AJIS vol. 8 no. 1 September 2000 IS A PICTURE WORTH A THOUSAND WORDS? THE USE OF NON-TEXT ITEMS EN INFORMATION SYSTEMS RESEARCH. Alison Parkes Department of Information Systems Massey University Private Bag 11 222 Palmerston North New Zealand a.j.parkes@massey.ac.nz ABSTRACT The use of non-text items (graphs, graphics and tables) is firmly established in the publishing tradition of information systems research. Non-text images are used to convey complex information in a pictorial or diagrammatic sense, adding additional meaning to the discussion. Non-text images also highlight results, and meaningfully represent both quantitative and qualitative data. This paper examines the use of non-text items in information systems journals. Three measures of use are employed: rate of use of non-text items; academic and practitioner journal usage of non-text items; and citation rates of articles with varying levels of non-text items. The sample research articles represent a 30- year timeframe and include articles from six prominent information systems journals. The research results show an increasing use of non-text items in information systems research. In addition, articles that utilise a higher rate of non-text items were more likely to be cited in other works. Practitioner journals were found to utilise more non-text items than academic journals, and to utilise a different mixture of non-text items. It would appear that articles that use higher levels of non-text items are more widely cited, so if citedness is the definition of 'success', then it may be practical to incorporate a range of non-text items. If consideration is given to the motive behind selection of articles for citation when preparing research, it seems possible that increased use of non-text items improves the clarity of research findings, leading to their increased usage, uptake and citation levels. INTRODUCTION Information accumulated during research endeavours can be presented in a variety of ways. The most common methods are via text, (or narrative), or by non-text items depicting the information. There is minimal published research available in this area however a summary of some findings to date follows. An examination of how user perceptions and decision making varied under three different methods of information presentation (tabular, graphical and combined tabular graphical) found that when time limits were low, tabular reports produced better decisions, whereas graphic reports produced quicker decisions. Combined tabular and graphic representations produced both faster and better decision making, and was found to be the preferred report format for most decision makers. Benbasat & Dexter. (1986). Readers tend to assign higher levels of trust in data presented in a scientific manner. It was found that using numerical tables tended to make readers assume the existence of scientific rigor. Davis (1981). Information presented in a combination of text and non-text styles is more likely to be comprehended by the recipient of the information presentation. Hamilton & Ives (1983). The increasing use of graphics in technical writing may be related to the emergence of technology that facilitates usage of graphics in both oral and written information presentations. Owen & van Halm (1989). Advances in publishing technology are likely to have facilitated increased usage of non-text items, however slow uptake of new technology may act as an obstacle to innovation in the publishing industry. McGill (1990). Research on the impact of various combinations of format and summarisation levels on different personality types defined by the Myers-Briggs typing instrument found that where the user personality type is unknown, raw data is the best offering, whereas if the personality type is known, there is a preferred option for each cognitive style. It is interesting to note that perceptions of value of information will vary between different personality types. Moran & Journet (1985). Research Question The purpose of this research is to determine whether the use of non-text items is increasing in published research, whether the use of non-text items affects the level of citedness of articles, and if there are 64 vo1- 8 n»-1 September 2000 differing levels of usage of non-text items between practitioner and academic journals Three hypothesis have been formulated. HI That the rate of occurrence of non-textual information in published Information systems research is increasing over time. Because of increasing awareness of the value of non-text items, and the improved ability of publishers to present non-text items easily, it was theorised that research would increasingly use non-text representations of data H2 That practitioner journals would exhibit higher frequency of non-text occurrences than academic journals. There are two main types of journals where information systems research is published, categorised as academic and practitioner journals. Practitioner journals tend to be 'glossy', with a magazine style format. It was thought that due to the form of the journal, and the target audience, practitioner journals would use more non-text representations than academic journals. H3 That those articles that contain a higher number of occurrences of non-textual data are more often cited. The existence of higher numbers of non-text items in an article increases readability and comprehension of the content. It is thought that successful information systems research would use higher levels of non- text information. Those articles that contain a high number of non-text representations may be more widely used, and, if so, would be cited more frequently. To test these hypotheses, a representative sample of journals was selected for examination. As an aid to determining a representative sample, a typology of journals was used to guide the selection. Vogel & Wetherbe (1984). A representative sample of articles was examined and occurrences of non-textual data recorded. Definition of Non-text items For the purposes of this research the following definitions apply: Table: Information displayed in matrix format, which is labelled or identified as a Table within the article e.g. Ihrr 1-plidp.lloa. CQ.HJCI. ud CooDkt Koolallco TABLE I nin* Sialiain t* StHtly 1'iirwNri Roporaibuiiy llanfeflfi Artivilicl UOT-4S Rebuoninip Influence — Hem 1 Influence — item 2 Influence — item 3 Conflict— item 1 CnnOKI— item 2 Connie!— iiem J Sililfacliwy Conflict Resolution — item 1 SiUMunry fi«l«l Knolulinn— hem 2 Man 0.29 042 OJI 4.11 497 4.49 1.5! 3.46 367 4.11 4.70 Std. dev. 0)1 0.32 0.30 2.20 1.94 2.06 1.76 2.11 19) 112 1.79 Figure J. Example of Table 65 AJIS vol. 8 no. 1 September 2000 1st 2nd 3rd 4th Qtr Qtr Qtr Qtr Figure 2. Example of Graph Graph: A pictorial representation of quantitative data made in a non-numeric illustration, for example a bar chart or pie graph, e.g. GraphicA picture or model illustration e.g. 1. Model of otpntnlioul poULka perceptual. Figure 3. Example of Graphic METHODS Selection of Journals Journals were stratified into four levels of importance, and those levels divided into practitioner and academic orientations. Two journals were selected from each orientation for level 1, and one journal was selected from each orientation for level 2 of the taxonomy; a total of six journals. Levels 3 and 4 of the taxonomy were not considered as the information provided was too generalised to aid in journal selection. In order to select the most appropriate journals to utilise within each classification, the citation frequency was utilised. The citation frequency was combined with the corresponding taxonomy of journal importance. Journals that were identified as attaining a higher number of citations within each level, for each orientation, were selected for this research. The rationale behind the utilisation of the citation frequency as the selector is that testing H3 will involve the use of a citation index. Sample articles are more likely to appear in this index if high citation frequency has been advised for their publishing journal. The most important and frequently cited journals from levels 1 and 2, for both academic and practitioner orientation were as follows: 66 AJIS vol. 8 no. 1 September 2000 Level 1 2 Orientation Academic Practitioner Academic Practitioner Selected Journal Management Science MIS Quarterly Datamation Harvard Business Review Communications of the ACM Communications of the ACM Table 1. Selected Journals As the same journal was selected for both academic and practitioner orientation for level 2, it was decided to examine the second most frequently cited journal for those levels, the Journal of Systems Management. To determine which journal was most representative of academic orientation, an analysis of submission preferences was conducted. This showed a higher academic preference for submission to Communications of the ACM. The journals selected for the final sample were: Level 1 2 Orientation Academic Practitioner Academic Practitioner Selected Journal Management Science MIS Quarterly Datamation Harvard Business Review Communications of the ACM Journal of Systems Management Table 2. Final Journal Selection Selection of Articles In order to test H1: That the rate of occurrence of non-textual information in published Information systems research is increasing over time, a longitudinal sample of journal articles was required. It was decided to sample from the years 1991, 1993, 1995 & 1997. The first issue for each year of the selected journals was selected. Every article in the selected issues was examined. This resulted in a sample of 201 articles; containing 916 identified non-textual information representations. Additional articles from 1971 and 1981, selected in the same manner, were added during the research process, to broaden the longitudinal scope of the research. This resulted in a total sample size of 271 articles containing 1148 non-text images. Selection of Citation Index There are two possible information sources relating to citation frequency, the Social Sciences Citation Index and the Science Citation Index. To determine the most appropriate index to use, an examination was conducted of the source listing for each index, with the results shown in table 3: 67 AJIS vol. 8 no. 1 September 2000 v • Journal Management Science Datamation Harvard Business Review MIS Quarterly Communications of the ACM Journal of Systems Management Social Science Citation Index Source »/ X •/