheses. Additional ANOVA models were run to test the significant effect of gender and class level on the number of comments made by the respondents. Consistent with the previous case, the results indicates that gender (Sum of Squares = 34.380, F Ratio = 0.2843, Prob > F = 0.5944) and class level (Sum of Squares = 388.899, F Ratio = 1.0762, Prob > F = 0.3599) are not significant in predicting the richness of the recall. When controlling for year and gender, we found that the video treatment was not statistically significant in predicting the number of comments that the respondent made. H3a was statistically rejected. Again, consistent with the previous case, results show that structuring the recall method (whether the respondent used a SOAP recall procedure or not) was statistically significant in predicting the number of comments that the respondent made. H3b is supported. Tables IX and X Table VII Analysis of variance results of the effect of knowledge exchange protocols on the number of methods recalled by the respondents Source Tacit to explicit knowledge conversion: knowledge exchange protocols DF Sum of squares Mean square F Ratio Model 2 14.40408 7.20204 6.4391 Error 235 262.84381 1.11848 Prob > F Total 237 277.24790 0.0019 show the results of the analysis of variance and the parameter estimates respectively. Effects of knowledge exchange protocols on richness of recall The last measure that we looked at in our experiment was the richness of recall indicator, which we used as another surrogate for a study participant's richness of understanding. The richness of recall score was based on a 1-7 Likert scale where a score of 7 meant that the analysis of a study participant's comments suggests a clear and comprehensive understanding of the director's film-based message. Richness for all subjects was a subjective measure that was assessed by a Master's of Accounting student who had seen both films. The average richness of recall score was 2.5 with a median score of 3. As in the two previous cases, to study the effect of knowledge exchange protocols on the richness of the recall score, two hypotheses were analyzed: H4a (Structuring knowledge presentation improves the richness of the recall of the knowledge exchanged) and H4b (Structuring knowledge recall improves the richness of the recall of the knowledge exchanged). ANOVA analyses were run to test the two hypotheses. Moreover, as in previous analyses, ANOVA models were run to test the significant effect of gender and class level on the richness of the recall. The results indicates that gender (Sum of Squares = 1.6678894, F Ratio = 0.9443, Prob > F = 0.3322) and class level (Sum of Squares = 8.00187, F Ratio = 1.5206, Prob > F = Table IX Analysis of variance results of the effect of knowledge exchange protocols on the number of comments made by the respondents Source DF Sum of squares Mean square F Ratio Model 2 2342.423 1171.21 10.5773 Error 235 26021.157 110.73 Prob > F Total 237 28363.580 < 0.0001 Table VIII Parameter estimates for the knowledge exchange protocols on the number of methods recalled by the respondents 1. Term Estimate Std Error t Ratio Prob > t Intercept 0.9675005 0.126059 7.67 < 0.0001 Knowledge presentation (SOAP vs. free form) 0.0445183 0.137663 0.32 0.7467 Knowledge recall (SOAP vs. free form) 0.4929158 0.138691 3.55 0.0005 113
Table X Parameter estimates for the effect of knowledge exchange protocols on the number of comments made by the respondents 2. Term Term Estimate Std Error t Ratio Prob > t Intercept Intercept 3.2556576 1.254264 2.60 0.0100 Knowledge presentation (SOAP vs free form) Video ±1.578405 1.369718 ±1.15 0.2503 Knowledge recall (SOAP vs free form) Web 6.2146277 1.379949 4.50 < 0.0001 0.2099) are not significant in predicting the richness of the recall. When controlling for year and gender, we found that the video was not statistically significant in predicting the richness of understanding of the knowledge exchanged. H4a is rejected. However, we found that the structuring the recall method (whether the respondent used a SOAP recall procedure or not) was statistically significant in predicting the richness of understanding of the knowledge exchanged. H4b was supported. Tables XI and XII show the results of the analysis of variance and the parameter estimates respectively. MANOVA (Multivariate analysis of variance) procedure uses the data covariance structure to simultaneously test the quality of means from different responses. The use of MANOVA is warranted when there is a strong belief that dependencies between measures may exist as indicated by correlation of the residuals in individual ANOVA models. That is, when the value of each measure may be related to the others. In this case, separate univariate analyses such as ANOVA would ignore the information contained in data due to the correlation. This may lead to falsely Table XI Analysis of variance results of the effect of knowledge exchange protocols on the richness of the recall by the respondents Source DF Sum of squares Mean square F Ratio Model 2 26.69158 13.3458 7.9850 Error 235 392.77061 1.6714 Prob > F Total 237 419.46218 0.0004 concluding a significant effect for a variable, when in fact there is none or conversely, concluding no effect for a variable, when in fact there is one. As an added measure of the robustness of our statistical conclusions reached by the four ANOVA models, we also performed MANOVA with our four dependent measures. We wanted to assess whether the respondents were able to recall the main point, the number of steps recalled by the respondents, the number of comments made by each respondent, and the richness of the recall. Tables XIII-XV show the results of our MANOVA model. These results further validate the statistical conclusions reached earlier. Therefore, since the results of statistical analysis using MANOVA are consistent with those of our earlier four ANOVA models, ignoring the dependencies among our four dependent measures did not obscure the analysis. Discussion We were surprised by our results. The only thing that statistically mattered was what recall form a participant in the study had used! That is, regardless of what film someone saw (free form or SOAP), participants who used the SOAP recall form substantially outperformed those who had been asked a simple question about what they had learned from a film! Table XII Parameter estimates for the effect of knowledge exchange protocols on the richness of the recall by the respondents Term Estimate Std error t Ratio Prob > t Intercept 2.1580223 0.154097 14.00 < 0.0001 Knowledge presentation (SOAP vs free form) ±0.059967 0.168282 ±0.36 0.7219 Knowledge recall (SOAP vs free form) 0.6769592 0.169539 3.99 < 0.0001 114
Table XIII MANOVA whole test results with the standard set of four multivariate test statistics Test Value Approx. F DF Num DF Den Prob > F Wilks' Lambda 0.8298142 3.7226 12 611.46 < 0.0001 Pillai's Trace 0.1724207 3.5520 12 699 < 0.0001 Hotelling-Lawley 0.2023993 3.8737 12 689 < 0.0001 Roy's Max Root 0.18823 10.9644 4 233 < 0.0001 Table XIV MANOVA test results with the standard set of four multivariate test statistics for the knowledge presentation methods (SOAP vs NON SOAP) Test Value Exact F DF Num DF Den Prob > F Wilks' Lambda 0.994959 0.2926 4 231 0.8826 Pillai's Trace 0.005041 0.2926 4 231 0.8826 Hotelling-Lawley 0.0050665 0.2926 4 231 0.8826 Roy's Max Root 0.0050665 0.2926 4 231 0.8826 Table XV MANOVA test results with the standard set of four multivariate test statistics for the knowledge recall methods (SOAP vs NON SOAP) Test Value Exact F DF Num DF Den Prob > F Wilks' Lambda 0.9258334 4.6262 4 231 0.0013 Pillai's Trace 0.0741666 4.6262 4 231 0.0013 Hotelling-Lawley 0.080108 4.6262 4 231 0.0013 Roy's Max Root 0.080108 4.6262 4 231 0.0013 What this means is that rich narratives may facilitate tacit to explicit knowledge conversion, but their effectiveness may critically depend on whether the recall process is structured. And, intuitively this makes sense. Ask someone ``How was your vacation?'' and you may get a few vague sentences in response. But ask them to give you a day-by-day account of what they saw and did and you'll probably get much more information. Our findings are consistent with previous studies where structure is used to facilitate tacit knowledge exchange (Reber, 1989, 1993; Seger, 1994). In these studies, information was presented to experimental subjects in either a somewhat randomized manner or in a structured manner. Results indicated that subjects using a structured process routinely performed better than subjects did where the process structure was unclear. However, the results of our study suggest greater specificity as to where and how structure affects knowledge transfer outcomes. That is, when structure is employed in the tacit knowledge sharing process, employing it in a narrative recall procedure seems to be the most critical factor for enabling effective knowledge transfer. 115 For those involved in knowledge management efforts, our findings offer good and bad news. The bad news is that rich narratives are still necessary and vital in the tacit to explicit knowledge conversion process. These narratives will take time to produce and for information to be shared. However, our findings suggest that the good news is this:. How tacit knowledge is shared may not matter relative to structure ± that is, it may be more important to worry most about the adequacy and richness of the narrative content than how it is structured!. Use of knowledge exchange protocols may substantially increase people's ability to articulate what they learned from the rich narrative ± structuring recall may matter in helping to convert tacit knowledge to explicit knowledge. Stated another way, contrary to prior research findings, structuring information (in this case, a rich narrative) may not be as essential to the expressed understanding of that information as compared to the nature of the recall format.
Our findings, though preliminary, suggest opportunities for future research. There is of course the opportunity to replicate the study. Moreover, there is a need for more research on how differing knowledge exchange protocols affect a given narrative's recall relative to effectiveness, efficiency, and quality. While the SOAP process works well in the medical community and while a variation of it appeared to make a difference in our study, the nature of the structuration questions may also differentially affect recall performance. References Alavi, M. and Leidner, D. (1999), ``Knowledge management systems: issues, challenges and benefits'', Communications of the Association for Information Systems, Vol. 1 No. 7, pp. 1-37. Bousfield, W.A., Esterson, J. and Whitmarsh, G.A. (1958), ``A study of developmental changes in conceptual and perceptual associative clustering'', The Journal of Genetic Psychology, Vol. 92, pp. 95-102. Burns, J., Clift, J. and Duncan, J. (1990), ``Understanding of understanding: implications for learning and teaching'', British Journal of Educational Psychology, Vol. 61, pp. 276-89. Choo, C.W. (1998), The Knowing Organization: How Organizations Use Information to Construct Meaning, Create Knowledge, and Make Decisions, Oxford, New York, NY. Cole, M., Frankel, F. and Shap, D. (1971), ``Development of free recall learning in children'', Developmental Psychology, Vol. 4, pp. 109-23. Davenport, T. and Prusak, L. (1998), Working Knowledge, Harvard Business School, Boston, MA. Dennis, A. (1996), ``Information exchange and use in group decision making: you can lead a group to information, but you can't make it think'', MIS Quarterly, Vol. 20 No. 4, pp. 433-55. Dennis, A., Hilmer, K. and Taylor, N. (1998), ``Information exchanges and use in GSS and verbal group decision making: effects of minority influence'', Journal of Management Information Systems, Vol. 14 No. 3, pp. 61-88. DeSanctis, G.L. and Gallupe, R.B. (1987), ``A foundation for the study of group decision support systems'', Management Science, Vol. 33 No. 5, pp. 598-609. DeSanctis, G., D'Onofrio, M., Sambamurthy, V. and Poole, M.S. (1989), ``Comprehensiveness and restrictiveness in group decision heuristic: effects of computer support on consensus decision making'', ICIS Proceedings. Earl, M. and Scott, I. (1998), What On Earth Is A CKO?, London Business School, London. Gordon, M.E., Slade, L.A. and Schmitt, N. (1986), ``The `science of the sophomore' revisited: from conjecture to empiricism'', Academy of Management Review, Vol. 11 No. 1, pp. 191-207. Greenberg, J. (1987), `The college sophomore as guinea pig: setting the record straight'', Academy of Management Review, Vol. 12 No. 1, pp. 157-9. Hansen, M., Nohria, N. and Tierney, T. (1999), ``What's your strategy for managing knowledge?'', Harvard Business Review, Vol. 77 No. 2, pp. 106-18. Horowitz, A.B. (1969), ``Effect of stimulus presentation modes on children's recall and clustering'', Psychonomic Society, Vol. 14, pp. 297-8 Nonaka, I. and Takeuchi, H. (1995), The Knowledge Creating Company, Oxford University Press, New York, NY. Nunamaker, J.F., Dennis, A.R., Valacich, J.S., Vogel, D.R. and George, J.F. (1991), ``Electronic meeting systems to support group work'', Communications of the ACM, Vol. 34 No. 7, pp. 40-61. Patel, V., Arocha, J. and Kaufman, D. (1999), ``Expertise and tacit knowledge in medicine'', in Sternberg, R. and Horvath, J. (Eds), Tacit Knowledge In Professional Practice, Lawrence Erlbaum Assoc., Mahwah, NJ, pp. 75-99. Reber, A. (1989), ``Implicit learning and tacit knowledge'', Journal of Experimental Psychology: General, Vol. 118, pp. 219-35. Reber, A. (1993), ``Implicit learning and tacit knowledge: an essay on the cognitive unconscious'', Oxford University Press, New York, NY. Seger, C. (1994), ``Implicit learning'', Psychological Bulletin, Vol. 115, pp. 163-96. Weick, K.E. (1995), Sensemaking in Organizations, Sage Publications, Thousand Oaks, CA. 116