ORGANIZATIONAL INFORMATION SYSTEMS: DETERMINANTS OF

I

MANAGEMEhT SCIMCE V d . 24 No. 2, Febmuy I982 ?rwd ia U.S.A.

ORG

ORGANIZATIONAL INFORMATION SYSTEMS: DETERMINANTS OF THEIR PERFORMANCE AND BEHAVIOR* GEORGE HUBERt This paper is conEmral with organizational infomlation systems. Exampla of such systems include intelligence system& communic.ations system& management infol7Mtion systems decision support systems, and administrative control systems. Systems such as these are critical to UL orgaaixation’s functioning; indeed to its survival. Tbe paper is intended to be of use to three typed of professionah those who sr+y organizations; those who h&n information systems; and those whomassage. T’bis fact causa the paper to differ in a number of ways from earlier works dealing with organihonal communiutioaa And related topics. For example, the paper draws on three different literatures: the psychological litmrure dealing with perception and cognition, the soci& psycbologiul litcnturc dealing with tie role of motivation in communications, and the orgaaixatiooal literature dealing witb behavior in information systems. Further. tbe paper gives greater attention than do earlier works to logistical determhnts of information system effectiveness, such as tbe workload of tbc unit procuaing the message or tbe priority assigned to a mesaage, as wntraated with social-psychological determinants. Fiiy, the explicit introduction of judgment and argument by analogy when direct evidence is unavailable isa third way in which the paper diffm from utlier works. l-be paper focuses in pallicuhr 00 the de&zminan ts of the performance and behavior of systems such as those mentioned above. It discussa in some derail the impact of these detcrmhants on four processes that are of key importance ia tbc operation of information +ems. Two of these procesacs arr used to increase system efficiency. One. masage routing causes any padulu message to be distributed to relatively fewo~tionaf unit. and thus grcdtly reduca tbe informatioo proa&n g load of the auny units that might otherwise be involved in receiving or relaying tbe message. The other. musup swfvwrhing. plays a similar role. It has as its pqose reducing tbe sixc of the message while at the same time faithfully reproducing its meaning. The Wmaining Iwo procemes follow from the fact that organizational units necusAly exercise some discretion in lbe way tlmt they handle messages. M-ge &by ia a wnscqucoa of the priority assignment given a mege, and in many cases enhances tbe effectiveness not only of the operating unit but of tbt organktion a~ a whole. Murap nmfiificario refers to the distortion of meauge meaning. Its source may k either the cognitive limitations or tbe motivations of the sender or receiver. Modifications may be conscious or unwnscious, well-intended or malicious. Each of these four processu is discussed in some depth. More specifically, tbe several determhanu of tbe probability or extensiveness of each process’s occurrence are identified. Tbe lituahue related to tbe various proceaAetermhn tpairingsistbenreviewed,andis summanzd in tbe form of propositions. Severalpluu where additional rcaearcbisoeaiedare noted and recommendations are made concerning whattbe nature of such raarcb should be. (ORGANIZATION DESIGN; INFORMATION SYSTJZMS)

1 . IllThis paper is concerned with the performance and behavior of organizational information systems, (e.g. intelligence systems, communication systems, management information systems, decision support systems, and administrative control systems). In their traditional and noncomputer-aided forms, such systems are critical to

l Aacpted by kie Y. L&n; received June2.1980. Tbis paper has been with the autbor 6 mootbs for I =?ELty of wiiti-Muiison.

organizational functioning external environments, the organizational units that “ implementation-related ir units, and they transmit t t& monitoring and s&r performance and behavioing that a number of orga information processing Nadler, [76]; O’Reilly an of our knowledge cancer organizational informati organizational informatic Although the concept it is to design or manage important to the thcori whose profession it is tc standing organizational conringency rheory (Chik of environmentally-relat mental uncertainty, a b organization’s current [79); Campbell and Pri and valences or utilitir outcome and payoffs a be affected by the n; components in theories of power, the availab decisions. These varia important to the field. organizational conflict In view of these ma of information is impc tional behavior, but tc that the treatment of i or design of organkt GaIbraith, [31]; Tush1 this may be especial& Jemiscm, [4OD. Givenallofth.&it literature dealing wit systems, and in this v the world of practice science. This paper r The paper differs uxnmunications (cf. [I6]; Redding, [63]: O’ReilIy and Pondy. logical literature dea

138 ‘RadmiBcaacaiin wiabtorcfertotbetic

ORGANIZATIONAL INFORMATION SYSEMS

139

functioning; they monitor and scan the organization’s internal and external environments, they transmit the resulting observations and interpretations to organizational units that “decide” if actions are called for, they relay the decisions and implementation-related information from these deciding units to the implementing units, and they transmit the progress and results of these impiem+ations as part of the monitoring and scanning activities mentioned earlier. Indeed, organizational performance and behavior are so closely linked to organizational information prm ing that a number of organizational scientists have advocated viewing organizations as information processing systems (cf., Simon, (70); Galbraith, [31]; Tushman and Nadler, (761; O’Reilly and Pondy, [58].’ These thoughts suggest that a summa&ation of our knowledge concerning the determinants of the performance and behavior of organizational information systems would be useful, useful to those who design organi&onal information systems and useful to those who manage such systems. Although the concept of information is clearly important to those whose profession it is to design or manage information systems, we should note that the concept is also important to the theories and concepts employed by those management scientists whose profession it is to develop, test, and refine theories for predicting and understanding organizational performance and behavior. For example, a major thrust of conringev theory (Child, [17]; Borsch, [461) concerns the processing and distribution of environmentally-related information. Other aspects of this theory deal with environmental uncertainty, a variable frequently viewed as a lack of information about the organization’s current or future environment. Similarly, expecfuncy theory (Vroom, [79]; Campbell and Pritchard, (14D deals with variables (expectancies or probabilities and valences or utilities) that are thought to be affected by information about the outcome and payoffs associated with certain behaviors. Other variables that seem to be affected by the nature or availability of information and that are important components in theories of organizational performance or behavior are the distribution of power, the availability of organizational slack, and the nature and quality of decisions. Th&e variables in turn have a considerable impact on other variables important t6 the fields of organization theory and organizational behavior, such as organizational conflict and organizational effectiveness. In view of these many linkages, it is perhaps not surprising to fmd that the concept of information is important not only to the fields of orga.n&tion theory and or@tional behavior, but to that of orgonirpron design as welL Many would argue, in fact, that the treatment of information may be the key issue to be dealt with in the analysis or design of organizations (cf. Thayer, [73]; Wile*, [82]; Miller, [56]; Simon, 1701; Galbraith, 1311; Tushman and Nadkr, [76]; O’ReiIly and Pondy, (58D. It m that this may be qccially true when designing boundary spanning units (Tushman, [74]; Jemiso~ [4OD. Given all of this, it seem reasonable and important to examine and summariz the litem~~~ dealing with the performance and -behavior of organizational information systems, and in this way enable management scientists to operate more confidently in the worid of practice and to claim a fuller understanding of their field in the world of science. This paper reports the results of having attempted these efforts. The paper differs in three ways from earlier works dealing with organization communications (cf. Guetzkow, (351; Thaycr [73]; Voos, [78]; Ferencc, [28]; Carter, [ 16); Redding, [63]; Farace and McDonald, 127; Porter and Roberts, [613; and O’Reilly and Pondy, (58D. Fit, it draws upon three separate literatures--the psychological literature dealing with perception and cognition, the social-psychologid litera-’

organizational

YSTEMS: 0’“E AND

~mp1c.s of such systems information system.% 1ch as Lhsc are critical ld.3: thOSC

Who JtlU+

oqe. This fact causa ng with organwioMl n three different litcracognition, the socialnmunications, and the Further, the paper gives of information system or the priority assigned 3. Finally, the explicit .cncc is unavailable is a mance and behavior of 3il the impact of the pclaticm of information :y. One message routing liuticmal units, and thus * chat might othwise be vnwihg, plays a simihr the same time faithfully ational units necuwiiy is a cmscqucnce se nc , effativenesa not age wuxiij?cation refers to gnitive iimitations or the msciocs or unconscioug

: qxtcifically.

the several

occurrence are idcntifiai. is thc3 rwicwcd, and is nal rcs%arch is needed are ,I such .zscarch should be.

ehatior of or-tional tion systems, management ;trativc control systems). :h systems are critical to 50 with rhe aurhor 6 months for I

‘Readers intetwtal in a ntorc detailed analysis of the role of information systems in oqddoar my wish IO refer to the articla by Milk 1561 and Hubez (38).

,

-

140

GEORGE HUBER

ture dealing with the role of motivation in communications; and the organizational literature dealing with behavior in information systems. Second, it gives somewhat greater attention than do earlier works to logistical determinants of information system effectiveness, such as the workload of the unit processing a message or the priority assigned to a message, as contrasted with social-psychological determinants. This feature does not, of course, limit the scope of the manuscript to the formal information systems of bureaucratic organixations, as messages transmitted within informal or ad hoc networks are also affected by the workload of the people involved and by the processing priorities that these people assign to the various items of information that they encounter. Obviously both logistical and so&i-psychological variables affect both formal and informal information systems. In view of this, it seemed important to highlight the importance of logistical variables, since the earlier works cited above had focused heavily on social-psychological variables. The explicit introduction of judgment and argument by analogy when direct empirical evidence is unavailable is the third way in which the paper differs from earlier works, although the difference is more one of degree than of kind. The usefulness of drawing on reasoning by analogy when attempting to identify relationships among variables in organizational information systems is made clear in Miller’s portrayal of the organization as a living system (Miller, [56D. In this classic work, Miller argues and demonstrates that the relationships among variables, that hold true at one level of organizational analysis generally also hold true at other levels. For example, the effects of stress on the information processing effectiveness, of an organizational unit are essentially the same whether the unit is a person, an ad hoc group, or an established organ&tion (cf. Miller, [54], [55]; Meier, [52]; Driver and Streufert, [25D. As will be seen, the literature review leading to the present manuscript identified additional support for this position, and identified no evidence to the contrary. As a consequence, the summarizing statements put forth on the following pages are not qualified so as to pertain to only the levels of analysis (or organizitional levels) for which empirical evidence is presently available. For example, in these statements the word “unit” is intended to be broadly defined, and subject to being interpreted as an individual, an ad hoc committee, or a formal work group, even a corporate division. Thus the statements, which are labeled “propositions” are in general not conclusions based on empirical studies covering all possible levels of analysis. Rather they are statements that represent what it seems reasonable to believe, based on the aggregate evidence available. In a few instances, the literature pertaining to the topic was quite sparse. In these instances, I have stated and supported with argument what I believe the literature wilI say when it appears. So that the reader will not be misled, these few instances are clearly identif&i2 By highlighting these facts, and by referencing under each proposition the supporting evidence, I hope to aid those researchers interested in fiUing the voids in our knowledge about the treatment of information in organizations. In addition, by explicating with these propositoins what it seems reasonable to believe, based on the evidence at hand, I hope to be helpful to those management scientists who are being asked to analyxe and “repair” malfunctioning information systems, and even to aid in the design of such systems. Let us turn now to a discussion of four processes that affect the availability, form, ‘If the ruder would like to reserve the term “propaitions” for relational rt~tanentn that are well N~bythCempiriullirmtur+hCorrhC~ych~torrudrhCierrwellNpporcsdrrktionrl statmmnta u “hyp~tha~,” or “cmjcctwa.”

For sume rudera the las well supported rtatenmtr may be the more intcrutin& as they may identify the mole likely oppommitia for significmt cmpiriul contributions.

0

and meaning of messagperformance and beha. will be used in this ant refers to that which U r in the information the0 ty-reducing potential of which information is cc out of the ad hoc or assess, distribute, alter “organizational inform. to a broad scope of act analyzing a report, a: includes activities that as those that take plac: 2 . PA Organizations acquithe critical functions requires the processing other hand, because cognitive and logistica the information system of organizational mess Two processes that systems are message particular message tc selective distribution potential receiving un intermediate units in summarizing plays a 5 while at the Same tim numbers are replace appropriately derivet reduce the cognitive c Messages vary car attributes. As a conse units responsible for the way they hand processing phenomer message delay and m There is no value _ delay. Since the prior time it will be delay part) a deiegated an judgments about the that the sources of categorized as ex= categories. Message wwa?ficaf either the cognitive ‘The word klit” is tL pomm;wrr . rr--1 “.--

A;

and the organizational

nrlier works to logistical Al/ &load of the unit *a’L rs contrasted with course, limit the scope of lucratic organizations, as are also affected by the -iorities that these people counter. Obviously both al and informal informablight the importance of had focused heavily on by analogy when direct ;h the paper differs from agree than of kind. The .pting to identify relations is made clear in Miller’s 561). In this classic work, .g variables that hold true true at other levels. For :ssing effectiveness of an 2it is a person, an ad hoc j; Meier, (521; Driver and to the present manuscript tified no evidence to the put forth on the following analvsis (or organizational le example, in these net, and subject to being ormal work group, even a i in general not conclusions

analy&~ Rather they are .ve, based on the aggregate ning to the topic was quite :h argument what 1 believe ill not be misled, these few Tc and Driver and Sue The empirical stt under high load con processing capa+ appropriate than +A highlight the fact ti will cause occasion transmission withir beyond the unit’s c Proposition R.2 1 especially multiple one another. PRo~osmo;u R.related to the perct T h e literanUr +

contractual grieva that achieves 0rg.a message-sending 1 tional penaltie% p. in determining u relevance critcti:

ORGANIZATION A L INFORMATION SYSTFMS

ticious. They range from modification of substitutrom message sueration does not. Although :ez a formal or-, arhysis of the “message mine the availability of routing, the process that

9 in ding ordinary men, ,rganizacion, arranging to ually ordinary men. This

1, pp. 60-611 highlights a rtmentalization of many zh specialization and deation does not have to be : process that principally not to others, is a logical d to information processions of messages allows iizational goals, probably :Idom notice its presence. organizational literature,* ‘* and “transmitting” the to much of this literature, .ay 9 role. :r g and transmission lg; \A) workload of the ens from communicating ;ceiver; and (6)’ frequency jr may have used terms variables referred to were ~1 values of these variables :re are two reasons for the it the objective values are ICC’* or “power,” whereas rtant to managers, is that >jective, values cannot be, eem to be most useful to riables and the perceived nay be that the perceived rbles. ys about message routing, puted to a unit is inversely

2, Chapter S].

143

In order to conserve their resources, we expect organizational units to communicate more frequently with units easily contacted than with other units. This certainly would be so if we could extrapolate the social-psychological findings that messages will be directed toward (a) persons in close proximity, (b) persons in the same work group, and (c) persons in the same socioeconomic status (Collins and Guetzkow, [19. p. 187D. On a more macro basis, and focusing on the concept of physical accessibility, Miller states that “In general the farther components of a system are from one another and the longer the channels between them are, the less is the rate of information flow among them” and “the less . . . encoding a channel requires, the more it is used” (Miller, [56, p. 31). In subsequent empirical studies, Brenner and Sigband [ 111, Conrath [20], and Bacharach and Aiken [7] found that either physical or structural accessibility was a determinant of the frequency with which subordinates communicated with superiors. The difficulty in communication may be interpersonal as well as physical or structural, as observed by Brenner and Sigband [ll], and Goldhar, Bragaw, and Schwartz [33] and as noted by Jam [39), “If the subordinates perceive the communication from the supervisor as generally positive, they would be encouraged to exchange a great deal of information about task related matters . . . *’ Anticipating these findings, March and Simon argued that “the greater the communication efficiency of the channel, the greater the communication channel usage” [49, p. 1671. Ference [28] offered similar propositions that focus on avoidance of time losses as a variable that determines routing. Studies concerning the choice of a communication medium or information source are also supportive of this proposition (cf., Johnston and Gibbons, [41)). P ROPOSITION R.2. The probability that a message will be transmitted from a unit is inversely related to the workload of the unit.

It is reasonable to expect that the transmission behavior of organizational units would be affected by their workload. Meier, for example, found that overloaded units “destroy lowest priorities’* when carrying out their functions [52, p. 5351. Research supporting this hypothesis for individuals and small groups is reported by Miller [54] and Driver and Streufert [25], respectively. The empirical studies relevant to this proposition have found support for it only under high load conditions where the unit seemed to be near or beyond its information processing capacity suggesting that the phrase “work overload” might be more appropriate than “workload.** We call attention to this issue of word choice in order to highlight the fact that the nonuniform arrival of demands for message transmissions will cause occasional work overloads, and thus temporarily lower the probability of transmission within time periods where the average workload observed would not be beyond the unit’s capacity. Proposition R.2 highlights the usefulness of providing for multiple message sources, especially multiple sources whose workload magnitudes are somewhat independent of one another. ~OPOSITION R.3. The probabihty that a message will be routed to a unit is positively related to the perceived relevance of its contents for that unit.

The literature on management by objectives, on personnel evaluation, and on contractual grievance procedures suggests that organizations tend to reward activity that achieves organizational goals and to punish activity that does not. It follows that , message-sending units, in order to achieve organizational rewards and avoid organizational penalties, would use the relevance of a message for some other unit as a criterion in determining whether to route the message to that unit. Certainly it seems that if relevance criteria were formalized with standing orders directing certain types of

144

GEORGE HUBER

messages to certain units, it is more likely that the message would be sent to the designated units than to undesignated units. In addition, our everyday encounters with overworked colleages suggest that the receiving units themselves tend to provide penalties in the form of cornpaints to units that send irrelevant messages. An early study by Davis (231 offers strong support to this proposition, and a later study by Sutton and Porter [72] offers weak support. Relevance, is, of course, influenced by tasks, assignments and responsibilities. Tushman found that, for high performing units, the greater the task interdependence, the greater was the frequency of communication (Tushman, [75D. We should note, however, that while relevance influences routing, this influence is not entirely dictated by formal considerations, as shown by Festinger, Schachter, and Back [29] and by Wickesburg 1811, who found that individuals seek information “‘wherever in the organization information, advice counsel, and expertise may be found . . . . Formal organization boundaries and levels yield to the demand of the task and situation” (Wickesberg, [81, p. 25fD.

ORGX

Some nonsupportive evic communications within an communication flow was do might be a consequence of i.n reality initiated by lower routmixed technologies ma scheduled quality control rf management by exception, PROPOSITION R-6. The pr related to the jrequency wil recent past.

This proposition follows that the sender’s perceptior short term interval. It also

R.4. The probabiliq that a message will be routed to a unit is inverse@ related to the decrearr in its goal attainment that the sending unit believes will occur as a result of the routing.

channel usage Len& to character, develops sidct channd for either kind o Simon. [49, pp. 167- 1681

As a result of his survey research in three industrial organizations, Read concluded that “Individuals in power hierarchies tend to screen out information passed upward, and to withold or refrain from communicating information that is potent$lly threateni ing to the communicator” (Read, (62D. Direct support for this proposition comes from three lines of empirical research. One is the social-psychological research dealing with the suppression of “bad news” in interpersonal communications (cf., Rosen and Tesser, [65); Rosen, Johnson, Johnson, and Teaser, [66D. A second is the field research dealing with the suppression of information that reflects adversely on the organizational performance of the unit controlling the information (cf., Carter, [IS]; Ullman and Huber, [771; and McCleary, [SOD. The third is the research on bargaining (cf., Cummings and Hamett, [21D and organization power (cf., Pfeffer, [6OJ) indicating that information is a critical resource to joint decision situations and that withholding information from one’s competitors is often useful in attaining one’s goals in a competitive environment. As we will see when we discuss message modification, a good deal of research indicates that individuals distort those messages that might adversely affect their goal attainment (cf. the laboratory studies of Cohen, [18], and O’Reilly and Roberts, [59], and the field studies of Mellinger, [53]; Gore, (341; Read, [62]; Athanassiades, (61; Kaufman, [42]; Roberts and O’Reilly, [64]; and O’ReilIy, [57]. Since message suppression (nonrouting) is an extreme form of distortion, it may be argued that this body of literature also supports the present proposition.

Let us t&n now t o a availability of information

~oPosrnoN

~OPoSrTtON R.5. The probabili~ that a message will be routed to a unit b patitiveb related to the perceived power and statw of the unit (except for the situation described under Proposition R .4).

This proposition follows from the findings that (1) persons of low status and power tend to direct messages to persons with more status and power, and (2) persons of high status and power tend to communicate more with their peers than with persons of lower status and power (Bamlund and Harland, 181; Collins and Guetzkow, 119, p. 1871; Allen and Cohen, [3D. The proposition might also be inferred from the finding that persons in high organizational positions, positions which often have more power and status, are better informed (Davis, (231; Zajonc and Wolfe, [83]; Sutton and Porter, [72J), although other variables such as seniority or perceptiveness may also play a role in this finding.

The time that elapses summarized or modified chain we will call “de! necessary to process the other tasks, and (3) the waiting perhaps for its inclusion as a combinat of these components is sender assigns to prcces course, be influenced b matter of timing. The next several par delay: (1) timeliness of number of links in the associated propositions assessing their usefulne ~OPOSTTtON D.1. Tj to the perceived rimelin

out every&y obsr Gerstenfeld and Berg sending messages in s impact, such as when receiving the message Assuming that orE sending units would t delay least those mes way of putting the sendinguniutowt messages.

ORGANIZATIONAL MFORMATION SYSTEMS

lge would be sent to the

everyday encounters with mselves tend to provide ev messages. An early .d a later study by ,a ems and responsibilities. the task interdependence, n [75]). We should note, Ice is not entirely dictated :r, and Back [29] and by nation “wherever in the y be found. . . . Formal f the task and situation’*

143

Some nonsupportive evidence is the conclusion of Davis [23]. in his study of communications within an industrial management group, that “the predominant communication flow was downward or horizontal.” It seems, however, that this finding might be a consequence of (1) higher organizational levels issuing directives that were in reality initiated by lower level staff groups or (2) the fact that in organizations with routinized technologies many upward “messages” are uncounted by observers, e.g., scheduled quality control reports or the absence of “out-of-stock” reports which is, via management by exception, a message in itself. PROPOSITION R.6. The probability that a message will be routed to a unit is positively related to the frequency with which similar messages have been routed to the unit in the recent past.

This proposition follows from the Propositions R.l, R.3, and R.5, since it is unlikely that the sender’s perceptions of cost, relevance, and status will change significantly in a short term interval. It also follows from the argument of March and Simon that

7uted to a unit is inversely (nit believes will occur PP a

Channel usage tends to be self-reinforcing. informal communication. much of it social in character, develops side-by-side withtasksrientcd formal communication, and theuse of the channel for either kind of communication tends to reinforce its use for the other (March and Simon. (49, pp. 167- 1681).

libations, Read concluded ‘ormation passed upward, Iat is potentially threaten-

Let us turn now to a discussion of delay, another of the processes affecting the availability of information.

If empirical research. One ession of “bad news” ‘in Rosen, Johnson, Johnson, with the suppression of performance of the unit lber. [77]; and McCleary, :S Hamett, [21]) and I&._ d a critical resource from one’s competitors is ment. a good deal of research :dversely affect their goal *Reilly and Roberts, [59], [62]; Athanassiades, [6]; 1. Since message suppreaargued that this body of

The time that elapses between when a message is received and when, in perhaps a summarized or modified form, it is passed on to the next link in the communication chain we will call “delay.” The delay time has three components: (1) the time necessary to process the message, (2) the time lost while the processing unit tends to other tasks, and (3) the time lapse while the processed message is held in storage, waiting perhaps for its relevance to increase or for another message to arrive for inclusion as a combination of messages to be transmitted simultaneously. The second of these components is a function of, among other things, the relative priority that the sender assigns to processing the particular message. The sender’s prioritization may, of course, be influenced by organizational directives. The third component is basically a matter of timing. The next several paragraphs highlight three variables that seem to affect message delay: (1) timeliness of the message, (2) work overload of the sending unit, and (3) number of links in the communication chain. Because the empirical support for the associated propositions is relatively weak, the reader should exercise some judgment in assessing their usefulness in fulfiig his or her needs.

rated to a unit is positiveiy ‘or the situation described

of low status and power r, and (2) persons of high zrs than with persons of s and Guetzkow, [19, p. nferred from the finding I often have more power Wolfe, [83]; Sutton and reptiveness may also play

4 . Deb&q

PROFQSITION D. 1. The probability dr duration of message aWay will be inverse& related to the perceived time&tejs of the message for the receiving unit.

Our everyday observations and some research (cf., Ulhnan and Huber, [77); Gentenfeld and Berger, [32D suggest that message processing units tend to delay sending messages in situations where premature delivery would reduce the message’s impact, such as when the message receiver is overloaded or has a mental set against receiving the message. Assuming that orwtions tend to reward good performance, it foilows that sending units would tend to delay most of those messages that were not timely and to delay least those messages where delay would be costly to the receiving unit. Another way of putting the argument is that use of organizational sanctions would cause sending units to use timeliness as a criterion for assigning priorities to the processing of messages.

146

OR

GEORGEHUBER

~OmSlTION D.2. The probabiliry or duration of message delay will be positively related to the workload of the sending unit.

Further evidence in I managers attempt to cr overstating costs (Lowe senders do not trust the more than otherwise (( O’Reilly and Roberts, modification is influcnc the receiver (Watson an the sender’s mobility E Athanassiades, [6D. 0’1 variables of trusf influe Additional support fc example, Collins and C communication from 1~ status person has learn kow, [19, p. 187D.

in labmato~ setting, Miller [54], [55] has observed that delay in processing of information is one way in which individuals deal with information overload, and he argues that the delay would also be used by organizational units when faced with overload. Additional evidence is the case study of work overload in libraries by Meier [52]. In a more clinical vein, as a result of his interviews with administrators in bureaucratic organizations, Downs concluded that “The most common bureau response to communications overloads is slowing down the speed of handling messages without changing communications network situations or transmission rules” [24, p. 2701. P~oposrrro~ D.3. The probabiliv or duration of message delay is pasitiveb related to the number of sequential links in the communication chain connecting the receiver to the message source.

This proposition relies on the reasoning, supported by Allport and Postman [4], Higham [36], and Miller [56], that the greater the number of sequential links in a communication chain, the greater would be the overall effect of a phenomenon that took place at each link. Let us move on now to examining the variables that determine changes in the form and meaning of organizational messages. We begin by looking at those that affect message modification, as it is this process, rather, than message summarization, at which the greater amount of empirical research has been directed.

Proposition

Some research sugg reducing the stress on Campbell notes that be he (the sender) make: remembered detail th gaps are conspicuous’ even after conuolhn sender and receiver . receiver against the s to not distress the re that ‘%rformation, c perceptions of the re

5. Modification Altogether, eight propositions are presented that relate message modification to other variables. We begin with a few propositions concerning the motivational bases for modification, then turn to those dealing with perceptual and cognitive bases, and conclude with a proposition identifying an organizational determinant of message modification. In some cases a proposition could have been decomposed into subpropositions that would focus on a particular form of modification or form of the associated determining variable. I have chosen to avoid extensive divisions of this nature and hope that I have thereby decreased the chance that either I or the reader would miss the forest for the trees. The references identified with the propositions contain the more specific propositions or findings from which these propositions were constructed.

PROPOS~OX

M-3

related to the discret.

PRoposmo~ M. 1. The probobiliry or extent of message modification is positively related to the increase in its goal attainment that the sending unit believes will result from introducing the modification.

As a result of his extensive interviews with administrators, Downs concluded that “Each official ten& to distort the information he passes upward in the hierarchy, exaggerating this data favorable to himself and minimidng those data unfavorable to himself’” (Downs, [24, p. 266D. His conclusion is strongly supported by the laboratory studies of Cohen [ 18) and O’Reilly and Roberta [59] and the field studies of Mellinger [53], Gore [34], Read [62], Athanassiades [6], Kaufman [42],,Roberts and O’ReiUy [64], and O’Reilly [571. The independent variables in these studes were the receiver’s status and influence over the message sender’s goal achievement, the sender’s trust in the receiver, and the sender’s attitudes towards his own upward mobility. The dependent variables were quite varied, and included revising the message format (the mildest form of modification) and eliminating the message or substituting an incorrect mes-

M.2.

related to the decrease or her modification.

I

It seems that the discretion in chaos; empirical study rela extensive interview: distortion is to use (except through ou nated definitions ar i.e., they are of fiie &ings, the use 0 discretion in the st those instances wh as in those where * We turn now tO they and the __mot perceive is affect perceive-it is use

_ e delay will be positively ..t delay in processing of marion overload, and he -1 s when faced with oab ,.II libraries by Meier s with administrators in ,.ost common bureau reted of handling messages ansmission rules” (24, p.

Aay ir posirively refated to necling rhe receiver to the Uport and Postman [4], of sequential links in a ,zt of a phenomenon that .nine changes in the form J&g at those that affect G:ssage summarization, at rcted.

message modification to .g the motivational bases and cognitive bases, and determinant of message inr, Jubpropositions that the associated determinIS nature and hbpe that I would miss the forest for untain the more specific constructed. modification is positively .rir believes will result from ‘r, Downs concluded that upward in the hierarchy, hose data unfavorable to sported by the laboratory field studies of Mellinger (oberts and O’Reilly [64], were the receiver’s status the sender’s trust in the mobility. The dependent sage format (the mildest ltuting an incorrect mes-

Further evidence in support of the proposition follows from the findings that managers attempt to create slack in their budgets by understating revenues ad overstating costs (Lowe and Shaw, 1471; Schiff and Lcwin, [67], [68D and that when senders do not trust the motives of the receivers, they tend to modify the mes~g~ more than otherwise (cf., Mellinger, [53]; Loomis, [48]; Read, [62]; Zand, [84]; O’Reilly and Roberts, (591; Roberts and O’Reilly, (641. Related findings are that modification is influenced by the sender’s perception of the receiver’s influence over the receiver (Watson and Bromberg, [80]; Alkjre, Collum, Kaswan, and Love, [2] and the sender’s mobility aspirations (Read, [62]; Lawler, Porter, and Tenenbaum, [44]; Athanassiades, [6]). O’Reilly [57] provides a particularly articulate discussion of the variables of trust, influence, and mobility. Additional support for the proposition follows from the small group literature. For example, Collins and Guetzkow’s review led them to conclude that ‘The content of communication from low to high power-status persons will depend on what the low status person has learned is most likely to obtain reinforcement” (Collins and Guetzkow, [ 19, p. 1871). ~OFOSXTION M.2. The probabiliv or extent of message modification is positively related to the decrease in stress on the receiver that the sender expects will result from his or her modification.

Some research suggests that message modifications are made for the purpose of reducing the stress on the receiver. In his review of the early psychological research, Campbell notes that “through an anticipatory monitoring of his own intended output, he (the sender) makes an active effort to produce a coherent output, by suppressing remembered detail that does not now seem to fit and by confabulating -detail where gaps are conspicuous” (Campbell, [ 13, p. 3423). Further, Rosen and Tesser [65] found, even after controlling for any possible prior or subsequent interaction between the sender and receiver and for the possibility of any punitive action being taken by the receiver against the sender, that senders still attempted to modify their messages so as to not disttess the receiver. As a result of his reading of the literature, Ference stated that “Information, once evaluated and integrated, will tend to fit the transmitter’s perceptions of the recipient’s needs” (Ference, [28, p. B-851). htOF+OSlTION M.3. The probability or extent of message modijication is positivet’y related to tk discretion allowed in choosing the masage format.

It seems that the tendency to modiry messages would be less if the senders had less discretion in choosing the format of their communications. I know of no specific empirical study relating to this proposition, but note that Downs concluded from his extensive interviews with the administrators that “One way for officials to avoid distortion is to use messages that cannot be altered in meaning during transmission (except through outright falsification). Such messages usually involve both predesignated definitions and coding or easily quantifiable information” (Downs, [24, p. 126]), i.e., they are of fixed format. It is interesting to note that predesignated definitions and codings, the use of checklists and forms, and most other mechanisms for reducing discretion in the selection of message format would tend to reduce modifications in those instances where the modifications were perceptually or cognitively based as weU as in those where they were motivationally based. We turn now to the perceptual and cognitive bases of message modification. mile they and the motivational bases of message modification are interactive-what we perceive is affected by what we are and what we are is affected by what we perceive-it is useful for both administrative and research purposes to make diitinctions between them wherever possible. Propositions M.4 and MS identify modifica-

¶

.--

--148

,

- ORG

GEORGE HUBER

would be less. Empirical SC Closely related is the idea and if the anticipated cost One contribution to antic: get the correct message fr for having made the mod: information channels mc (DOW-IS, [24, p. 2691). ‘Ihe one might expect (cf.. Hs

tions introduced by the message receiver. Propositions M.6 and M.7 return again to modifications introduced by the sender. ~OPOSl’TlON M.4. The probabiIiry or extent of message madification is positiveb related to the difference between actual message content and its expected or desired content.

_

In his review article, Campbell [ 131 notes that both cognitive limitations and personal motivations cause transmitters to imperfectly modify messages during assimilation, stating that the “tendency to distort messages in the direction of identity with previous inputs is probably the most pervasive of the systematic biases” @. 346), and “that . . . the human transmitter is prone to bias away from input in the direction of the transmitter’s own attitudes” (p. 3%). As a result of his interview study, Downs concluded that “Officials’ perceptions will operate so as to partially screen out data adverse to their own interests, and magnify those favorable to their interest” [24, p. 2721. These conclusions suggest that information inputs are transformed in the direction of the receiver’s prior information, expectations, or wishes. McLeod (51, p. 2181 reviewed a number of studies suggesting that the transformation is less if the sender expects to receive further information on the subject.

~OPOSmON

More links in a COTdistortions to occur. Tha the rumor transmission [36]. Other writers have [56]. In their discussior. “Selective filtering take every state in the trani Simon, (49, p. lSSl>, ar through many officials. the next, the final outp is, significant distortio have addressed these c A dramatic example

PnoposmoN M.5. The probabiliry or extent of message modification is positively related to the perceived ambiguity of the data on which the message is based. &una summarized the early work relating to this proposition in his review article “On Perceptual Readiness” as follows: “Presented with a complex stimulus, the subject perceives in it what it is ‘ready’ to perceive; the more complex or ambiguous the stimulus, the more perception will be determined by what is already ‘in’ the subject and the less by what is in the stimulus’* (Bruner, [12]). Porter and Roberts, in their review of findings related to this idea, stated that ‘These results would indicate that the more. tangible and objective the subject matter . . . the more likely it is that subordinates and their superiors will feel that they are communicating accurately, whereas when the messages involve more subjective opinions and feelings there is greater doubt about accuracy” (Porter and Roberts, (6 11).

A rCpOrtCr -aaS F D$fision in 1967. It brigade was: “On no l-he brigade rack absolutely COntiUCe The battalion rat COng in tlxhade:

PRomsmoN M.6. The probability or extent of message modification is positiveb related to the extent of the sender’s work overload.

The company cc

It seems reasonable to believe that, if the sender is either cognitively or logistically

69D.

overloaded, message modifications would be greater. His early literature review led Campbell to conclude that “Whenever human beings operate at near maximum capacity, selective information loss-undesired reduction of message complexity-is apt to be involved . . . ”(Campbell, [13, p. 3363). Miller [54], [55] has given considerable attention to information overload and found that a wide variety of modifications, e.g., filtering, approximation, and omission, and other devices, are used to deal with it. Of some interest was his observation that “At slow rates of transmission subjects used few adjustment processes. At mediumlrates they attempted them all. At higher rates filtering was preferred, but as the ma&mum channel capacity was reached, both subjects used chiefly omissions’* (Miller, [55, p. 941). Additional support is given to this proposition by the empirical study of Lanzetta and Roby (431, the case study by Meier [52], and the review by Driver and Streufert [25]. PRomsmoN M.7. The probability or extent of message modification is inverseh related to the cost that the sender expects to incur as a result of making the modif?cation. If cognitive limitations were the cause of message modification, and if the anticipated cost to the sender of these modifications was high, then it seems likely that the sender would put forth a greater effort not to make errors and that the modification

M.8. 7%

related to the number 9: receiver to the message SC

This phenomenon example, Kaufman civilians by Americ military hierarchy t levels had no idea o: for transmission of declared that “eve: (Kaufman, 142, p. :

Message summz military outpost C

L

command the nun truck traffic count truck traffic even average number c taken in summar

and M.7 return again to modij5cation 13 positively expected or &sired ;d

cognitive limitations and ‘y messages during assimidirection of identity with latic biases” @. 346), and I input in the direction of s interview study, Downs partially screen out data e to their interest” [24, p. transformed in the direcrhes. McLeod (51, p. 2181 lation is less if the sender 8 modijication is positively lessage is bared.

jsition in his review article nplex stimulus, the subject omplex or ambiguous the is already ‘in’ the subject )rter and Roberts, in their results would indicate that rhe more likely it is that :ommun.icating accurately, JO end feelings there is

would be less. Empirical support for this belief is provided by Adams and Swanson [I].

Closely related is the idea that if motivations were the cause of message modification and if the anticipated cost of making them were high, then modification would be less. One contribution to anticipated cost would be the knowledge that the receiver could get the correct message from another source, and would hold the sender accountable for having made the modification. For example, Downs noted that “use of redundant information channels increases the probability of obtaining accurate information” (Downs, [24, p, 2691). The empirical support for this latter idea is more equivocal than one might expect (cf., Hsia, [37]; Anderson, [S]). PRoPOsXTtON M.8. The probabiliq or extent of message modification is positively related to the number of sequential links in the communication chain connecting the receiver to the message source. More links in a communication chain provide the opportunity for additional distortions to occur. That these additional distortions do occur was a central finding of the rumor transmission research by Bartlett [9], Allport and Postman [4] and Higham [36]. Other writers have also addressed the matter (cf., Downs, [24], p. 2691; and Miller, (561. In their discussion of organizational innovation, March and Sunpn stated that “Selective filtering takes place not only at the boundary of the organtzation, but at every state in the transmission and elaboration of program proposals” (March and Simon, [49, p. 189]), and Downs concluded that “When information must be passed through many officials, each of whom condenses it somewhat before passing it on to the next, the final output will be very different in quality from the original input; that is, significant distortion will occur” (Downs, [24, p. 2691). Smart and Vertinsky (711 have addressed these same ideas in the context of decision making under,crisis. A dramatic example of what Downs was describing is the following: A reporter was present PI a hamlet burned down by the U.S. Army’s 1st Air Cavalry Division in 1967. Investigation showed that the order from the division headquarters to the brigade was: “On no occasion must hamlets be burned down.” The.‘bripde radioed the battalion: “Do not burn down any hamlets unless you are absolutely convinced that the Viet Cong are in them.”

P mocijkation is positively

The battalion radioed the infaavy company at the scene: “If you &ink there are any Viet Gong in the hamlet, bum it down.”

r cognitively or logistically

The company commander orderedhis uoopr: “Bum down that hamlet” (Millet, [56, p. 69D.

early literature review led lperate at near maximum of message complexity-is 11, [55] has given considerle variety of modifications, es, are used to deal with it. transmission subjects used J them all. At higher rates tpacity was reached, both Inal support is given to this .3], the case study by Meier :e modification is inversely of rrraking the modification.

dication, and if the anticihen it seems likely that the I and that the modification

This phenomenon takes place in upward as well as downward communications. For example, Kaufman notes that “An official study of a mass kiIling of Vietnamese civilians by American troops disclosed that at each successive higher level in the military hierarchy the reported number of victims was reduced, so that the highest levels had no idea of the extent of ‘the tragedy despite two separate command channels for transmission of news about events in the field. A field commander subsequently declared that “every large combat unit has similar episodes ‘hidden somewhere’” (Kaufman, 142, p. 14J). 6 . S-n Message summarization can perhaps best be illustrated with an example. If a military outpost monitors enemy truck traffic, it may report to some higher level command the number of trucks seen each day. The higher level command, recei+g truck traffic counts from many such outputs and having to communicate a measure of truck traffic even further upward through the chain of command, may’ well report the average number of trucks per day observed by this outpost. The fit step has been taken in SUmmarizB tion-a frequency distribution has been reduced to a descriptive

.

11

,

150

GEORGE HUBER

I ‘

’

1 t

i I

i I I i

t

i

statistic, in this case the mean. The next higher echelon will receive many such reports from its several subordinate units and may be required to conclude ðer enemy truck traffic in its area is increasing or remaining constant. When it does this, the second step has been taken-a statistical inference has been draw, a mean value has been encoded as a 1 or a 0, depending on whether the null hypothesis was rejected or accepted. The echelon receiving this coded inference may receive such inferences about the increase in truck traffic from many subordinate units, and may pass upward the descriptive statement that “in twc+thirds of the sectors polled, truck traffic has increased.” Here, data in the form of binary digits were combined into a descriptive statistic, a proportion. It may well be that proportions dealing with other variables such as rail traffic, construction of artillery sites, etc. would also be received by a still higher echelon and would be subjectively combined with the truck traffic proportion to facilitate the drawing of still another inference concerning whether or not the enemy is in the process of a major buildup of its forces in a particular region. Depending on the number of echelons involved, we can envision the continuing repetition of the following cycle: (1) data are combined into a descriptive statistic, (2) the descriptive statistic is compared to some standard and a statistical inference is drawn, and (3) inferences are treated as data. It is not necessary that only one step in this cycle takes place at any particular node in the communication network, or that a summarization takes place at every node. Of course summarization also occurs with qualitative information. For exampie, information obtained in an employment interview is subjectively aggregated and transformed into a descriptive rating, the rating is compared to some (perhaps implicit) standard, and a conclusion is drawn concerning the acceptability of the candidate. As we noted earlier, message summarization is a process purposefully employed by organ&dons and tends to have as an outcome a faithful representation of the original meaning These features are generally in contrast to those of message modification. Message summarization greatly reduces the cognitive and logistical loads on organktional units. In addition, in that the conclusions that follow from inference-drawing become guidelines and directives for organktional action, it aids considerably in planning and coordination (cf., the discussion of “uncertainty absorption” in March and Simon, 149, pp. 165-1661). For these reasons, message summarization is a pervasive organizational process. In spite of its pervasiveness, however, there is very little empirical literature that deals with the subject The relative shortage exists in contrast with the abundance concerning message modification. Perhaps the disparity exists because findinga of successful summarization, i.e., condensation without distortion, are less tantalizing than are findings of modification, e.g., alteration with distortion. Whatever the reason, there is very little literature on which to base propositions about message summarization. I have chosen, nevertheless, to offer four propositions on the subject. I hope that their presentation will lead organizational scientists to undertake empirical studies of information summarization and that it will heighten the sensitivity of organization designers and analysts to this organizational process. There are four variables that it seems reasonable to believe would affect the summaking of information in or&tions: (1) savings in transmission costs, (2) cost of summarGng the message, (3) workload of receiving unit, and (4) number of links in the communication chain. It is important to note that “costs” include the expenditure of any resource, including time, space, and intellectual effort. S.1. The probabiliry or exzent of message summarization is positively related to the perceived savings in twwnission costs obtained from summarizing tk message. hOPOSmON

It seems reasonable to message than an (2) the greater the summarization. On the senders seek and use cases, transmission Rather, only (1) where are transmission costs here the fact that the prc PROPOSZION s.2. The related to perceived cost

Working against the requires time and effo associated with transmi: sion of unsummarixed observed by Kaufman managen to the U.S. Proposition S.2. Because of the would expect an and the cost of difficult to validate the dependent variablt relationship between difficulty of successfu: valid may explain relationships. PROPOSITION S.3. related to the perceivec

A number of extent if it is known the sending unit message in accordant thought that the be desired, leads us by the sender and observation of becomes overloaded so &at the lower messages’* (Downs, of its workload reduc receiver. ~ROPOXT’ION

related to tk receiver to tk It seems reasonat communication cha took place at [9j, Allport and

receive many such reports j conclude whether enemy ot. When it does this, the .n *’ w, a mean value has lesis was rejected or 4. such inferences about the nd may pass upward the polled, truck traffic has ,,mbined into a descriptive Aing with other variables i also be received by a still he truck traffic proportion 4 whether or not the enemy :ular region. n envision the continuing o a descriptive statistic, (2) d a statistical inference is :ssary that only one step in nication network, or that a urformation. For example, .bjectively aggregated and 1 to some (perhaps implicit) ability of the candidate. purposefully employed by presentation of the original ‘: of message modification. ogistical loads on organixaJW n inference-drawing aids considerably in )n, inty absorption” in March ssage summarization is a less, however, there is very relative shortage exists in ttion. Perhaps the disparity mdensation without distorttion, e.g., alteration with zrature on which to base jn the subject. I hope that lertake empirical studies of sensitivity of organization believe would affect the I transmission costs, (2) cost , and (4) number of links in

ts” include the expenditure XT. sutnmari2ution is positive+ ined from summarizing the

It seems reasonable to believe that (1) if it is less costly to transmit a summarized message than an unsummari& message, then the sender would summa&e, and that (2) the’ greater the reduction in transmission cost that is possible, the greater the summarization. On the other hand in today’s technology where high volume message senders seek and use very low cost transmission machines, it may be that, in many cases, transmission costs are not important determinants of message summarization. Rather, only (1) where they are large and (2) where the sender is accountable for them, are transmission costs likely to be important dete rminants. In view of this, we highlight here the fact that the proposition deals with perceived savings.

PROPOSITION

S.2. The probability or extent of message summarization is inversely

related to perceived cost of summarizing the message.

Working against the impact of Proposition S.l are the facts that sum.marixing requires time and effort, and that the resultant costs could exceed the savings associated with transmitting a summarized message. This could lead to the transmission of unsummarized messages even when transmission costs are significant, as observed by Kaufman in his study of the communications from individual forest managers to the U.S. Forest Service (Kaufman, (42D. This idea is captured in Proposition S.2. Because of the fact that summarization does take time and effort to carry out, one would expect an inverse relationship between the extent of message summarization and the cost of message summarization. If this inverse relationship exists, it would be difficult to validate both propositions, since S.l posits a positive relationship betwetn the dependent variables and the extent of summarization and S.2 posits an inverse relationship between the dependent variables and the cost of summarixation. The difficulty of successfully validating these two propositions, even if both are in fact valid, may explain the lack of reported empirical studies dealing with the stated relationships. PROPOS~ION S.3. The probability or extent of message sumnuui2ation is positively related to ,tk perceived work&d of rhe receiving unit.

A number of arguments suggest that the message will be summa&cd to a greater extent if it is known that the receiving unit is heavily loaded. One argument is that if the sending unit wants the message to have an irnpacf it will attempt to format the message in accordance with the desires of the receiving unit. This, combined with the thought that the greater the receiving unit’s workload, the more summarization would be desired, leads us to conclude that a relationship would exist between summarization by the sender and workload of the receiver. A final line of reasoning follows from the observation of Downs that “when the topmost level of communication intermediaries becomes overloaded for any reasou, it can react . . . by changing the tntnsmission rules so that the lower levels in the network screen out more information before sending messages” (Downs, [24, p. 129D. Thus, an overloaded receiving unit might have some of its workload reduced with a directive that sending units condense their inputs to the receiver. PnomsmoN S.4. The probability or extent of message eation is positive& related to tk number of sequential link3 in tk commum ‘cation chain connecting tk receiver to the message source. It seems reasonable to believe that the greater the number of sequential links in a communication chain, the more intense would be the effect of a phenomenon that took place at each link. Evidence supporting this is suggested by the work of Bartlett [9], Allport and Postman [4], and Higham [36].

:

--- _--

_

-7

GEORGE HUBER

152

ORG.%

7. summary

2.

This paper reviewed the literature concerning the determinants of the performance and behavior of organizational information processing systems. In most instances, these detemlinan ts and their impact were identified from an examination of the empirical literature, and thus most of the propositions are supported by this literature. The review did identify two areas, however, where there seems to be an especially acute need for further empirical research. One of these concerns message routing The presently available literature related to message routing deals primarily with informal and not necfss(iTily work-related messages. It may be that, in some instances, empirical studies of the more formal&d information systems that deal primarily with workrelated messages would lead to somewhat different propositions than those that followed from the presently available literature. The second area in particular need of further empirical study concerns message summa&&ion. The literature pertaining to this topic is extremeiy sparse. It seems that the results of studies dealing with message routing in formal settings and with message swtion would be quite useful to the designers and managers of information systems, since the availability, form, and meaning of messages are such important determinants of. the quality of organizational decisions, and hence of the effectiveness and viability of the organization itself. There is clearly much research to be done before management scientists can tell managers or information systems designers how to prevent or resolve any information system malfunction that might occur. I hope that highlighting this fact fl accelerate the amount of research that wiIl be brought to bear on the matter. In the meantime, however, management scientists are being called upon to address problems that have deadlines for solution. They cannot ask the clients to wait for future research results, but mu& work with the knowledge that is available today. By searching out and gathering together this knowledge, by summarizing what we know, or think we know, about the determinants of the performance and behavior of organizational information systems, I have attempted to take a first step in responding to this need. Two matters seem to be worthy of special mention at this point. One is that the forthcoming inclusion of extremely sophisticated computing and comminications technology, irito our organizational information systems may affect the nature of the impact of the determinants identified in the review presented here. Or, it may generate additional determinan ts. Thus there will be a need for further research of a relatively basic nature. The second matter concerns a more immediate need, the need for applied research to respond further to the needs of today’s managers and systems designers. I believe we can expect to see, and should see, studies that include the following steps: (1) developing normative guidelines for the design and management of information systems, guidelines based on behavioral research such as that reviewed here, (2) testing, revising, and retesting the usefulness of these guidelines in the actual design and management of organiz&ional information systems, and (3) reporting the results of these teats so that when the guidelines are valid, they may be adopted by others and so that when they are not, further and more targeted research can be initiated?

3.

‘The ruthot would like to thank Chula O’ReilIy, Daniel Power, Michael Tubman, and the reviewers for their vay UsefIll comments on 8n earlier draft. This reacuch was supported in put by the Army Research Institute for the Behavioral and SociJ Sciences end in part by the National Science Foundation’r Division of Infonnatioo Science and Tcchnolo@ Relereoces J. R. AND SWANSON. L A., ‘Information Procuing Behavior and Estimating Accuracy in @wations Managemen&” Acnd Mono- I.. VoL 19. No.1 (1976). pp. 98-110.

t. ADAMS

4. 5. 6. 1.

ALKlm~c0UUKM”KrJ\ Behavior Under Social Powe. AUZN, T. J. A N D Cow. S. Ad&IL sci. Quar¶. VOL 14, > Aupoin. G. AND mu, L. And-n, J . A_ “Single Ch. Meaning” Aua?o-VinKLl corr ATHANASSUDES, J. C, me I Acad Matulgmenl 1.. VOL 1

BACHAUC& S. B. AND Aum meru 1.. Vol. 20, No. 3 (197‘

8.

BARNLUND, D. C. IWD HARL Networks,” Sociomcrry, Vol 9. Btiwztr, F. C, Rmmmbenn London and New York_ 19: 10. Beu, D., ‘“Thi&ag Abeah” 11. BILENNEX. M. H. A N D SICX Emptid Da&&” Acod .wC 12. BRA J. S.. “On Pauprur 13. CANF.B~U, D. T, Systnaa Information and CunffoL Vt 14. CAJWBELL J . P . m PRJT Psycholtigy.” in M. D. Du McNally. Chicago, IJL 19‘ IS. CAR- R K, “clients R; Evaluation Studies,” Amer 16. CARTEL. R. M.. Cow?. Company. Book Towet, C 17. CHaD, I., ‘0quizatiooaf Sorio@, Vol. 6, No. L (I 18. COHEN. A. R “Upward c Vol. 18 (1958). pp. 41-53 19. Cows, B. E m G~rn Wiley, New York. 1%. 20. CONIATH, D. W.. “Comm: Managemenl Sci. VoL 2C 21. CvMlwO~ L. L. em HA: 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 3s.

tions, Houston. Tcx. 19F Cuat. R M. AND MARCI NJ., 1963. D~vu. K., “A Method of (1953), pp. 301-312.

DOWNS A.. hide Burrmc* DRNEII, M. J. AND Snurr Informalion Pwxsing j DUNCAN. R B.. “Multiple fmpacl on Org?nization F.uuc& R. V. AND MAC tion,” Pmonnei P3’ch. FEIIWCE. T. P.. “Grw= Sci.. Vol. 17 (1970X pp F~GER, L.. SCHA~R Pmjecr. Harper, Nerw \ Guurrit,J.K..Tlu~ Gwun-fr, J. R.. 0qm Gnsnsnt~. A. *D BE Information.” Monrrgrr GOLD- J. D, BUG* Tc~hn~lo&ai fnnovat Gouq W. I.. ‘Admixus (1956). pp. 281-291. Gunzxow. H, -Cow Rand McNally. chi=

ORGANIZ ATIONAL INFORMATION SYSTEMS 2.

nants of the performance ,ter+- In most instances, l xamination of the pported by this literature. ,eems to be ‘an especially ems message routing. The Is primarily with informal some instances, empirical LeaI primarily with work>ositions than those that I study concerns message zmely sparse. It seems that settings and with message managers of information isages are such important hence of the effectiveness

sgement scientists can tell or resolve any information ing this fact wilI accelerate 2 matter. In the meantime, :ddress problems that have for future research results, .ay. By searching out and e know, or think we know, organizational information 3 need. :i th_ _ lint. One is that the g and comminications techf affect the nature of the zd here. Or, it may generate [her research of a relatively .e need for applied research systems designers. I believe :de the following steps: (1) zanagement of information as that reviewed here, (2) delines in the actual design ud (3) reporting the results iy be adopted by others and .ch can be initiated.’

3. 4. 5. 6. 7. 8. 9. 10. II. 12. 13. 14. IS. 16. 17. 18. 19. 20. 21. 22. 23 24. 25. 26. 27. 28. 29.

:ael Tushmaa, and the reviewers for _.

30. 31. 32.

ute for the Behavioral and Social lfonaation Science and Technology.

33. 34.

Tavior and Estimatiag Accuracy in 1976). pp. 98-110.

35.

A>,,

w A-. Couuu, M.. Kxsw.u& J. AND Love. L, ‘Iafon~tioa Exchange md Accuracy of Verbal Behavior Under Social Power Condition* 1. PcrJowlity Social Plyrh. VoL 9 (1968). pp. 301-308. 4LLm, T. J. AND COHEN, S. I.. ‘Information Flow ia Research aad tkvclopmeat bboraton-~dmin Sci Qwrr. Vol. 14. No. 1 (1969). pp. 12-19. eLpo~r, G. AND Posx~~. L, Qc Basic Psychology of Rumor,” Trans. New York ~cod Sci., 1941. ~dersoa, J. A.. Single channel aad Multi Channel Maaaga: A Comparison of Connotative Meaniag” ~Udi~Visuof Comm Rev.. Vol. 17 (1%9), pp. 428a34. A~~ANUSLUIU, J. C., ‘“The Distortion of Upward Coaxnunicatioa in Hierarchical Mtions.” Acad Management 1.. Vol. 16, No. 2 (1973). pp. 207-226. B~cn~~cti, S. B. AND m, M.. “Communication ia Administrative Bureaucracies,” Acad MaMgc menr 1.. Vol. 20, No. 3 (1977), pp. 365-377. BARNLLI?Q D. C AND HAJUAND. C.. “Propiaquity aad Prestige as Determinants of Communication Networks,” Socionutry. Vol. 26 (IW3). pp. 466-479. BARB, F. C., Ronmtbwing, A Sru& in Experimental and Social P@obgy, Cambridge Uaiv. Press, London and New York, 1932. Ba D., vg Ahead,” Harvard Businuc RN. (May-June 1979). pp. 20-42. B-t& M. H. A N D SlGBAND, N. B.. “Grganiutiond &xnmuni~atioa-Aa Analysis Based on Empirical Data,” &ad MoMgemenr /., Vol. 16. No. 2 (1973), pp. 323-324. BRIJNER, J. S., “On Perceptual Readiness,”P+&gical Rev., Vol. 64 (1957). pp. 132-133. cA&QB~ D. T., “Systemntic Error on the Part of Humaa Links ia Commuaication Systems” fnfommtion and ContmL Vol. I (1958). pp. 3-369. CuSBeu. J. P. m PIUTCHARD. R. B.. “Motivation Theory ia Industrial aad organizational Psychology,” in M. D. Dunnctte (Ed.) Handbook 01 Indtu~rial and Organizational Plychofogy, Raad McNally, Chicago, III.. 1976. pp. 63-130. CARTF,R, R. K., “C&au’ Resistance to Negative Findings and the Latent Conservative Function of Evaluation Studies” Amer. Socio/o@. Vol. 6 (1971). pp. 118-124. CARP& R. M., Co- ‘carions in Organizerions. A Guide IO Informofion Sourcr~, Gala Research Compmy. Book Tower, Detroit. Mich.. 1972. Cws. J., “OrgpnLatioaal Structure, Enviroment and Performance, The Role of Smtegic Choice.” Sociology, Vol. 6. No. I (1972). pp. l-22. COWN, k R, ‘Upward Communication in Experimentally Created Hierarchies.” Human Relotionr. Vol. 18 (1958). pp. 41-53. Coums, B. E. AND Gut?tz~ow. H. A.. A Social Pyhology of Gray Pmusu fur De&ion Makbtg. Wiley, New York, 1964. C~NMT& D. W., “Commuaialtonr Environment aad Its Rclrtioaship to Grgaaizatioaal Structure,” Mana&men~ Sci. Vol. 24 No. 4 (1973), pp. 586-603. CU~~WGS,LLANDH~, D. L. Bargaining Beinviar: An Intemtiomi SW. Dame Publicatioas, Houston, Tex., 1980. CYERT, R. M. AND M ARCH . J. G.. A Behavioml7%oty oj the Firm, Prentic+Hti. Eaglewood Cliffs NJ., 1963. DAWS, K., “A Method of Studying Communication Patterns ia @ganizstioas,” PC-~ PJych4Io&y (1953). pp. 301-312. Dams, A_, Inride Bweaucraq. Little Brow Boston. MPU, 1964. T, S.. “Integrative Complexiry: Aa Appfr~cb to Individuals and Groups as Duvzu.M.J.mS Iafonaation Pmcasiag 7 stems,” A&a Sci Qwrt. Vol 14. No. 2 (1969). pp. 272-285. DUNCAN, R. B., uMultiple Deoi&a-making Structures ia A&ping to Environtneaul Uncertainty: The Impact oa vtioaal Effectiveness,” HWWI R&ions. VoL 26 (1973), pp. 273-291. FAUCE. R V. m hhCDoNALD. D:. ‘New Directioas ia the Study of vtioaal Communication,” Perso~4 Psj&, Vol. 27 (1974), pp. I-19. Famrce T. P., yOrganizatioaal Commuaicatioas Systems and the Decision Rotas”ManagSci, Vol. 17 (1970). pp. B83-B%. F-G= L. SCHACKR. S. *ND BACX, K.. SoriakPruwcs in in/annrJ Grasp: A St* of a Howing Project. Hatper. New York, 1950. G-w J. K.. Thr Nm Indwtriaf State, Houghton Miiflis Boston. Mass., 1971. G~~RAIT& J. R. O+atiun Design. Addison-Wesley, Rdit~g, Mm 1977. Gs, A. AM) BURGER, P, ‘An tiysti of U~tioo Differencea for Scientific aad Technical Iaformatioa.” Managewmf Sci. VoL 26, No. 2 (1980). pp. 165-179. GCI~DHAM, J. D., BRAGAW, L. K. AND Sctiw~n. J. J, ‘Information Flows, Maaagement StyI= nnd Technological Inaovatioa,” IEEE Tratu Enginebng Mana-, Vol. EM-23 (1976). Golle. W. J., ‘Ad&t&ntive Decision-Makiag in Federal Field office&” P&c Admin &., Vol. 16 (1956). pp. 281-291. Gmow, H., “f&~unic.&oa~ ia Organizations,” in J. G. Much (Ed.), Hun&ok 0~0rgMiLofioN. Rand McNally, Chicago, 111.. 1965.

I

I

i 1 , I

ORGA

GEORGE HUBER

154

36. HIGHAM, T. M, 7he Expcrimcnlal Study of the Transmission of Rumour,” Bririrh 1. Psych (1951). pp. 42-55. 31. HSIA, H. J, “Gn Channel Effectiveness, Audio-Yinrol Comm Rev.. Vol. 16 (1968). pp. 245-261. 38. HUBHI, G. P.. ‘Gqanizational Decision Making and the Design of Decision Support System” Ma~grmmr Infonnryion Sysrmu Quarr.7 Vol. 5, No. 2 (1981). 39. JAIN, H. C, Jupcrvisory Communication and Performance in Urban Hospitals” /. Comm. Vol. 23 (1913), pp. 103-111. 40. J-N, D. B, “Stnt&c Decision Making InEiucn~~ in BOU~~UY Spanning RoIcs,” Acad MUM~Cmcnl Proc, (3979). pp. 118-122. 41. JOHNSIUN. R AND GmBooNs, hf., “Characteristia of Information Usage in Technological Innovation,” IEEE TIWU Engineering hha~enwnt, VoL Em-22, No. I (1915), pp. 21-34. 42. I(AupyAN. H, Adminisrrarive Feedbuck ‘The Brookingr Institution, Washiigton, D.C., 1913. 43. -A,J. ‘I: m ROBY, T. B.. “Effects of Word-C+up Structure and Grtain Task Variables on Group Performance,” J. Abnormal Social &vch. Vol. 53 (1957), pp. 307-314. 44. L\m E, PORTI?& L AND TENZNBAUN, A., “Managers’ Attitudes Toward Interaction Episodes,” 1. A& Psych, Vol. 52 (1968X pp. 432-439. 45. L,A~ P. R, and L.oascn, J. W., O+arion and Emkvvnent, Graduate School of Business Administration. Harvard Univenity, Boston, Mass., l%1. 46. Loasrx, J. W.. Qntingcncy Theory and Gqanization Design: A Ptnoaal odyssey.” Chapter 8, R H. Kilmm. L R Pondy. and D. P. Skin (tds.), The Managemenl o/ Organization Design, NorthHolland, New York_ 1976. 41. LOWE, A. E AND SHAW. R. W., ‘An Analysis of Managerial Biasing: Evidence from a Company’s Budgetins Proceu,” 1. iUanugemenf S&es. Vol. 5. No. 3 (1968). pp. 304-3 IS. 48. Loorag J. I_., ‘Communication, the Dcvelopmcnt of Tnut. and Ccapcrativc Behavior,” Hunan R&tio~. Vol. 12, No. 4 (1959), pp. 305-315. .’ 49. M_ucn, J. G. AND SDION, H. A., Orgunizationr, Wiley, New York, 1958. 50. MCCL&UY, R, -How Parole Gffii Use Records,” Social Pro6lems. Vol. 24 (1917j, pp. 576-589. 51. McLEOD, J. M., 7he Contribution of Psychology to Human Communication Theory,” in F: E. X. Dance (cd.) Hunwn Commwukation Theoty: Originai Eksayx, Halt, Rinehart and Winston, New York, 1967. 52. MEIER, R L., “Communications Overload: Proposals from the Study of a Uaivenity Library,” Admin Sci @IV?, Vol. 4 (1963) pp. 521-544. 53. M~Ex_ G, “Inkrpersonal Trust as a Factor in Communications,” /. Abtwnd Social kych. Vol. 52 (1956-), pp. 304-m. 54. m J. G, ‘Information Input Overload and Psychctpatbology.” Amer. /. Pzychior~, Vol. II6 (19&l), pp. 695-104. 55. -,“Adjusting to Ovcrloada of Informtion,” in D. M. Rioch and E. A. Weinstein (cds.), DLcordrn oj Communica&n, Research Publicationa, Association for Research in Nervous and Mental Disordm, Hafacr, New York, 1964. 56.. -. “Living Systems: The orgnintion,” Behaviors Sci., Vol. Il. No. I ( 1972). 51. O’RmLy. C ?lte Intentional Dktortion of Information in Organizational Communication: A Laboratory and Field Approach.” Huwmn Rclationr, Vol. 31 (1918), pp. 113-193. 58. -AND m; L, wtional Cummunication,” in S. Kerr (Ed.), Organiz~iond Behmiw, Grid, cobmbw. Ohio, 1980. 59. -AND ROBEX~ K., “Information Filtration in Organktions: Three Experiments,” Organiza!k7dBdkd?dH-Prrformmr e. VoL 1 I (1974x pp. 153-265. 60. ppeppm. J, Pmw in Oqandatti, Pitmsn. Marshfield, Mass., 1980. 61. PO~TP& L W. AND ROIQYS, K. H, ‘Communication in Gqanizations,” in M. D. Dunncttc (Ed), Hadbock of Industrial ad Orgdzatiod P~Mogy, Rand McNally, Chicago. Ill., 1976. 62. m, W. M., ‘Upward Cummunicatioa in Industrial Hierarchies,” Human Rebdou. Vol. I5 (I 962). pp. 3-15. 63. ReDDmo. c. W., chmnniC ation within th Organization. Industrial Communication C&cil, New York 1972. 64. ROBERT K. H. AND O’REULY, C. A., “Failures in Upward Commuakatioa in Orga~G=tions: Three Pcuiblc Culprits,” Acad Mmgm J.. Vol.- II (1914). pp. 205-215. 65. ROSFN, S. AND Tm A., “00 Reluctance to Communicate Undesirable Information: the MUM Effect,” SoeromCrry, VoL 33. No. 3 (1970). 66. -t J OHNSON, R D, JOHNSON, M. J. AND Tw A_, me Interactive Effccu of News Valence and Attractjon on Co~~~~uniutiot~~ Behavior.” J. PerxomafiiQ Social PJycir. VoL 2%-(1913). pp. 298-300.

AND Lmm, A. Y., “where Traditioaal Bud&n8 tiaib,~ Fhawiul Exrcurivr, VoL 26. No. 5 (1968), pp. 51-62.

67. SCHIFF, M.

68.

SCHI-FF,M.ANDL~~P~.~’ (1970). pp. 259-268.

69.

SHEPARD. J.

10.

M. AND HOCG Organization?“,” Acad .+lSIMON, H. A., “Applying Inf

71.

pp. 268-278. SMART. c. AND vExtmsKY. : pp. 640-657.

12.

SUllDN,

H. S. m &RE Personnel Psych, Vol. 21 13. r~YEll, L.. “communicau fieory: Or@71 ties. F 74. TUSHMAN, M., “Sperial 60: (1977). pp. 587-605. -, *Work Character Aahin Sci. Quart.. Vol. I 16. - AND NADLER_ D. .Managemeti Rev., Vol. > 77. ULLMAN, J. C. AND HL~ER

75.

18.

Heati Lexington. Mas Voos, H.. Organizario& 1961. H.. Wwk and

19. 80.

VROCJM, v.

81.

WICXESBERG,

WMSON,

D. m BROVB

so&l Pyh, Vol. 2 ( I ( A . K, -( Manugemenr 1.. VoL 11 82. W~ENSKY, H. L, Organ Books, New York. 1% 83. ZUONC. R. AND WObon.” Hwnsn ReLrior. 84. ZAND, D. E., -Trust an< 229-239.

ORGANIZATIONAL MFORMATION SY!5l-EMS

.umour,” Btirish /. Psych (1951).

4.

01. ” ’‘968). pp. 245-261. ision support system.&*-0

69. 70.

zn Hospitals,” 1. Comm, Vol. 23 71. Spanning Rob,” Acd Managt72. age in Technological Innovation,”

‘p, 27-34. v’asbington, DC., 1973. -e and Certain Task Variables on 1. 307-3 14. Toward Interaction Episodes,” J.

73.

mr. Graduate School of Business

76.

,~nal odyssey,” Chapter 8. R. H. lt of Orgcnirarion Design, North-

77.

74. 75.

78. .ing: Evidence from a Company’s

pp. 304-315. 1 Cooperative Behavior,” Hum0

79. 80.

1958. a, Vol. 24 (1977). pp. 576-589. nmunication Theory,” in F..E. X. Rinehart and Winston, New York,

81. 82. 83.

.y of a University Library,” ~dmin 84. m4” J. Abmmal Social Psych, Vol. sy.

x J. Pq&iatry. Vol. I16

nd E. A Weinstein (e&z.). Disorders rch io Nervous and Mencal Disor17, No. I (1972). )rganizationaI Communication: A 8). pp. 173-193. lerr (Ed_), Organizariond Behavior, ns: Three Experiments+” Organiza265. 30. rations,” in M. D. Dunnctte (Ed.), dally, Chicago, Ill.. 1976. .*’ Hwmn Rrhiom, Vol. I5 (1962). rial Communication Council, New muniution in Organizations: Three -215. ndcsirable Information: the MUM nteracrive Effects of News Valence sa&I P-h, Vol. 28 (1973). pp. s,” Finmviol Execurive, Vol. 26, No.

155

SCOW. M. AND LEWIN. k Y.. me Impact of People on Budpq” Acematting Rev., VoL 45. No. 2 (1970). pp. 259-268. SHEPAICD. J. M. AND HOUGLAND. JI, J. G.. ‘Contingency Theory: Qmpla Mm’ or “Cornpica Grganiration7”.” Acad hiamgement Rev., Vol. 3. No. 3 (1978). pp. 413-427. SIMON. H. A.. “Applying Information Technology to Gqaniratiott Design,” public Admin Rev. (1973), pp. 268-218. SMART . C. AHD Vumssx. I.. “Designs for Crisis Decision Units,” A&in Sci Quart., VoL 22 (1977). pp. 640-657. %lTl-ON. H. S. A ND PORW L W., ‘A Study of the Grapevine in Governmental orgaoiution.” ksonntl PJych, Vol. 21 (I%&, pp. 223-w). THAYER, L.. “Communication and Grganiration Theory,” F.E.X. Dance (cd.), Humn Communication 77woty: Original Ewoys, Holt, Rinehart and Winston, New York, 1967. TUSHLUN, M., uSpecial Boundary Rola in the Innovation Procey” A&tin Sci Qwzrt.. VoL 22, No. 4 (1977). pp. 587-605. -, ‘Work Characteristics and Subunit Communication Structure: A Contingency Analysis,” A&in Sri. Quart., Vol. 24, NO . I (1979), pp. 02-90. -AND NADL,?R, D. A., “An Information Prccusitq Approach to Gr~tionaJ Design.” Aced Managewtenf Rrv., Vol. 3, No. 3 (1970 pp. 613-624. ULLQAN, J. C. AND HUB= G. P.. Thr Local Job Bank Program Pqformon~e, Stmfwe ond Directionr. Heath. Lexington, Mass., 1973. VOOS, H., Organikationol Commumkatotr A Bibiiog@y, Rutgers Univ. Press, New Brunswick NJ.. 1967. Vnoou, V. H.. Work and Motivation. Wiley, New York, 1964. WATSON, D. - Bnohtszsto, B, “Power, Communication and Position Satisfaction.” J. PersomIig Social PJyck. Vol. 2 (1965). pp. 059-864. WICKESSERG, A. K.. “Communication Networks in the Business Grganization Structure,” AC& ~anagemenr 1.. vol. 1 I (I%@, pp. 253-262. WRENSKY, H. L., Organizational Xnrelligence: Knowledge and Policy in Govmnwwr~ and Industry, Basic Books, New York, 1967. ZAJONC, R. AND WOLF& D.. ‘Cognitive Consequences of a Person’s Position in a Formal Gqanization.” Hwnm Relarioru. Vol. 19 (1966). pp. 139-150. ZAND, D. E, ‘“Trust and Managerial Problem Solving,” Admin Sci Qwrr., VoL 17, No. 2 (1972). pp. 229-239.