University of Groningen

Automatisering in bedrijf. Een empirisch-theoretisch onderzoek naar de effecten vanautomatisering op de functiestructuur van bedrijvenBatenburg, Ronald Simon

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.

Document VersionPublisher's PDF, also known as Version of record

Publication date:1991

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):Batenburg, R. S. (1991). Automatisering in bedrijf. Een empirisch-theoretisch onderzoek naar de effectenvan automatisering op de functiestructuur van bedrijven. Thesis.

CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.

Download date: 02-01-2021

Summary

The impact of automatization on job structures has been the subject of abroad discussion since the seventies. The origin of the discussion l ies in thecontroversy between optimists and pessimists. Optimists point out theopportunities of new technology, such as new jobs, the demand for new skillsand the improvement in labor conditions and career development of workers.Pessimists stress that automatization will lead to degradation of work andpolarisation of job structures leading to negative career changes for lessskil led workers. At the level of the organization, the debate was concentratedin particular around the finding that job structures in Íirrns tend to to besrmultaneously upgraded and downgraded.

The dm of this study was to arrive at a general conclusion on (a) thcimpact and content of job stnrcture changes after automatization in firms and(b) the most prominent deterrninants of these changes.

Changes in job structures can be defined as the result of changes in work(the content ofjobs). Through automatization jobs can change in four ways:(l) Jobs are downgraded (the level of skil is diminish);(2) Jobs arc upgraded (the level of skil ls improve);(3) Jobs are created at a certain skill-level in the organizatíon (job g,ain) and(4) Jobs are removed at a certain skill-level in the organizatton Qoh loss).

To rnark out the resulting changes in the job structure of firms the expla-nandum of this srudy is l imited to:(A) Changes in the mean skill level of firms, namely downgradin,g andupgrading on job structures) and(B) Changes in the division of skill levels in the firm namely polarisarion.

Existing empirical research results can be divided into those supporting thedegradation or polarisation hypothesis, those supportrng the upgrading orconvergence hypothesis and those supporting the hypothesis that automatiza-tion is not at all related to changes in job struchrre. For the greater part, thisresearch is based on case study material. The constant contradiction betweenthe outcomes of research is striking and persistent. Case studies on theeffects of CNC-machines in metallurgic lactories are well known because oftheir controversial outcomes (Bilderbeek e.a., 1985).

The studies which support the hypothesis that automatization does rolaffect the job structure of organizations have influenced the central question

132

of this study. Beforecould ask why strurquestlon was refocustchance of regradatiortion of empirical supersists. The absenctIimited generality of ,question why, in appttion differ.

Therefore a theolwhich predict in whitpected. The sociologytions. Different apprchave been more popuquestion, we selected

Ihe first is the terwork of Woodwardorganrzational structurby members of the Astnology and type of orgnot be confirmed. Thrctzed on theoretical grintroduced the idea ofresearch today. In shrcertaln technological ctives to (re)organize rconcept caused an imdeterrninism in organiz

The contingency aJ.tional studies. This apBurns & Stalker and Lplaced on the influencThe stability of the (nirtures more successful tidea elaborated by Buigreater the stability of tto automatization. Althcthis hypothesis is the crmarket, such as publi<approach concems thethat the only differencedeterminism lies in thetions determine organizmarket conditions. Morout into concrete hypotl

ob structures has been the subject of a. The origin of the discussion lies in thed pessimists. Optimists point out the:h as new jobs, the demand for new skillsions and career development of workers.n will lead to degradation of work andng to negative career changes for lesslrganization, the debate was concentratedat job structures in firrns tend to to belraded.rrive at a general conclusion on (a) the:hanges after automatization in firms ands of these chanses.: defined as the result of changes in worknatization jobs can change in four ways:of skil ls diminish):skil ls improve);l-level in the organization Qob gain) and<ill-fevel in the organizatton Qob bss).

in the job structure of firms the expla

rel of firms, narnely downgrading and

levels in the firm namely polarisation.

can be divided into those supporting thessis, those supporting the upgrading orupporting the hypothesis that automatiza-n job structure. For the greater paÍ, this:rial. The constant contradiction betweenng and persistent. Case studies on thergic factories are well known because of'beek e.a. , 1985).hypothesis that automatization does nolions have influenced the central ouestion

of this study. Before focusing on the different changes in job structure, onecould ask why structures change at all. Therefore, the central researchquestion was refocused towards explaining the chonce of polarisation and thechance of regradation in automated organizations. As a result, the contradic-tion of empirical support for the different effects of automatization stillpersists. The absence of standardized interorganizational research and thelimited generality of case studies does not allow a preliminary answer to thcquestion why, in apparently similar organizations, the effects of automatiza-tion differ.

Therefore a theory must be elaborated which can generate hypotheseswhich predict in which firms which changes in job structr"rre are to be expected. The sociology of organisations is a diverse field of theoretical tradi-tions. Different approaches have developed simultaneously, although somehave been more popular in certain periods than others. Given our researchquestion, we selected three different approaches in organisational sociology.

The first is the technoLogical determinism approach. Based on the carlywork of Woodward and Touraine the idea behind this approach is thatorganizational structures vary with the level or state of technology. Studiesby memtrers of the Aston School empirically tested the relation between teoh-nology and type of organizational structure. This relation could in most casesnot be confirmed. The technological deterministic approach was also crit icized on theoretical grounds. One such contribution comes from Child. Heintroduced the idea of 'org,anizationaL choi<:e' which is used in theory andresearch today. In short, this concept states that in organizations, 'within

certain technological constraints, managers always have a choice of alterna-tives to (re)organize work and organizational structures. The rise of thisconcept caused an important decline in the significance of technologicaldeterminism in organizational studies.

The contingency approach was selected as a second tradition in organrza-tional studies. This approach has a strong tradition in the classic work ofBurns & Stalker and l-awrence & Lorsch.

'Ihe ernphasis in this approach is

placed on the influence of market conditions on organizational structures.The stability of the (market) environrnent makes some organizational stnrc-tures more successful than others. With regard to technological change, thisidea elaborated by Buitendam c.s.. They arrive at the hypothesis that thegreaíer the stability of the market, the greater the change in polarisation dueto automatization. Although attractive because of its clarity, a shortcoming ofthis hypothesis is the omission of organizations which do not operate on amarket, such as public firms. More generally, a major critisism of theapproach concems the deterministic character of the idea. One could arguethat the only difference between the contingency approach and technologicaldeterminism lies in the fact that on the one hand only technological condi-tions determine organizational structure, and on the other only features ofmarket conditions. Moreover, contingency theory has seldom been workedout into concrete hypothesis which can be empirically tested.

t -)J

As athird and recent tradition in the sociology of organisations, the laborprocess approach was discussed. This approach is based on the central ideasin Bravermans' writings on the control and division of labor. Marxists and'verstehende' theories of power relations are combined in the concept ofcontrol to illuminate organizational structures. In the Netherlands manystudies on the effects of automafization were done within this approach, forinstance Doorewaards' study of power relations in the context of automatiza-tion in insurance companies. Ho',,rever, hypotheses on how the decisionmaking process interferes with the relation between automatization and thechanges in job structures have not been worked out, so that the empiricalgenerality of the idea remains uncertain. The importance of this approach liesin the centralization of decision making as the 'interaction mechanism'through whích persons in organizations decide on certain job adaptions.

Not only as a critical reaction, but also as an extension of these three ap-proaches, a different theoretical approach was used to provide a moreinformative and complete answer to the central question of this study. Moreexplicitly, choices of managers in organizations are assumed to be the coreof the explanation of why jobs and, at aggregate level, job structures changeafter automatization. The conditions which influence the decisions of manag-ers to alter jobs in certain ways are dealt with stepwise modelling of manage-ment behaviour. In all four steps are taken leading to several hypotheseswhich explain the change in jobs and job structures in terms of up- anddowngrading and polarisation as described earlier.

In the first step the model starts with a situation of one manager in thefirm who has to decide upon the degradation, regradation, creation orremoval of jobs after automatization.

As a first modification, the manager is influenced by the technologicalfeatures of the automation equipment. This influence is apparent in the coststhat will arise when the skills of users of the automation equipment do notmatch the demand for skills. It is expected that managers will choose regra-dation or the creation of new jobs when autornation equipment is complex orapplied broadly in the firm. Under these conditions managers expect the costsof this 'skillgap' to be higher. On the other hand, it is to be expected thatsuchjob changes will be superfluous ifequipment has already been in use forsome time. In this situation, job changes such as downgrading and loss ofjobs become more attractive to the manager. This led to the hypothesis thatthe chances ofjob loss and job degradation are higher when the equipmenthas been in use longer or is introduced in a sinration where other similarequipment is already in use.

As a second. modffication to the first step of the model it is assumed that,besides intemal goals such as controlling skill gaps, managers also haveexternal goals such as the continuation of market production. In this way theidea is theoretically grounded that market situations are related to changes inorganizafions. The proposition is that stable market conditions stimulate thedivision of labor, since this is an efficient way of maximizing production.

t34

The result ocreation on tTherefore itchances ofjcIn addition, tthe quality renvironment.new jobs areity through a

To avoicapproach, thezations outsicfirms. The pctheir economabsence of cconly way maquality of thein combinaticlabour-intensipublic than in

In the seciour, the assuran interest in raltered. This iwhich the deli

First, empdecision makirsince this beaexpect that inchance ofjob

Second, c,tization in orgiadvisory worksystems. Withthe scope of tlthe involvemetthe chances ol

Membersdistinguished.are obliged toa right to inteÍstructure. sinrexpectation isof upgrading edowngrading i

rlogy of organisations , Íhe laborlch is based on the central ideasdivision of labor. Marxists andre combined in the concePt ofres. In the Netherlands manY: done within this aPProach, for

lns in the contexl o[ autornaLiza-

/potheses on how the decisionbetween automatization and the

rrked out, so that the emPiricalimportance of this approach lies

as the 'interaction mechanism'de on certain job adaPtions'.s an extension of these three aP-

was used to provide a more

tral question of this studY. Moreions are assumed to be the corergate level, job structures changenfluence the decisions of manag-

h stepwise modelling of manage-n leading to several hYPothesesstructures in terms of uP- and

earlier.siruation of one manager in the

lat ion, regradat ion. creat ion or

influenced by the technologicalinfluence is apparent in the costshe automation equiPrnent do not

that managers wil l choose regra-

lmation equiPment is comPlex or

ditions nlanagers expect the costs:r hand, it is to be exPected that

rrnent has alreadY been in use for

uch as downgrading and loss of

r. This led to the hYPothesis that

r are higher when the equiPmenta situation where other similar

p of the model it is assumed that,

skrll gaps, managers also have

arket production. In this waY the

tuations are related to changes in

: market conditions stimulate the

way of rnaximizing Production'

The result of the division of labor in order to increase productivity is jobcreation on the one hand and job loss and degradation of work on the other.Therefore it is to be expected that under stable market conditions, thechances ofjob loss and downgrading and simultaneous job gain are higher.ln addition, under znstable market conditions managers are eager to optimizethe quality of their product to cope with the uncertainfy of a changingenvironment. This leads to the hypothesis that regradation and creation ofnew jobs are to be expected because they are needed to support this flexibil-ity through automatization.

To avoid the previously mentioned shortcoming of the contingencyapproach, the influence of market conditions was also elaborated for organi-zations outside the commercial environnrent, such as non-profit and publicfirms. The position of managers in non-conrmercial firms is different becausetheir economic basis is dependent on budgets and government funds. In theabsence of competition there is a constant demand for their products, so theonly way managers can improve their position or status is to improve thequality of the services. Automatization to improve quality wil l only be usefulin combination with upgrading of, jobs, since the work remains relativelylabour-intensive. Therefore a higher chance ofjob upgrading is expected inpubl ic than in pr ivate organizat ions.

ln the second J/ep towards constructing a model of rnanagement behav-iour, the assumption that the manager is the onl1' person in the firm who hasan interest in and preferences concerning job changes after automatization isaltered. This alteration is realized by picturing a decision making process inwhich the deliberations of the manager are influenced by three other actors.

First, employees as usc^' of Íhe aLttomation equipmenÍ play a role in thedecision making process. 'fheir interest is especially in the upgrading ofjobs,since this bears upon the content of their own work. Therefore, one canexpect that in situations where users are involved in decision making thechance ofjob upgrading is larger.

Second, automotion experts are relevant actors in the context of automa-tization in organizations. Experts have an interest in expanding specialized andadvisory work in organizations, such as the implementation of new computersystems. With reference to job changes, the upgrading of jobs wil l benefit ofthe scope of their work in the organization.

'Iherefore the hypothesis is that

the involvement of automation experts in the decision rnaking process heightenthe chances of job upgrading and job creation accompaning automatization.

Members of the n-orks council form collectively the third actor to bcdistinguished. In the Netherlands, f irms with 35 fullt ime ernployees or moreare obliged to instinrte a works council. Members of this council legally havea right to interfere with important decisions such as changes of work and jobstructure. Since their interest is in all possible sorts of job changes, theexpectation is that when the work council members are involved the chanceof upgrading and job gain wil l be positively influenced, while the chance ofdowngrading and job loss wil l be negatively influenced.

t 3 5

In deriving these hypotheses, the underlying proposition is that the threeactors actually influence the managers' deliberations when they are involvedin the decision making. The third sÍep in the elaborating the model consistsof altering this assumption. The three extra hypotheses derived frorn thisstep are specifications of interaction effects. The influence of an actor'sinvolvement (step two) is expected to be reinforced by the technological andmarket conditions (added frorn step one). This theoretical step was based onthe idea of decision making as a process of exchange (Coleman, 1990,Stokman & van den Bos, 1989). Managers irnaginarily exchange theirpolicy position towards job changes for latent goods controlled by the threeother actors.

Pursuing this insight, managers are dependent on the user.; of automationequipment where these employees control an efficient usage of and gain frornthe ar-rtomatization. The first part of the adjusted proposition is the nrorecomplex, and the more recent the introcluction of automation equipment inthe firm, the more dependent managers wil l be on the cooperation and skil lsof the users and the more the decisions wil l be influenced by the interests ofthe users. The second part states that the influence of users 'will be reinfc.rrcedunder unstable markets and in non-conrmercial Íirrns.

Automation experts are engaged by managers on the basis of theirspecialized skil ls. The dependency increases with the cornplexity, novelty andscope of automation in the firni. The proposition is that thcse eqrripnrc'ntfeatures demand grealer knowledge and skil ls, especially in the lase ofimplementing the system.

' l 'he influence of experts cn managers' decisionmaking is therefore expecled to be stronger rvhen the ecluipment is complex,new and introduced broadly in the firm.

Mernbers of the works unmcil, the third actor, irnaginarily exchange co-operation for influence over the rnanagers' decisions concerning job adjust-ments. Managers have an interest in a cooperative works council. Theadjirsted proposition is: the broader the scope of automatization in the firrn,the greater the chance of discussion with the works council on this matterand the more managers are influenced by them in their decision making.

With this hypothesis the decision model is complete, as far as the chanceofjob changes such as upgrading, downgrading, and gain and loss ofjobs isc<rncerned. The transformation to changes ín job sÍructur€ is the fourth andlast step in the model. This transformation is a matter of applying rules ofaggregation and definit ion.

The first sort of job structure change is the increase or decrease in thernean job level in the organisation. This depends on whether job upgradingor downgrading is occuring and whether job gain and job loss take place onaverage at a high or a low job level in the firm. In addition, such job changescan occur simultaneously, which can result in a net absence of change in themean job level.

The second change in job structure is the division of job levels in thefrrm. We decided to concentrate on one sort of division: polarisation. It is amatter of definition when polarisation is said to be the result of job changes.

136

We have decidrjob gain at a haverage level wprobabilistic phresult of iob ch;

Finaily, tecsituations are recomplete, predi(downgraded or:

The complet.9. 1).

In the model sevelntegratlon was pcthe result of choicconditions. In thi

I T - ^ r ,

I I f.trf-1t ^luomp

F--l Scopr| | NovelI lDura t i

t=I IN /ARKI lS tab i l

l*II ''"

Figure S.l Schet

ing proposition is that the threerations when they are involvedelaborating the model consistshypotheses derived fron'l this

. The influence of an actor'sforced by the technological andis theoretical step was based onof exchange (Coleman, 1990,rs imaginarily exchange theirt goods controlled by the three

dent on the u.lers of automationlff,rcient usage of and gain fromjusted proposition is the moreon of autornat ion equipment inDe on the cooperation and skillsre influenced by the interests ofence of users 'will be reinforcedal firms.anagers on the basis of theirvith the complexity, novelty andosi t ion is that these eqrr ipmentiills, especially in the fase ofexperts cn matragers' decision-vhen the equipment is cornPlex,

actor, imaginarily exchange co-lecisions concerning job adjust-roperative works council. Thee of automatization in the firm,re works council on this matter:m in their decision makrng.is complete, as far as the chanceing, and gain and loss ofjobs ist job structure is the fourth andis a matter of applying rules of

the increase or decrease in therends on whether job upgradingI gain and job loss take Place onm. In addition, such job changesn a net aLrsence of change in the

the division of job levels in the: of division: polarisation. It is aI to be the result ofjob changes.

Figure 5.1 Scheme of the explanotory *"a"tïorait,,r"p,

We have decided to use a broad definition, meaning that job upgradrng orjob gain at a high average level orjob downgrading orjob ross at a lowaverage level will cause polarisation. In this way polirisatitn is defined as aprobabilistic phenomenon than can occur to a greater or lesser extent as aresult. of_job changes foliowing automarizar.ion..

l-rnally, technology factols. marker condirions and decision makingsituations are related to these three job structure changes. The model is nowcomplete, predicting the conditions under which job siuct.rres are upgraded,downgraded or split up (polarisarion) following automarization.

I ne comptete model can be visualized in the following schema (figures.1).

ln the model several insights concerning organisations are integrated. Thisrntegration was possible because of the starting assumptions of jo'b change asthe result of choice and choice as a deriberation by n,inug"., under differentconditions. In this way hypotheses are founded by one consistent set of

TECHNOLOGYComp lox i t yScope o Í au tomat ioNove l t y DEOIS ION MAKIN

U S E T SEx per tsWorks counc i l

MARKETStab i l i t y

/ Q + ^ ^ í l

JCBCHANGESupgrad ingDowngrad ingJob ga inJob l oss

JCB STRUCTURE CHANGESUpgrad ingDegrac la t i onPo la r i sa t i on

( S t e p

( S t e p

theoretical premises and logically derived as ernpirically testable statements.Since the model is developed according to the nrethod of decreasing abstrac-tion (compare Lindenberg, 1983), it can be tested rvhether the theoreticalelaboration of the rnodel (rn three steps) coincides with an improvem.'nt inthe statistical f it of the rnodel. This allows an extra evaluation and empiricaltest of the complete model.

The hypotheses and the model are empirically tested using two data sets.The first data set consists of the results of a mailed questionnaire arnong

491 autornated organiz-ations in Lirnburg (a province in the south of theNetherlands) in 1986. lJsing these data it was possible to operationalize andmeasure the theoretical concepts, but at the sarne l ime several objectionsbecame clear. First, the data on the state of tcchnology, the market situation, the decision rnaking and the changes in jobs are provided by only oncperson in the firnr. Secor.rd, questions on firrn profits, involvernent of workscounci ls in decis ion nraking and. lob loss are scnsi l ive to socia l ly desi rablcanswers, especially when using rnailecl questionnaires. And third, wlr.iài thehrnrs it is not possible to drstingLrish betu,ccn differcnt applications ofautomation equiplnent. Such diÍfercntiation is dcsirable because withinorganizations equipment can difÍèr significantly in conrplexity and noveltyand in the decision making and job changes which are associated with theequipment.

For these reasons, extra data was collecteci in a subs:rmple of 2l organi-sations.

'fhese data arc r.r.ruch more in Iine rvith the tlreoretical concepts clí the

model arrd accomodate the indicated objcctions. First, data was gatlreredfrorn severaL persons within the organization. Job changcs are judged by theemployees themselves, decision making by the participants themselves,market conditions by the financial n'ranager, the statc of autonratzation by theautomation personnel, etcetera. Second, the sarne questions were answereclby a nr.rmber of persons in the organisation (several users of the sanreequiprnent, several decision makers for the sanre decision). And third, withinthe 2l f irms, 62 different cases of ar.rtomatizatiorl were exanrined. Thcsecond data set consists of 62 observations (cases of equipment or decisions)with the sarne variables as measurecl in the l lrst data set of 491 organizat ions.

The resr-rlts of the analysis of both data sets can be presented sirnul[a-neously. There is a quite a rernarkable sinri larity between the results. In bothdata sets, job upgrading is the most frequently occuring job change (first datase t 4 l%; second 61%) . I ob ga in (16%; l 0%) and j ob l oss (5%;3%) <>ccu rmuch less frequently. Job loss wil l not be involved in further analysisbecause of these low frequencies. Job downgrading cannot be distinguishedfrom job upgrading in the first data set, but the more accurate data (N:62cases) show that such change was hardly present. Regradation of the jobstructure (a net increase in average job level) occurred most frequently inboth data sets (56%; 4I%). Degradation of the job stnrcture can be left outof further analysis (2%;5%). Polarisation, the rnain subject of the polit ical

1 3 8

discussion, can be ob:automation cases. Thisslnce the definit ion of I

In testing the hypordata set N:491 and into test for significancestrength of the expectecto the changes in jobbivariate analysis.

Some of rhe hypoth,chance of upgrading arelated to complexityorganizations. Novelty irelation to changes in jrgreatest chances of pol;relatively sÍ.able and relThere was no differenct

The hypotheses frorcomes of the analysis. 1data set (N:a9t); 29%positively related to th,structure. The relation t,be analysed because of

lnteraction effects, ,ment of the model, coul.data. Only the main eÍwere reliable.

These results are ,analysis using only the Ifirst step - as in the(complexity, scope ofincluded in the regressirtion of the data varianAdding the market varinot funher improve tht- as in the modificatioproves the fit of the mocbetween the technologicregression equation. Thr

Evaluation of all theconsequences of automadation ofjobs and job sfrequencies that it maketest the theoretical modetwo data sets collected i

as ernpiÍically testable statentents.the method of decreasrng abstrac-be tested whether the theoretical

coinc ides wi th an improverncnt in; an extra evaluation and empirical

rir ically tested using two data sets.s of a r r ra i led quest ionnaire arnong1 (a province in the south oÍ' thewas possible to operationalize andthe sanre time several objcctionsr of technology, the market situa-s in jobs are provided by only onefirm profits, involvenrcnt of worksare sensitive to socially desirable

restionnaires. And third, within rhebetwecn different applications oftion is desirable because withinicantly in conrplexity and noveltyrges which are associated with the

ected in a subsample of 27 organi-rvith the theorctical concepts of the

lections. Flrsl, data was gatheredion. Job changes are judged by the1by the par t ic ipants themselves,)r, the state of automatT-ation by thethe same questions were answeredjation (several users oÍ' the sarnee same decision). And third, wíthintomatization were examined. Thes (cases of equiprnent or decisions)the f i rs t data sel of - l9 l organiza-

ata sets can be presented sirnulta-ri larity between the results. In both:ntly occuring.lob change (first dataL0%) and j ob l oss (5%, 3%) occu r't be involved in further analysiswngrading cannot be distinguishedbut the more accurate data (N:62y present. Regradation of the job

level) occurred rnost frequently inof the job stnlcture can be left outn, the main subject of the polit ical

discussion, can be observed in 3g% of the 491 firrns andautomation cases. This can be considered a relatively low ratesrnce the definit ion of polarisation is broacl.

in 20% of theof occurrence,

. In testing the hypotheses ancr the nrodel it nrust be noted that in the first

data set N:491 and in the second N:62. Therefore we use the first data serto test for significance ancr the second to anaryse the direction, linearrty andstrength of the expected rerations. we will sulnnrarize the resurts with regardto the changes in job structure. The hypotheses are first analysed usingbivariate analysis.

. Some of the hypotheses frorn the first srep are supported by the data. The

chance of upgrading and polarisation of tire job si^ctur" are posrtivelyrelated to comprexity of equipment and the scope of automatizatron inorganizations. Novelty ancl duration of use of equipment show a weak, or no,relation to changes in job struct're. on theoreticar grounds *. e*p"ctea tnegreatest chances of polarisation and regradation when rnarket .onàitiun, ur"relatively stable and reratively unstabre. only the ratter could be supported.There was no difference between private anj public organisations.

The hypotheses from the second stcp were partry zupported by the out-comes of the analysis. The invorvement of users (frequency 49% ;; the firstdata set (N=491); 29% in the second (N=62)) and experts (2g%;4g%) ispositively related to the chance of upgrading and polarisation of the .,obstructure. The relation to the invorvernent of works council mernbers wrlr notbe analysed because of Iow f requencies ( Ig%; 6%).

Interaction effects, expected according to the thircr step in the deverop-ment of the moder, courd not be demonstráted by multivariáte analysis of thedata. only the moin effects of technorogicar and decision rnaking factorswere reliable.

These results are confirmect by a final stepwise rogistic regressionanalysis using only the large data set of 49r automated orgánisationJ. In thefirst step - as in the theoretical moder - only technálogicar variabres(complexity, scope of automatization, duration of use anJ novelty; areincluded in the regression equation. using this model, a significant propor-tion of the data variance compared ro the base model càn be exilained.Adding the market variables (stability, public/private) to the equatiàn doesnot further improve the moder fit. Adding the decision making variabres- as in the modification of step two in the theoretical model r do", ,--proves the fit of the moder. Finally, suggested in step 3, the interaction termsbetween the technological and decision rnaking va.abres were added to theregresston equation. This did not further improve the model fit.

Evaluation of all the results reads to the first main conclusion that sameconsequences of automatization cannot be demonstrated empirically. Degra_dation of jobs and job structures such as with job loss occur with such lowfrequencies that it makes no sense to examine these phenomena in order totest the theoretical model and hypotheses. This conclusion is by supported bytwo data sets collected in two different wavs.

t39

The second main conclusion is that in general there is suppoÍ for themodel we developed to explain variations between organizations in thechances of job structure changes. The support for the model is in favour ofthe effect of technological constraints and the influence of decision makingpartners as conditions under which managers imaginanly compare the prosand cons of certain job changes. Since not all technological features anddecision making factors are significantly related to job structure changes, itis worth reconsidering the theoretical construction of these hypotheses. Asuggestion with regard to the absence of an effect of market conditions mightbe a more accurate specification of the environmental features (for instancethe temporal financial situation of the firm instead of stability of the marketlevel).

Other possible modifications lie in the assumptions of the rnodel. Onecould consider for instance distinguishing different kinds of managers, since'the manager' is treated here as one actor in the model. With regard to thedecision making process one could also consider introducing interactionsbefween the three different actors.

Geraadpleel

Altmann, N., en G. BecBetriebliche Henscl

Asch, A. van en T. JansNaar een verbeter,autr.tmatisering. Eenvernieuwingen. Den

Batenburg, Ronaltl S. l9Automatisering enLimburgse bedriiven

Batenburg, Ronald S. 19,Ftrnctiestructuurverar2, pp. 126-144

Berentsen, Boudewijn l9lEn de invloed houde,Amsterdam: FNV-Ct

Bethlehem, Jelke G. en VA general method fol

Bilderbeek, R.H. en p.J.Automatisering van d,

Birchal l , D.W.19i7Technological changeThames: Administrati

Bjórn-Andersen N., en anrComputersystems andbanking. Wenen: CenSociales.

Blauwhoff, Gertrud en LoeTechnologie en ArbeitHaag : Staatsuitgeverii.

Boer, Hans de l9g9Automatisering en pervan organisatiekenmerwerving en scholing. C

Brandt G. en anderen ló7gComputer und Arbeítspgen des Computereinsatdes Bankgewerbes. Kra

Braverman, H. Ig74Ittbor and monopoly ccYork: Monthly Review

140