REPRESENTATION, TESTING AND POLICYIMPLICATIONS OF PLANNING PROCESSESKIMBERLY B, BOALDepatiment of Managerial Sciences, University of Nevada-Reno, Reno, Nevada,U.S. A.
JOHN M, BRYSONHubert H. Humphrey Institute of Public Affairs. University of Minnesota, Minneapolis,Minnesota, U.S.A.
Various way.s to represent planning processes lo test and draw policy implications fromthem are presented. We argue that three kinds of variables should he included: context,process and outcome. Four different models which incorporate these three variables areproposed, and data from a large public sector planning .study are fitted to these four models.The findings suggesi that the process-outcome link is the key to understanding theeffectiveness of planning processes.
INTRODUCTION
Managers frequently are exhorted to includevarious plannltig processes in their tool kit.Rarely, however, are matiagers given any evi-dence about the comparative effectiveness ofthese processes (Bryson, 1983), The most fre-quently cited tests sitnply measure the effect ofthe presence or absence of formal planningsystems on profitability, rate of return, marketshare, or some other desired outcome {e.g,Harold, 1972; Malik and Karger, 1975; Thuneand House, 1970). These tests usually indicatethat the use of a formal planning system doeshelp, but exactly how is unclear (Hofer andSchendel, 1978).
One difficulty in evaluating planning processesis knowing how to represent them for testing.Previous research on the nature of the planningprocess has followed one or more of fourbasic approaches, The first approach emphasizescontext and outcomes, while ignoring processaltogether (e,g. Porter, 1980), The implication isthat one has only to match desired outcomesto context and the appropriate process willautomatically follow. In fact, some research on
goal setting does suggest that individuals doadjust different strategies if goals are altered(Locke et al., 1981),
The second and more traditional approachfocuses on normative or descriptive sequences ofphases or steps (e.g, Delbecq and Van de Ven,1971; Etzioni. 1967; Undblom. 1959; Mintzberg,Raisinghani and Theoret, 1976; Simon, 1947;Van de Ven, 1980b), The assumption is thatfollowing a certain problem-solving sequence willlead to desired outcomes,
A third approach focuses on specific tactics ortasks to be completed within one or more stepsof a general problem-solving model. Exampleswould include such tactics or tasks as establishinga program coordinating committee, doing acost-benefit analysis, or conducting a pilot test(e,g, Bryson el al., 1979; Bryson and Delbecq,1979; and Nutt, 1982), The assumption is thatuse of specific tactics or the performance ofspecific tasks will produce desired outcomes.
The final approach focuses on generic activities(e.g. communication, conflict resolution, or goal-setting activities) either within a sequence ofphases (e,g. Van de Ven. 1980a) or across anentire problem-solving sequence (e,g, Bryson,
1979). This paper will utilize this last approachto the representation of planning processes.
Several theorists have suggested recently thatthe appropriate choice of a planning process iscontingent upon any number of factors, such asthe structure and complexity of the problem, theenvironmental context, and the desired outcomes(Beach and Mitchell, 1978; Bryson et al., 1979;Bryson and Boal, 1983; Bryson and Delbecq,1979; Nutt, 1976a; Payne, 1982; Rainey, Backoffand Levine, 1976; Kartez, 1984; Alexander,1984), More research focused directly on planningprocesses is necessary, however, before it ispossible to know which is the most effectiveprocess in which situations and why.
Evidence that good planners do behave contin-gently if given the chance was first obtained byBryson and Delbecq (1979) through the use ofa project planning simulation in which plannerswere asked to design a planning process inresponse to different situations, or contexts,Bryson and Delbecq were able to investigate theinfluence of the following contingencies: whetherthe project goal was easy or difficult, and whetherthe situation was easy or difficult politically andtechnically. Each contingency was found tohave some infiuence on the design of planningprocesses. Political difficulties prompted themost dramatic contingent variations in planningprocesses, followed by goal and technical con-cerns. As an example, the planners respondedto politically difficult circumstances by payingmore attention to coalition-building, shared dis-cussion and problem-solving, bargaining andnegotiation, and so forth, than they did in theabsence of political difficulties. In summarizingtheir research results, the authors noted:
The prineipal eonclusion is that project plannersdo appear to behave contingently if given theehoiee. They do appear to change their strategiesand tactics as the situation changes in the hopeof increasing the likelihood of goal achievement.Specifically, there appear to be some thingsplanners always do. never do, and do contingentupon the situation, given constraints on budgetand staff time (Bryson and Delbeeq, 1979: 177),
We believe that even the simplest contingentmodel of the planning process should includefour basic elements: the context within which theplanning occurs, the planning process itself, theresult or outcome of the planning, and the
interconnections among these three elements.Among the very few studies that have eonsideredall four constructs are McCaskey (1974). Nutt(1976b), Bryson and Delbecq (1979), Bryson andBoal (1983), and Van de Ven (1980a-c),
The remainder of this paper is divided intofour parts. The first section will discuss fourplanning models that link context, process, andoutcome variables, A brief example of eachmodel will be presented, along with a briefoutline of the policy implications of each model.The second section will discuss how the fourmodels can be tested using cross-sectional data.Alternative research approaches will also bementioned. The third section will use data froma study of Minnesota's Land Planning Act of1976 to demonstrate the utility of the four models.Finally, we will outline some implications of thefour models for practice and research.
CONTEXT, PROCESS, OUTCOMES,AND THEIR INTERRELATIONSHIPS
Four simple theoretical models of the inter-relationships of context, process, and outcomewill be discussed: the Intervening. Independent.Moderating, and Interaction Effects models (seeFigure 1), More complex models involvingindirect paths and feedback loops will not bediscussed due to space limitations.
The Intervening Effects model has two ver-sions. In one, context has no independent effecton outcomes. Instead, context affects process,and process in turn affects outcomes. For example(following Delbecq and Van de Ven, 1971; Nutt,1976a; and Thompson, 1967) planning contextsthat differ in the analyzability, variability, andinterdependence of decision tasks or technologywould cause planners to choose different groupprocesses or patterns of communication as ameans of producing desired outcomes.
In the other version of this model, process hasno independent effect on outcomes. Instead,process affects context, and context in turn affectsoutcomes. An example would be the process ofconvening a large group of very different expertsto create an innovative solution to a problem. Itis then the context of an interdisciplinary teamthat results in a far different solution than mighthave occurred otherwise (Delbecq, Van de Venand Gustafson, 1975).
Representation, Testing and Policy Implications of Planning Processes 213
I
to
o5
>Oo
CO
o
o '5(j_ra
3 oa"c O
3
a'Era
E >
S8C Ora
» c
is0) 0)
c £O cU ro
" •S•o 2,E O
ra o _2 '£ (D ^ O
XI V) o Pra 0) (- t̂
0 E « •2 o d) o'
«f o|.J- O _ . o •
1 ^E °ra ^ Qj
— ra I
o w E
lC 3O O
o ,.c
B o
•fc (fl tJ) _o ^.E -
c L:O 03*; ,c_tO " ^ .
>— ora (D „,
r- O *D ^
o g ^ .E i2
O
AX
co roc
a.
QO
o,£ a
LUoo
3 Q.
c ora uU (1)
UQ, O)X C
Xa>*-•o
CJ
Hi
Eo(J
O<
U)tn0)
O
CL
T3C
O
B
u•aoB
214 K. B. Boal and J. M. Bryson
In the Independent Effects model, context andprocess have statistically independent effects onoutcomes. For example, consider a contextvariable such as group composition. No matterwhat process is used, planning projects thatrequire the participation of heterogeneous groupswith different norms and perhaps incompatiblegoals are likely to result in different outcomescompared to similar projects with homogeneousand cohesive participants (Huber, 1980; Janisand Mann, 1977; Seashore, 1954). Conversely,the use of certain group decision-making processes{e.g. the Nominal Group technique versus atraditional interacting group) will have an effectindependent of group composition (Delbecq, Vande Ven and Gustafson, 1975).
The Moderating Effects model also has twoversions. In the first the direct effects of contexton outcomes are moderated by process. Forexample, scarce resources (context) directly affectthe possible outcomes for groups competing forthose resoruces (Pfeffer and Salancik. 1978). Onthe other hand, conflict resolution processes thatmoderate the competition for resources caninfluence the connection between scarce resourcesand outcomes (Filley, 1975).
In the second version, context moderates thedirect effects of process on outcomes. Forexample, task complexity (context) has beenshown to moderate the effects of differentcommunication processes on task performance(Shaw, 1964).
In the pure Interaction Effects model thereare no simple main effects. The only effect resultsfrom the interaction of process and context. Forexample, Vroom and Yetton (1973) suggest thatproblem attributes (e.g. problem structure) anddecision-making style (e.g. autocratic, consulta-tive, and group) interact to effect decision qualityand acceptance.
Knowing which theoretical model best describesthe relationships among context, process, andoutcome variables should help answer severalplanning-related questions. First, can an inter-vention affect outcomes in a desired way? Second,can planners focus on either context or processvariables alone, or must they worry about both?And third, where in a causal sequence shouldplanners intervene? These questions cannot beanswered in detail without reference to specificvariables and settings. There are, however,general implieations of the four models.
The basic implication of the Intervening Effects
model is that process (in version 1) or context(in version 2) must be manipulated by manipulat-ing its antecedents. That is, in version 1 contextdirectly determines the appropriate process whichin turn determines outcomes. The implication forpolicy-makers is that they should manipulatecontext—and not worry so much aboutprocess—as they seek to affect outcomes.
This model is implicit in Porter's (1980)competitive strategy framework, in which marketconditions (context) basically determine strategicresponse (process), which in turn should stronglyinfluence outcomes. If the model is accurate,government policy-makers should try to manipu-late market conditions in light of desiredoutcomes—for example, through anti-trust poli-cies, environmental regulations, and taxschedules^—and not worry much about corporatestrategic responses. Following this model, corpor-ate planners should simply make an accurateassessment of market conditions and the appropri-ate competitive response should become apparent.
The Independent Effects model, in contrast tothe Intervening Effects model, implies thatplanning can affect outcomes by manipulatingboth context and process. As a result, policy-makers have more strategies at their disposal,but they probably do not have the option ofignoring either context or process. Moreover, itis unlikely that large changes in outcomes willresult from small changes in either context orprocess variables unless the relevant causal linksare very strong.
The policy implication of version 1 of theModerating Effects model is that one must knowsomething about process in order to predict theeffects of context on outcomes. For version 2one must know something about context in orderto predict the effects of process on outcomes. Inthe Interaction Effects model, manipulations ofcontext are dependent on process and vice-versa,since there are no simple main effects. Moreover,interaction effects may make it possible toproduce large changes in outcome variablesthrough small changes in either context orprocess.
TESTING THE DIFFERENTTHEORETICAL MODELS
Tests of the first three theoretical models arestraightforward. A test of the fourth model is
Representation, Testing and Policy Implications of Planning Processes 215
more difficult. In the research discussed below,we first were guided by the first version of theIntervening Effects model and by part of theIndependent Effects model (see Figure 2). Weassumed, in other words, that the genericprocess variables intervened between context andoutcome variables in some cases, and that inother cases context variables had an independenteffect on outcomes. We first tested our databased on these assumptions, and then pursuedthe exploratory task of fitting our data to othertheoretical models.
In order to test the first version of theIntervening Effects model we used the followingprotocols; the first order partial correlations ofcontext and outcome variables were comparedto the bivariate correlations. If the zero-ordercorrelations were significant, but the partialswere non-significant, we then concluded that theprocess variables do intervene between thecontextual and outcome variables. If the zero-order correlations were non-significant, but thepartials were significant, then the process vari-ables were acting as 'suppressor" or 'distorter'variables. (Suppressor variables, also known as'distorter' variables, tend to mask the actualrelationship between an independent and depen-dent variable. This usually occurs because thesuppressor variable is positively correlated withone of the variables but negatively correlatedwith the other.) If the zero-order correlationsand the partials were both significant or non-significant, we concluded that contextual andprocess variables had independent or no effects,respectively.
To test the second version of the InterveningEffects model, we would follow the sameprocedure, except that here we would partial outthe effects of the context variables. In theexample to be discussed below we have onlyanalyzed the data in terms of the first version ofthe Intervening Effects model, because of our apriori theoretical assumptions about how thevariables in our study might be related.
The above procedure suggests whether theIntervening or Independent Effects models ismore descriptive of the data. However, theprocedure is insufficient because the bivariatecorrelations do not reveal if both process andoutcome variables explain variance independentof each other. To test for their joint effects it isnecessary to use standard multiple regressionanalysis.
To test for the Moderating Effects model weused hierarchical moderated regression analysis(Saunders. 1956; Arnold, 1982). Moderatedregression is similar to multiple regression exceptthat cross-product terms for the independentvariables thought to interact are entered into theequation after the main (independent variables')effects are determined. The moderating effect isexamined by comparing the multiple correlationcoefficient (/?-) of the equation without the cross-product term(s) to the equation with the cross-product term(s). The increment in R~ is thentested for statistical significance. Since the testfor both versions of the Moderating Effects modelrelies on the same statistical model, differentiationbetween the two models must be on theoreticalgrounds.
Unfortunately, there is no accepted standardprocedure for differentiating the ModeratingEffects model from the Interaction Effects modelusing data such as ours. The mononacity,reliability, and non-ratio nature of the data,among other things, contributed to the difficulty(e.g. Arnold, 1982; Busemeyer and Jones, 1983).In the results reported below we used thefollowing procedure and reasoning. We comparedthe results of the hierarchical moderatedregression analysis with a free entry stepwiseregression analysis. We reasoned that if theinteraction model best described the data thenthe following should oecur. First, in the freeentry stepwise analysis only the interaction termshould enter in the equation. Second, in thehierarchical analysis the interaction effect termshould always enter into the equation. Third,with all three terms in the equation (i.e. the twomain effects and the interaction), only the betaweight for the interaction term shouldbe statistically significant. In other words, onlyif all three occurred would we conclude thatthe Interaction Effects model best described thedata.
The statistical tests discussed above are appro-priate for cross-sectional data. Other researchers(e.g. Bryson and Delbeeq, 1979; and Nutt,1976b, 1977) have utilized experimental orquasiexperimental designs to examine the interac-tion between process and context variables inaffecting outcomes. Nutt (1983) discusses thepros and cons of various research strategies (e.g.longitudinal, case, process reconstruction, andexperimental strategies) for studying the planningprocess (cf. Bryson, 1983).
216 K. B. Boal and J. M. Bryson
An Example: the Implementation of Minnesota'^Metropolitan Land Planning Act of 1976
Our example is based upon data drawn from astudy of the implementation of Minnesota'sMetropolitan Land Planning Act (MLPA) of1976, which was a pioneering attempt to manageurban growth and decline in a major metropolitanarea,' Our study is one of the few empiricalplanning studies of its kind (Bolan and Nuttall,1975; Gilbert and Specht, 1977; Van de Ven,1980a, b; Van de Ven and Koenig. 1976),
Like policy-makers elsewhere in the country.Minnesota legislators recognized the urgent needto develop effective methods for expanding,maintaining or shrinking public services inresponse to urban population shifts and fiscalcircumstances, Minnesota legislators, however,also realized that solving problems of growthmanagement and control requires the appropriatedesign and management of publie organizationalstructures and processes.
The legislature decided to adopt a design thatworked well in the corporate world. The designpermits centralized control over critical region-shaping systems, such as transportation networksand water and sewer systems; while allowing farmore decentralized control over less criticalregion-shaping decision such as sites for specificindustries or types of housing. The design is thepublic-sector, inter-organizational equivalent ofmany corporate planning systems, in whichstrategic decisions are made at the institutionallevel and less critical decisions are reserved forthe managerial and technical core levels (e.g,Hofer and Schendel, 1978; Mintzberg, 1979;Steiner, Miner and Gray, 1982).
The generalizability of our example dependsin part on how important the distinctions arebetween the public and private sectors, A varietyof authors have argued that there are importantdifferences between the public and private sectors(e,g, Rainey, Backoff and Levine, 1976; Lind-blom, 1977; Ring and Perry, 1985). Unfortunatelythese assertions rarely have been tested empir-ically. When they have been, the distinctions
' Cotnplete descriptions of the Metropolilan Council, itsfunctions, responsibilities, and limitations, and of theMetropolitan Development Guide may be obtained from theauthors, or from the Metropolitan Council, Public InformationOffice, 300 Metro Square Building, St Paul, Minnesota 55101.
between public and private sectors often appearto be minor (e.g. Rainey. 1983). Furthermore,a variety of authors from both the public (e.g,Cleveland, 1973, 1985) and private sectors (e.g.Freeman, 1984) argue that the publie and privatesectors are becoming increasingly indistinguish-able. We therefore conclude that while thequestion is still open, it is nonetheless useful tothink about the applicability of our four modelsto both the public and private sectors.
The Metropolitan Council, the MetropolitanDevelopment Guide, and the Metropolitan LandPlanning Act
The Metropolitan Land Planning Act of 1976was to be implemented by the MetropolitanCouncil (MC), a nationally unique, limited-purpose, government agency designed to guideand control growth in the Twin Cities metropoli-tan area. Established by the Minnesota Legis-lature in 1967, the council was patterned on theearly General Motors model of organizationaldesign. In that model, corporate policy decisionswere reserved for the board of directors andcorporate headquarters, while most operatingdecisions were left to the separate divisions(Sloan, 1963), In the case of the MetropolitanCouncil, regional decision-making was reservedfor the council, while most operating decisionswere left to the regional operating agencies (e,g,the Metropolitan Transit Commission, WasteControl Commission, or Airports Commission),or to units of local government (Harrigan andJohnson, 1978; Bryson, 1981). As a frameworkfor both regional and local decision-making, thecouncil produced a Metropolitan DevelopmentGuide,
The Metropolitan Land Planning Act of 1976required the council to prepare a metropolilansystems statement—based on the developmentguide—for each of the 195 local units ofgovernment and 49 school districts in the metro-politan area. The statements were to be completedby 1 July 1977, The act further required thateach local government use its metropolitansystems statement to develop a comprehensiveplan by the end of July 1980 (a deadline thatmany units did not meet).
Each metropolitan systems statement was tobe a compilation of information that the councildeemed necessary for the preparation of a local
Representation, Testing and Policy Implications of Planning Processes 217
comprehensive plan. Both the council and thelocal units had to agree to the statement.
Each unit's comprehensive plan had to conformto its systems statement and all applicableMetropolitan Council review criteria, almost allof which were drawn from the MetropolitanDevelopment Guide. The local unit's comprehen-sive plan had to contain: a land-use plan, publicfacilities plan, housing implementation program,capital improvements program, and an integratedset of official controls (e.g. zoning ordinances,sewer regulations, and subdivision ordinances).The council could force local units to revise plansthat did not meet review criteria.
Study design and methodology
The basic conceptualization that provided initialguidance and data for this study is presentedin Figure 2. The lines in Figure 2 representhypothesized bivariatc relationships. (These willnot be discussed in detail, but are presented onlyto aid the reader in the discussion to follow.)^In this partly Intervening and partly IndependentHtlccts model, context variables affect processand outcome variables, and process variablesaffect outcome variables.
Procedures and Subjects
Data were collected for this study from twosources—the U.S. census and a lengthy question-naire filled out by representatives of localgovernment units in the Twin Cities region,(Each representative was the key MetropolitanCouncil liaison for his or her government unit.)We invited a representative from each of the 195local units to complete the questionnaire. Of the69 (.̂ 5.2 percent) who agreed, 55 were employedby municipalities, six by counties, and eight bytownships. They had 15 different job titles; thefour most common being Planning Director (12),City Administrator (10), Clerk/Treasurer (8). andCity Planner (8). They had worked for theirorganizations an average of 6.6 years (S.D. =
' A full discussion of the logic behind, and test results of.ihc hypotheses underlying Figure 2 can be found in: Bryson.John M. and Kimberly B. Boal 'Strategic manugement ina niL'Iropolitan area: the implementation of Minnesota'sMetropolitan Land Planning Act of 1976". Hubert H.Humphrey Institute of Public Affairs. University of Minne-sota, Working Paper. \9H2.
5.50), and 54 of them had at least bachelor'sdegrees, 38 of which were in a planning-relatedfield. The larger, more professional units ofgovernment in the metropolitan area were some-what overrepresented in our sample.
MEASURES
The questionnaire collected data on the majorcontextual process, and outcome variables relatedto local governments' preparation of comprehen-sive plans in response to the MLPA. Mostvariables were measured on a five-point Likertscale, with differing anchors. Whenever possiblewe tried to develop multiple indicators ofa phenomenon. Factor analysis (not shown)sometimes indicated that these multiple indicatorsshould be collapsed into scales, sometimes not.(The Appendix presents a more complete descrip-tion, including operationalization. means, stan-dard deviations, and where applicable, coefficientalphas. The questionnaire itself may be obtainedfrom the authors.)
Contextual variables
Resources
The following resources were measured: (1) timeallotted for completing plan (Ample Time);(2) level of financial resources for completingplan (Financial Resources); (3) adequacy of dollarresources for decision-making (Decision-MakingResources); (4) dependency on grant money forpreparing plan (Planning Grant Dependency):(5) dependency on grant money for implementingplan (Implementation Grant Dependency); and(6) adequacy of full-time equivalent professionalsto prepare plan (Adequate Personnel).
Demography
Two demographic characteristics were measured:population size (1980 Population), and populationgrowth between 1970 and 1980 (PopulationGrowth).
Number of groups
Respondents indicated whether or not each of12 different groups had a formal or active rote
SUPPORTOFMCFORLOCAL UNITS ASPERCEIVED BYLOCAL UNITS(Goal Support)
COMMUNICATIONFREOUENCV OF
COMMUNICATION(Frequency ol
Communication)
CHANGES INCOMMUNICATION
PATTERNS(Communication
Changes)
IMPACTON UNITS' PLANSHOW DIFFERENTIS PLAN(How Plan DiflerentI
WHAT ACTIONWITHOUT MLPA(Alternative Action)
IMPROVeMENTS INUNITS- CAPABILITIESEFFECT ON DECISION-MAKING CAPABILITY(Decision Makirvg)
USEFULNESSOF LEARNING(Lessons Learned)
DEMOGRAPHYPOPULATION 1980
(1980 Population)
POPULATION CHANGE1970-19S0
(Population Growtti)
ROLE CHANGESROLE CHANGES
OF ACTORS(Role Changes)
NUMBER OFGROUPS INVOLVEDNUMBER OF GROUPS
INVOLVED OVERALL(Numbof ol Groups)
CONFLICT ANDCONFLICT RESOLUTION
AVERAGE CONFLICTISSUES # t - » 3
(Average Conflict)
AVERAGE NUMBEROF CONFLICT
RESOLUTION METHODSISSUES #1-#3
(Conllict ResolutionMettiods)
CONSULTATION ANDTECHNICAL
ASSISTANCE
CONSULTANT USED INPLAN PREPARATION
(consultant)
HELPFULNESS OF TAFROM COUNTY
(TechnicalAssistance)
SATISFACTIONTOTAL SATISFACTION(Sslislaclion)
PERCEIVEDEFFECTIVENESSOF MLPA
PERCEPTION OFEFFECTIVENESS OF MLPA(Effectiveness)
Figure 2. Hypothesized relationships among context, process, and outcome variables
in either plan preparation or implementation(Number of Groups).
Role changes
Also measured was the extent to which the rolesplayed by the 12 groups represented a changefrom earlier relationships or merely a continuationof established practices (Role Changes).
Process Variables
For our study, process was conceptualized asgeneric activities across an entire problem-solvingsequence. As noted in the introduction, thisapproach is only one of four that are typical ofplanning process studies. Four generic processactivities were investigated: goal-setting, com-munications, confiict, and consultation and tech-nical assistance.
Representation, Testing and Policy Implications of Planning Processes 219
Goal-setting
Three task-goal attributes were measured usingmulti-item scales. These attributes were GoalAcceptability, Goal Specificity, and Goal Sup-port.
Communications
Two communication attributes were measured:frequency of communication between organi-zations (Frequency of Communications), and theextent to which communication patterns betweenthese organizations had changed (CommunicationChanges).
learned from tbe comprehensive planning andimplementation process be useful for understand-ing future comprehensive planning and implemen-tation endeavors?'
Overall satisfaction ^
To assess overall satisfaction (Satisfaction),respondents were asked to what extent theyfelt their comprehensive plans would improvedecision-making in general operations, land-use,public facilities, housing and capital improve-ments. They also were asked whether the officialcontrols identified in the plan would actuallyachieve plan objectives.
Conflict
Two aspects of conflict were measured: averageamount of conflict (Average Conflict) and thenumber and type of conflict resolution techniquesused (Conflict Resolution Methods).
Consultation and technical assistance
Respondents were asked whether they usedoutside consultants (Consultant), and whether ornot technical or consulting services provided bycounty governments were helpful (TechnicalAssistance).
Outcome variables
Impact of the MLPA of unit's plans
Two questions were asked: the first was: 'Howdifferent is the plan prepared under the MLPAfrom your earlier plan?" (How Plan Different)The seeond (Alternative Actions), was if theMLPA (1976) had not been enacted, what actionwould the community have taken in the last threeyears with regards to the comprehensive plan?'
Improvements in units' capabilities
Two questions assessed the act's impact on units'capabilities. The first, labeled Decision-Making,asked 'How did the project affect your unitof government's capability for making futuredecisions allocating resources for undertakingfuture endeavors?' The second, labeled LessonsLearned, asked 'To what extent will what was
Effectiveness of MLPA in solving problems
To assess the effectiveness (Effeetiveness) of theMLPA, subjects were asked (a) How effectivewill the MLPA be in meeting regional needs andsolving regional problems? and (b) How effectivewill the MLPA be in meeting local needs andsolving local problems?
RESULTS
Table 1 shows the correlation matrix for all25 variables utilized in this example. To guide usin examining the utility of the four models wedecided to examine only those paths that resultedfrom the 52 hypotheses on which the originalstudy was based (see Figure 2).-' The resultsof these hypotheses are shown graphically inFigure 3. One must be careful in drawinginferences when examining a iarge number ofcorrelations, especially with a small sample size,since some will be significant or non-significantby chance. Also, the level at which all thesignificant correlations hold simultaneously isconsiderably less than p < 0.05. The alternative,of course, would be to examine only a fewvariables, thus reducing the likelihood of commit-ting type I and II errors. We do not believe,however, that the current state of knowledgeoffers any reliable way of identifying the most'crucial' variables.
Other qualifications are also necessary. Inaddition to possible sample bias, there was
See footnote 2.
220 K, B. Boal and J. M. Bryson
SUOSS91
uoiieDiunuimoj
JO
0861
llOI}BlU3I113|dlll|
a if l l
d 12. a.
s:a. o — (L s ^ o .
• S § i: - . d 5 =•=>
5 S » S
r 2 n.
^J
f i t
S a. o* £
iii a. cix; 6,
s a
f? i
3 3 13SS. Ss
Representation, Testing and Policy Implications of Planning Processes 221
SUPPORT OF MC FORLOCAL UNITS ASPERCEIVED BYLOCAL UNITS(Goal Support)
COMMUNICATIONFREOUENCY OF
COMMUNfCATION(Frequency ot
Communication)
CHANGES;NCOMMUNICATION
PATTERNS(Communication
Changes)
CONFLICT ANDCONFLICT RESOLUTION
AVERAGE CONFLICTISSUES #1-#3
(Average Con (hot)
AVERAGE NUMBEROF CONFLICT
RESOLUTION METHODS.ISSUES #1-#3
iConilict ResolutionMeltiods)
CONSULTATION ANDTECHNICAL
ASSISTANCE
CONSULTANT USED INPLAN PREPARATION
(consultant)
HELPFULNESS OF TAFROM COUNTY
(TechnicalAssislarvre)
IMPACTON UNITS' PLANSHOW DIFFERENTIS PLAN(How Plan Different)
WHAT ACTIONWITHOUT MLPA(Alternative Action)
IMPROVEMENTS INUNITS' CAPABILITIES
EFFECT ON DECISIONMAKING CAPABILITY(Decision Making)
USEFULNESSOF LEARNING(Lessons Learned)
SATISFACTIONTOTAL SATISFACTION(SBttBfaction)
PERCEIVEDEFFECTIVENESSOF MLPA
PERCEPTION OFEFFECTIVENESS OF MLPA(Effectiveness)
Figure 3. Statistically significant Pearson Product-Moment correlation among context, process, and outcomevariables
considerable opportunity for bias in the responsesdue to differences in our respondents' positionsand experience with their organizations, and tothe possibility that they interpreted our questionsdifferently. There also were no 'controi' organi-zations.
In spite of these qualifications, we feel largeexploratory studies such as this one have consider-able potential for helping us to understand thedynamics of planning.
Of the 31 bivariate correlations testing thehypotheses relating context to process and out-come variables, seven were in the predicted
direction and were significant at p < 0.05 level.An additional 17 were in the predicted directionbut were not statistically significant. Two corre-lations were statistically significant but not In thepredicted direction. Five correlations were notin the predicted direction but were not statisticallysignificant.
The seven supported hypotheses relating con-text and process variables are ds follows:
— The more ample the time and resourcesavailable to local units for plan preparation,the greater their acceptance of the MLPA
Representation, Testing and Policy Implications of Planning Processes 223
and its associated implementation process.(r = 0.23, p < 0.05)
— The greater the resources available to localunits for plan preparation, the less frequentthe communications among the potentiallyaffected organizations, {r = 0.23, p < 0.05)
— The greater the dependence of local unitson grant money for plan preparation (GreatDependency) and for decision-making basedon the plan (Money Dependency), the morespecific the local units perceived the MLPAand MC to be about goals, (r =̂ 0.21.p < 0.05)
— The more adequate the number of pro-fessionals available to local units for planpreparation, the greater the perceived effec-tiveness of the MLPA. (r = 0.28, p < 0.05)
— The greater a local unit's population in 1980,the less likely a consultant would be used toassist it with plan preparation, (r = -0.47,p < 0.05)
— The greater the role changes of actors as aresult of the MLPA, the greater the changesin communication patterns, (r = 0.80,p < 0.05)
As noted, two hypotheses were statisticallysignificant, but in the opposite direction. Un-expectedly, the more resources a unit hadavailable for plan preparation, the less theacceptance of the MLPA and its associatedimplementation process (/? = 0.27, p < 0.05) andthe less the unit was likely to feel it was supportedby the Metropolitan Council (r = -0.28,p < 0.05) Upon reflection, however, these resultsmay not be surprising. The more resources a unithas, the more autonomous the unit is likely tofeel, and, consequently, the less likely it will beto accept externally imposed goals and contacts(Pfeffer and Salancik, 1978).
Thirty-two bivariate correlations were exam-ined in order to test hypotheses relating processand outcome variables. Fourteen were in thepredicted direction and statistically significant atthe p < 0.05 level. An additional 14 were in thepredicted direction, but were not statisticallysignificant. One statistically significant correlationwas not in the predicted direction. Threeadditional correlations were not in the predicteddirection, but were not statistically significant.
The 14 supported hypotheses relating processto outcome variables are as follows:
— The greater the goal acceptance by localunits of the MLPA and its associatedimplementation process, the greater theperceived improvement in units' decisionmaking capability, the greater their totalsatisfaction, and the greater their perceivedeffectiveness of the MLPA. (r = 0.26,p < 0.05; r = 0.42, p < 0.05; r - 0.47,p < 0.05)
— The more local units perceived the MLPAand MC to be specific about goals, the morethe local units perceived the MLPA to beeffective, (r =0.39, p < 0.05)
— The more the local units felt they weresupported by the MC. the more the localunits felt their decision-making capabilitieswere improved, the more satisfied they werewith the outcomes of the process, and themore they felt the MLPA was effective.(r = 0.39, p < 0.05; r = 0.39, p < 0.05;r =0.56, p < 0.05)
— The more frequent the communicationamong affected units, the more they felt theyhad learned, and the greater their totalsatisfaction with the outcomes of the process.(r = 0.36, p < 0.05; r = 0.34, p < 0.05)
— The greater the average number of confiictresolution methods used by local units,the more they felt their decision-makingcapabilities were improved, the more theyfelt their learning was useful, and the morethey were satisfied with the outcomes ofthe process, (r = 0.32, p < 0.05; r = 0.30,p < 0.05; r = 0.31, p < 0.05)*
— The more a consultant was u.sed by localunits to help with plan preparation, the morethe actions of the local units were differentfrom what they would have been in theabsence of the MLPA. and the more thelocal units felt their learning was useful,(r = 0.39, p < 0.05; r = 0.23, p < 0.05)
As noted, one statistically significant correlationwas not in the predicted direction. We weresurprised that the use of a consultant usuallyresulted in a plan little different from previousplans (r = 0.31, p < 0.05). Anecdotal evidenceprovides two possible explanations. First, manyunits hired a consultant simply to make sure theirplans met the MLPA's requirements. Second,many consultants turned out standard, unexcep-tional plans that relied mainly on previous plans.
224 K. B. Boal and J. M. Bryson
intervening Effects model
After testing out a prior hypotheses, our nexttask was more exploratory—namely, the fittingof our data to the different theoretical models.Based on our original conceptualization (Fig-ure 2). we expected one of the intervening models(i.e. context leads to process which in turn leadsto outcomes) to be most representative of thedata. Thus this was the first model we examined.To test our assumption we examined all potentialpaths between context and outcomes controllingfor process effects."* Of the 82 partial correlationsexamined, in only two situations was there anintervening effect. It is interesting to note thatboth situations involved the size of the community(1980 Population) and the use of an outsideconsultant (Consultant) which acted as interven-ing variables. In the first situation the resultsindicated that smaller communities used outsideconsultants, which in turn resulted in actionswhich were different than they would have beenotherwise (Alternative). In the second case,because the use of a consultant (Consultant) waspositively correlated with satisfaction (Satisfac-tion), the zero-order correlation between sizeand satisfaction was suppressed. The resultsindicate that both larger units and units who usedconsultants were more satisfied. Thus there isvery little support overall for an InterveningEffects model. (We must note again, however,that we did not examine for the possibilitythat context intervenes between process andoutcomes.)
Independent Effects model
The results from the bivariate and partialcorrelations indicated that the IndependentEffects model might be the most representativeof the data. However, the bivariate correlationsdid not examine the simultaneous effects of bothcontext and process variables on outcomes.Therefore, to further test this model, we ranstepwise multiple regressions on the paths.
In all but one of the 42 applicable cases theresults indicated support for a single main effectsmodel linking process variables to outcomevariables (refer to Figure 3 and bivariate corre-
lations).'̂ In only one case did a context variable(1980 Population) contribute independently alongwith a process variable, use of a consultant(Consultant), to an outcome variable (HowPlan Different). Together these accounted forapproximately 22 percent of the explained vari-ance (F - 6.93, p < 0.00; R~ = 0.221).
The amount of explained outcome variancethat any single context or process variable canaccount for can easily be determined hy squaringthe correlation coefficient reported in Table 1.The reader will observe that the strongestrelationships occur generally between the goal-setting variables and the outcome variables. Thesingle strongest relationship was between goalsupport and effectiveness (r = 0.56: R- = 0.31).
Moderating Effects model
To test the possibility that contextual or processvariables moderated the relationship betweenprocessorcontext and outcomes, we ran hierarch-ical moderated regressions. Results supported amoderating effects model in tour cases. Theyare:
1. Ampleness of time available for plan prep-aration moderates the relationship betweenacceptance of the MLPA and satisfactionwith outcomes of the planning process. Inother words, units that accepted the goalsof the MLPA were satisfied with theoutcomes of the process; however, theirlevel of satisfaction was greater the morethey believed they had ample time forcompleting the plan.
(F, „ = 3.29, p < 0.05; R~ = 15.7)2. The use of a consultant moderated the
relationship between the size of the com-munity and action that community wouldhave taken without enactment of the MLPA.That is, larger communities would haveupdated their plans in a manner consistentwith the MLPA, even if the MLPA had notbeen passed. However, when they used anoutside consultant they tended to opt forplans similar to those they already had. In
* To save space we have omitted the partial correlations.They are reported in full in the paper cited in footnote 2.
•* Missing data, and thus a reduction in sample size, accountsfor the differences in the number of signiHcant bivari;Uecorrelations (46) vs. the number of significant multipleregressions (42).
Representation, Testing and Policy Implications of Planning Processes 225
other words, use of a consultant by largercommunities tended to result in plans lessin keeping with the intent of the MLPAthan would have occurred otherwise.
fF.,.54 = 6.01, p < 0.01; /?^ = 16.5)3. The ampleness of resources for plan prep-
aration moderated the relationship betweengoal support from the Metropolitan Councilas perceived by the local units and theirperceived effectiveness of the MLPA inmeeting local and regional needs. The resultindicates that units that felt the MC wasresponsive to their needs perceived theMLPA as more effective. This was especiallyso when they also felt they had ampleresources for plan preparation.(F3 3, = 6.90; p < O.Ol; R- = 0.400)
4. The ampleness of resources for plan prep-aration also moderated the relationshipbetween frequency of communications andthe usefulness of what was learned. Thus,frequency of communication was associatedwith greater usefulness of learning,especially when units had ample resourcesfor plan preparation.
(̂ 3.48 = 4.01, p < 0.01; R~ - 0.200)
Interaction Effects model
Comparisons of the results between the stepwisemultiple regressions and the hierarchical moder-ated regressions supported a 'pure" interactioneffects model in two cases. They were:
1. The effect of the MLPA on the units'decision-making capability was jointlyinfluenced by the frequency of communi-cation and the ampleness of resources forplan preparation. We interpret this to meanthat in the absence of either frequentcommunication with other units or ampleresources for plan preparation, the MLPAhad no effect on the units' decision-makingcapability.
(F^^^, = 4.51, p < 0.05; R- = 0.083)2. Units' decision-making capability was also
jointly affected by the total number ofgroups involved and the number of conflictresolution techniques used. Thus, whenmany groups were involved in the planningprocess, decision-making capability wasimproved only when many conflict reso-
lution techniques were used to resolvedifferences among the groups.
(F,,49 = 8.06, p < 0.01; R~ = 0.141)
DISCUSSION
The purpose of this study was twofold: (I) toexplore theoretically the interrelationshipbetween context, process, and outcome variables;and (2) to examine empirically these relationshipsto see if they hold some promise for futureresearch. Our findings indicate that on boththeoretical and empirical grounds the four modelswe have developed make sense.
The basic empirical findings were quite surpris-ing, given our initial conceptualization, namely,that the Intervening Effects model would pre-dominate. While there were numerous linksbetween context and process and between processand outcomes {i.e. bivariate correlations), thecontext variables generally were not causallylinked to outcomes. Certain contexts seem tomake the use of certain processes more probable.They are not determinant, however. Planners, itwould appear, have considerable discretion inthe use of planning processes.
This finding is significant, particularly in lightof recent theorizing based in the ecologicaltraditions of organizational research. The extremeversion of the ecological approach argues thatenvironments select successful organizations andthat managers or planners can do little to affectoutcomes. Managers or planners, in others words,are rigidly constrained by their contexts (Aldrich,1979; McKelvey. 1982). Our research indicatesthat at least for our sample of organiziitions suchwas not the case. Managers and planners coulddo quite a lot to affect outcomes positively.
In fact, the strongest relationships observed inour study were direct process-outcome linkages.Particularly strong were the relationships betweengoal-setting and outcome variables. The impli-cation is that efforts to improve goal acceptance,specificity, and support for subordinate units willpay off in increased impact on units" plans,improvements in units" capabilities, increasedsatisfaction of units, and greater perceived effec-tiveness of the overall inter-organizational plan-ning system. This finding extends to the inter-organizational level the large body of research thatdemonstrates the impact of task-goal attributes on
226 K. B. Boal and J. M. Bryson
performance at the micro-organizational level(see Locke et al., 1981). Our results stronglysupport the assertion that these micro-relation-ships hold at the macro-level as well.
We also found that more frequent communi-cations and greater conflict resolution effortsincreased the likelihood of favorable outcomes.Thus planners may be well advised to payattention to the development and application ofcommunication and conflict resolution skills.
Context variables (except for size) were alwaysimportant in conjunction with process variables.There may be several reasons for the relativeunimportance of the contextual variables. First,people really may not have been causally influ-enced by context as they constructed processesto affect outcomes. Second, we may have focusedon the wrong set of context, process and outcomevariables. Third, context may not constrainprocess-outcome relationships very much. Incomprehensive planning certain things—such asbasic studies—need to be done regardless ofcontext, and there may not have been muchroom left for variation in response to otherfactors.
Finally, our study illuminated how context andprocess can be manipulated differentially whenhierarchical levels of government are involved ina planning process. In the situation we studied,the MLPA and MC set much of the context andprocess for local units of government. Forexample, the MLPA outlined the general processto be followed, along with general plan contentsand actors to be consulted, while the MCprepared the systems statements for each unit.Local units therefore had limited discretion inmanipulating context variables, and somewhatmore discretion in controlling process variables.More numerous context—process—outcomerelationships might have existed in the absenceof the MLPA and the MC. But the MLPA andMC do exist, and given that fact we may nothave been able to detect morecontext-process-outcomc relationships becausethe MLPA and MC may have eliminated muchof the variation in our variables. This explanationactually is the most intriguing, because it impliesthat the real power of the MLPA and MC comesfrom establishing the premises underlying decisionmaking, rather than from prescribing specificactions or outcomes. March and Simon (1958),for example, argue that establishing the premises
underlying decision making is a far more powerfulbehavioral control than is prescribing specificactions or dictating specific decisions.
IMPLICATIONS FOR FUTURERESEARCH
This study has two basic implications for futureresearch. The first is the need for more longitudi-nal studies of the connections between context,process and outcome variables; the second isthe need for theoretical rc-examination of theplanning process.
The present study is one of the few large cross-sectional studies analyzing the interrelationshipsof context, process and outcome variables inpublic sector interorganizationul planning sys-tems. In several instances, statistically significantrelationship emerged. The need for longitudinalstudies is obvious, however, if we are tounderstand fully how context and process arerelated and how they affect outcomes. Van deVen's work (1980a-c) is virtually the onlyexemplary research of this sort. In addition tomeasuring process variables in a way similar tothis study"s measures, he was able to demonstratethe superiority of a specific normative sequenceof steps (the Program Planning Model of Delbecqand Van de Ven, 1971) over a more random,unstructured process. Only longitudinal studieswill reveal precisely which normative sequencesor phases work best in which circumstances, andprecisely how context, process and outcomes arerelated over time.
The need for a theoretical re-examination ofthe planning process is also implied in our study.The planning process typically is conceptualizedin one of four ways: (1) as not worthy ofattention, because the appropriate process isdetermined by an accurate specification of contextand desired outcomes; (2) as a normative ordescriptive sequence of phases or steps; (3) asspecific tactics or tasks to be completed withineach phase of a general problem-solving model;or (4) as generic activities either within phasesor across a general problem-solving sequence.No matter how the process is conceptualized,however, what is not known is exactly howcontext, process and outcomes arc related. Inour study we originally conceived of the planningprocess principally as generic activities intervening
Representation, Testing and Policy Implications of Planning Processes 227
in a causal sequence from context to process tooutcomes. We were surprised to find processvariables usually had independent effects—andonly occasionally intervening, moderating, orinteraction effects—on outcome variables.
Our research is only a very preliminary effortto answer the question of how to conceptualizeand test connections among context, process andoutcome variables. Although our research hasseveral implications for practice, obviously morework is needed. The reward, we believe, will besignificant improvements in public and privateplanning practice.
APPENDIX: VARIABLES AND THEIROPERATIONALIZATION
Contextual Variables
Resources: Five-point Likert scale where 1 indi-cates tight resources and 5 indicates ampleresources
1. Ample time: How ample has the calendar timebeen for completing your comprehensive plan?(.V = 3.1, SD = 1.05)
2. Financial resources: How generous is yourlevel of financial resources for the preparationof the comprehensive plan? {x = 2.55,SD = 0.80)
3. Decision-making resources: How generous is/will be the amount expressed above (theywere asked to indicate the approximate dollaramount of resources available to them eachyear for decision-making) for decision-makingbased on the comprehensive plan? {x = 2.38,SD = 0.79)
4. Grant dependency: What is your dependenceon grant money for the preparation of thecomprehensive plan? {x = 3.42, SD = 1.28)
5. Money dependency: What is/will be yourdependence on grant money for implemen-tation decision based on the comprehensiveplan? {x = 2.13, SD - 1.49)
6. Adequate personnel: How adequate do youfeel the above number of full-time equivalent(FTE) professional (they were asked to indi-cate the number), in addition to a consultant,was to prepare the comprehensive plan?{x = 2.86, SD = 0.95)
Demography: We report here only the onesrelating to size and growth.
1. 1980 population: Respondents were askedwhat their local units' population was in 1980(x = 44,110, SD = 133.086).
2. Population growth: Census data were used todetermine the percentage change in thepopulation base between 1970 and 1980(.V = 33.29. SD = 42.44).
Number of groups: Respondents were asked toindicate whether or not each of twelve groupshas/had either: (1) a formal (required) role inplan preparation; (2) plays, played, or will playan active role (whether required or not) in theplan preparation; (3) a formal (required) role inthe plan implementation; or (4) plays, played, orwill play an active role (whether required or not)in implementation. These were answered yes/no,and a summary index Cv = 1. /V = 0) wascalculated (x = 15.11; SD - 7.50).
Role changes: Respondents were asked whetherthe roles played by the 12 groups (referredto above) represented a change from earlierrelationships, or were they a continuation ofestablished practices. The responses were coded(yes = 1, no = 0) and summed {x = 2.21;SD = 3.02). This is treated as an indicator ofenvironmental stability.
1. How reasonable are the systems statementsrequired by the MLPA (1976) as a device forcoordinating metropolitan and local decision-making?
2. How reasonable arc the plan content require-ments presented in the MLPA (1976)?
3. Given the intentions of the MLPA (1976),how reasonable is the financial burden thatresulted?
4. How reasonable are Metropolitan Councilexpectations of communities regarding thecomprehensive planning and implementationprocess?
228 K. B. Boal and J. M. Bryson
5, How reasonable was the time schedule pre-sented in the MLPA (1976)? (jc = 3.2;SD - 0.60; a = 0,65)
Goal specificity (Two items): Five-point Likertscale where 1 indicates very unclear and 5indicates very clear.
1. How clear is the MLPA (1976) regardingthe expectations and requirements of theMetropolitan Council and affected govern-mental units?
2. How clearly does the Metropolitan Council,as an organization, present their expectations?{x = 2,92; SD = 0,63; a = 0.53)
Goal support (Four items): Five-point Likertscale where 5 indicates a high support orresponsiveness.
1. How responsive is the Metropolitan Council,as an organization, to regional needs andproblems?
2. Do you think the Metropolitan Council andits staff took the time required to becomefamiliar with your community and its circum-stances?
3. During the system statement and plan reviewprocess, did the Metropolitan Council and itsstaff give you sufficient opportunity to presentand explain your jurisdiction's views andpositions?
4. (Assuming the unit had submitted their planfor an informal review) How helpful was thereview in developing your final ComprehensivePlan, {x = 3,42; SD = 0,64; a - 0,52)
Frequency of communication: Five-point Likertscale where 1 indicates never, and 5 indicatesvery frequent. Respondents were asked: 'Howfrequent were communications related to thecomprehensive planning and/or implementationprocess between your unit and each of thefollowing types of organizations (nine werespecified from 1 July, 1977 to 1 July, 1981?Consider face-to-face conversations, telephoneconversations, letters, and reports,' {x = 2.46;SD = 0.06; a = 0,55)
Communication changes (Nine items): respon-dents were asked whether 'the communicationpattern noted (in the previous questions) rep-
resents a change from earlier relationships, oris it a continuation of established patterns?'Responses were coded (yes - 1; no ^ 0) andsummed {x = 2.77; SD = 2.46).
Average conflicts: The respondents were firstasked to identify three important areas of conflict.Then for each issue they were asked to indicatethe degree to which the conflict on each issueinvolved the following:
A: Philosophical conflict—over what should bedone
B, Conflict over location—over where somethingshould be done
C, Conflict over means—over how somethingshould be done
D, Conflict over timing—over when somethingshould be done
E, Conflict over extent—over how much ofsomething should be done
Responses were on a five-point Likert scaleranging from not at all (1) to almost completely(5). The coefficient alphas, means, and standarddeviations for each issue were:
Issue 1: a = 0.57; x = 2.95; SD = 0.82Issue 2: a = 0.67; x = 3.00; SD = 0.87Issue 3: a = 0.77; x = 2.96; SD = 0.98
(It should be noted that the specific issues werenot the same across respondents,) The threesubscales were then summed and averaged intoan overall scale of average conflict. The meansand standard deviation for this scale are: J: = 3.01;SD - 64.
Average conflict resolution methods used: Thisvariable was operationalized by first having thesubjects indicate how many of 20 possible conflictresolution methods were used to resolve theabove three areas of conflict. The respectivemeans and standard deviations for each issue are:
These three subscales were then summed andaveraged, (x = 3.83; SD = 2.00)
Consultant usage: Respondents were asked, 'Wasa consultant hired to assist with the preparation
Representation, Testing and Policy Implications of Planning Processes 229
of your comprehensive plan?" Answers werecoded (yes = 1; no = 0). (.r = 0.78; SD = 0.42)
Technical assistance: Five-point Likert scale where1 = hindered more than helpful, and5 = extremely helpful, 'How helpful was thereceived technical assistance or consulting servicessupplied by the county?' (,v = 3.87; SD = 1.10).Not all units requested or received technicalassistance or consulting services. The number ofrespondents to this question was 23.
Outcome Variables
Impact on unit plans: Five- and four-point Likertscales respectively where a low number indicateslittle impact
1. How plan different: 'How different is the planprepared under the MLPA from your earlierplan? {x = 3.20; SD - 1.20), Fifteen respon-dents indicated they had no earlier plan,
2. Alternative action: 'If the MLPA (1976) hadnot been enacted, what action would thecommunity have taken in the past 3 years withregard to the comprehensive plan? (JC = 2.55;SD = 0.95)
Improvements in unit's capabilities: Five-pointLikert scale where 1 indicates little improvementor usefulness and 5 indicates great improvementor usefulness
1, Decision-making: 'How did the project affectyour unit of government's capability formaking future decisions allocating resourcesor undertaking other future endeavors?'(,v - 2,66; SD = 0.54)
2. Lessons learned: 'To what extent will whatwas learned from the comprehensive planningand implementation process be useful forundertaking future comprehensive planningand implementation endeavors' {x = 3,41;SD = 0.82)
Satisfaction (Six items): Five-point Likert scalewhere 1 = extremely dissatisfied, and5 = extremely satisfied. Respondents indicatedhow satisfied they were with the comprehensiveplan's ability to improve decision-making in sixareas: local unit's ability in general, land use,public facilities, low and moderate housing
implementation, capital improvements, andwhether the official controls identified will actu-ally achieve objectives, {x = 3.49; SD = 0.69;a = 0.83)
Effectivenss (Two items): Five-point Likert scalewhere 1 = very ineffective, and 5 = very effec-tive.
A. How effective will the MPLA (1976) be inmeeting regional needs and solving regionalproblems?
B. How effective will the MLPA (1976) bein meeting local needs and solving localproblems? {x = 0.307; SD = 0,87; a = 0.68)
ACKNOWLEDGEMENTS
Earlier versions of this paper were presented atthe 43rd Annual Meeting of the Academy ofManagement, Dallas, Texas. 14-17 August 1983,and at the 25th Annual Conference of theAssociation of Collegiate Schools of Planning,San Francisco, California, 21-23 October, 1983.The authors would like to acknowledge theirimmense gratitude to the Hubert H, HumphreyInstitute of Public Affairs and to the Center forUrban and Regional Affairs at the University ofMinnesota for their financial support of thisstudy. In addition, the authors would like toexpress special thanks to Thomas Anding, Bar-bara Crosby, Richard Efford, Robert Einsweiler,Karen Fladmoe-Lindquist, Neil Gilbert, SusanSchmidt, Thomas Scott, Harry Specht, LouiseStraus, and Andrew Van de Ven. along withDan Schendel and two anonymous reviewers ofthis journal, for their helpful comments. Theauthors, of course, remain solely responsible forthe paper's content.
REFERENCES
Aldrich, H, E, Organizations and Environments,Prentice Hall, Englewood Cliffs, NJ, 1979,
Alexander, E, R. 'After rationality, what? A reviewof responses to paradigm breakdown'. Journal ofthe American Planning Association, 50(1), 1984,pp, 9-21.
Arnold. H, J. "Moderator variables: a clarification ofconceptual, analytic, and psychometric properties'.Organizational Behavior and Human Performance,29, 1982, pp. 143-174.
230 K. B. Boal and J. M. Bryson
Beach, L, R, and T, R, Mitchell, "A contingencymodel for the selection of decision strategies'.Academy of Management Review, 3, 1978.pp, 439-449,
Bolan. R, and R. Nuttall, Urban Politics and Planning.D, C, Heath, Lexington. MA, 1975,
Bryson, J, M, 'A framework for studying a majorexperiment in metropolitan growth managementin the United States', Proceedings of the 22ndAnnual Conference of the Midwest Dlvisioti of theAcademy of Management.. Division of Research,Graduate School of Business Administration, TheUniversity of Michigan, Ann Arbor. MI, 1979,pp, 102-117,
Bryson, J, M, 'A perspective on planrling and crisisin the public sector'. Strategic Management Journal,2, 1982. pp, 181-196,
Bryson, J, M, "Representing and testing proceduralplanning methods*, in Masser, I. (ed,). EvaluatingUrban Planning Efforts, Gower Publishing Co,.Aldershot. England, 1983. pp, 245-268,
Bryson, J, M, The policy process and organizationalform". Policy Studies Journal. 12(3). 1984.pp, 445-463,
Bryson, J, M, and K, B, Boal, 'Strategic managementin a metropolitan area: the implementation ofMinnesota's Metropolitan Land Planning Act of1976', Working Paper, Hubert H. HumphreyInstitute of Public Affairs, University of Minnesota.Minneapolis, MN. 1982.
Bryson, J. M. and K, B, Boal. 'Strategic managementin a metropolitan area: the implementation ofMinnesota's Metropolitan Land Planning Act of1976', in K, Chung (ed,), Academy of ManagementProceedings 1983. pp, 332-336,
Bryson, J, M, and J, W, Cullen, 'A contingentapproach to strategy and tactics in formative andsummative evaluations'. Evaluation and ProgramPlanning. 7, 1984. pp. 267-290.
Bryson. J. M, and A. L. Delbecq. 'A contingentapproach to strategy and tactics in project planning'.Journal of ihe American Planning Association, 45.1979, pp, 167-179,
Bryson,. J, M,. K, B. Boal. S, Poole and C. Terrell,'A contingent planning model for programs andprojects', Project Management Quarterly.. 10(1),1979, pp, 19-29,
Busemeyer, J, R, and L, E, Jones, 'Analysis ofmultiplicative combination rules when the causalvariables are measured with error'. P.svchologicalBulletin. 93, 1983, pp, 549-562,
Cleveland, H, The Future Executive. Harper and Row.New York, 1973,
Cleveland, H, The Knowledge Executive. E, P, Dutton,New York, 1985,
Deibecq. A. L, and A, H, Van de Ven, 'A groupprocess model for problem identification andprogram planning'. Journal of Applied BehavioralScience, 7, 1971, pp, 466-492.
Delbecq, A, L,, A, H, Van de Ven and D, Gustafson,Group Techniques for Program Planning, Scott,Foresman, Glenview, IL, 1975,
Etzioni, A, "Mixed-scanning: a ""third" approach todecision-making'. Public Administration Review,27(5), 1967, pp, 385-392,
Filley. A, C, Interpersonal Conflict Resolution. Scott,Foresman, Glenview. IL. 1975,
Freeman, R, F. Sirategic Management: A StakeholderApproach. Pitman. Marshfield, MA, 1984,
Gilbert, N, and H. Specht, Dynamics of CommunityPlanning. Ballinger. Cambridge. MA, 1977.
Harold, D. M, 'Long-range planning pays off, BiLsine.s.<iHorizons, 13, 1972. pp, 81-87,
Harrigan, J, J, and W, C. Johnson, Governing theTwin Cities Region, University of Minnesota Press.Minneapolis. MN, 1978,
Hofer, C, W. and D. Schendel, Strategy Formulation:Analytical Concepts, West. St Paul, MN. 1978.
Huber. G, P. Managerial Decision Making, Scott.Foresman, Glenview, IL, 1980,
Janis. L L, and L, Mann, Decisiott Making. APsychological Analy.tis of Conflirt. Choice andCommitment. Free Press. New York. 1977,
Kartez, J. D. 'Crisis response planning: toward acontingent analysis,' Journal of ihe AmericanPlanning Association. 50(1), 1984, pp. 9-21,
Lindblom, C. E, The science of muddling through'.Public Administration Review. 19. I959,^pp. 78-88,
Lindblom, C. E, Politics and Markets, Basic Books,New York. 1977,
Locke, E. A,, K, N, Shaw. L, M. Saari and G, P.Latham, 'Goal setting and task performance:1969-1980', Psyehological Bulletin, 90(1). 1981.pp. 125-152.
Malik, A, and D. Karger, 'Does long-range planningimprove company performance?" ManagementReview, 64, 1975. pp, 27-31.
March, J. G. and H. Simon, Organizations, BasicBooks, New York. 1958,
McCaskey, M, B, "Contingency approach to planning:planning with goals and planning without goals'.Academy of Management Journal. 17. 1974,pp, 281-291.
McKelvey, W, Organizational Syxtematics, Universityof California Press, Berkeley and Los Angeles,CA. 1982,
Mintzberg, H, The Structuring of Organizations,Prentice-Hall. Englewood Cliffs. NJ, 1979.
Mintzberg, H,, D. Raisinghani and A, Theoret. Thestructure of "unstructured" decision processes'.Administrative Science Quarterly. 21, 1976,pp, 246-275.
Nutt. P. C, 'Models for decision making in organi-zations and some contingencies with stipulateoptimal use'. AcadetJiy of Management Review. I.1976a. pp, 84-89,
Nutt, F, C, 'A field experiment which compared theeffectiveness of design methods'. Decision Science,7(4). 1976b, pp, 739-758,
Nutt, P, C, 'An-experimental comparison of theeffectiveness of three planning methods'. Manage-ment Scienee, 23(4). 1977, pp, 499-511,
Representation, Testing and Policy Implications of Planning Processes 231
Nutt, P, C, 'Stage-based.and process reconstructionparadigms for planning research', a paper presentedat the 43rd annual meeting of the Academy ofManagement, Dallas. TX. 14-17. August 1983,
Payne. J. W. "Contingent decision behavior'. Psy-chology Bulletin, 92. 1982. pp, 382-402,
Pfeffer. J. and G. R. Salancik, The External Controlof Organizations, Harper and Row, New York.1978,
Porter. M. E, Competitive Strategy, Free Press, NewYork. 1980,
Rainey. H, G, "Public agencies and private firms:incentive structures, goals and individual roles'.Administration and Society.. 15(2), 1983.pp. 207-242,
Rainey. J. G.. R, W, Backoff and C, H, Levine,"Comparing public and private organizations*. PublicAdministration Review, 36. 1976, pp, 23.3-244,
Ring, P, S, and J. L. Perry. 'Strategic managementprocess in public and private organizations: impli-cations of distinctive contexts and constraints',Academy of Management Review, 10(5). 1985.pp, 276-286,
Saunders. D, R, "Moderator variables in prediction'.Edtteational and Psychological Measurements. 16.1956. pp, 209-222.
Seashore. F, S, 'Group cohesiveness in the industrialwork group' Institute for Social Research, Univer-,sity of Michigan. Ann Arbor. MI. 1954,
Shaw. M, E, 'Communication networks", in L,Berkowitz (ed.). Advances in Experimental Social
Psychology, vol, 1. Academic Press, New York,1964. pp, 111-147,
Simon. H, A. Administrative Behavior, Free Press.New York, 1947,
Sloan, A. My Years with General Motors, Doublcday,New York, 1963,
Steiner. G, A,. J. B. Miner and E, R, Gray,Management Policy and Strategy. 2nd edn. Macmil-lan, New York. 1982,
Thompson. J, D. Organizations in Action, McGraw-Hill. New York. 1967,
Thune. F, F, and R, J, House, 'Where long-rangeplanning pays off. Business Horizons, 13, 1970.pp. 81-87.
Van de Ven, A, H, "Problem solving, planning, andinnovation. Part I. Test of the program planningmodel". Human Relations. 33. 1980a. pp. 711-740,
Van de Ven. A, H. 'Problem, solving, planning, andinnovation. Part 11, Speculations for theory andpractice'. Human Relations. 33, 1980b, pp, 757-779.
Van de Ven. A. H, "Early planning, implementation,and performance of new organizations', in J. R.Kimberly by R, H, Miles and Associates (eds,)The Organizational Life Cycle, Jossey-Bass, SanFrancisco. CA, 1980c. pp, 8.V134,
Van de Ven. A, H, and R. Koenig, 'A process modelfor program planning and evaluation'. Journal ofEconomics and Bu.sine.ss. 28. 1976. pp. 161-170.
Vroom. V, H, and P, Yetton. Leadership and DecisionMaking, University of Pittsburgh Press, Pittsburgh,PA, 1973,
![BORDERING CAPABILITIES VERSUS BORDERS: IMPLICATIONS FORsjs2/PDFs/Bordering_capabilities.pdf · Spring 2009] Border Capabilities Versus Borders 571 is a flawed representation and instead](https://static.documents.pub/doc/80x56/5ecd3f4fc597974194584eab/bordering-capabilities-versus-borders-implications-sjs2pdfsborderingcapabilitiespdf.jpg)
![[PPT]The - Welcome to Dr. Kim Boal's Website!kimboal.ba.ttu.edu/MGT 4380 Spring 2014/Presentations 099... · Web viewStrategy Canvas Diagnostic and an action framework for building](https://static.documents.pub/doc/80x56/5afb38237f8b9ad2209047fb/pptthe-welcome-to-dr-kim-boals-website-4380-spring-2014presentations-099web.jpg)
