research. 1100014-76-c-0362
TRANSCRIPT
DOCUBEhT,RESUMB
ED 137 338 TM 006 139
AUTHOR Pecorella, Patricia A.; Bowers, David G.TITLE Future Performance Trend Indicators: A Current Value
Approach to Human Resources Accounting. Report III.hultivariate Predictions of OrganizationalPerformance Across Time.
INSTITUTION Michigan Univ., Ann Arbor. Inst. for SocialResearch.
SPONS AGENCY Office of Naval Research, Arlington, Va. Personneland Training Research Programs Office.
PUB DATE Jan 76CONTRACT 1100014-76-c-0362NOTE 109p.; For related document see ED 132 160
EDRS PRICE MF-$0.83 HC-$6.01 Plus Postage.DESCRIPTORS *Accounting; Cost Effectiveness; Efficienc *Human
Resources; Industry; Management; ManpowerUtilization; Measurement; *Multiple RegressionAnalysis; -Organizational Development; *OrganizationalEffectiveness; *Performance; Performance Criteria;Prediction; *Productivity; Statistical Analysis;Surveys
ABSTRA TMultiple regression in a double cross-validated
design was used to predict two performance measures (total variableexpense and absence rate) by multi-month period in five industrialfirms. The regressions do cross-validate, and produce multiplecoefficients which display both concurrent and predictive effects,peaking 18 months to two years subsequent to the first wave of surveymeasurement. Some evidence of reciprocal causation occurs for absencerate. The human organization characteristics display remarkableconstancy across the years separating Waves 1 and 2 of surveymeasurement. Ho evidence of curvilinearity was found. It is concludedthat the value attribution steps, necessary for a current value humanresource accounting system, may he safely attempted. (Author
******* ***************************** ****************Documents acquired by ERIC include many informal unpublished
materials not available from other sources. ERICmakes every effortto obtain the best copy available. Nevertheless, items of marginalreproducibIlity are often encountered and this affects the quality *
* of the microfiche and hardcopy reproductions ERIC makes availablevia the ERIC Document Reproduction Service (EDRS).. EDES is notsponsible for the quality of the original document. Reproductions
supplied by EDRS are the best that can be made from the original.***************************g ********* ****************
TECHNICAL REPORT
January, 1976_
FUTURE PERFORMANCE TREND INDICATORS:A CURRENT VALUE APPROACH TO HUMAN RESOURCES ACCOUNTING
US. EPARTMENT OF NEA;.TH.EDUCATION IS WELFARENATIONAL INSTITUTE OF
EDUCATION
THIS DOCUMENT HAS BEEN REPRO-DUCED EXACTLY AS RECEIVED FROMTHE PERSON OR ORGANIZATIONORIGIN.ATING IT. POINTS OF VIEW OR OPINIONSSTATED DO NOT NECESSARILY REPRE-
',. SENT OFFICIAL NATIONAL INSTITUTE OFEDUCATION POSITION OR POLICY
MULTIVARIATE PREDICTIONS OFORGANIZATIONAL PERFORMANCE ACROSS TIME
Patricia A. PecorellaDavid-G Bowers
Institute for Social ResearchUniversity of MichiganAnn Arbor,liichigan
-This report was prepared under the Navy Manpower R&D Program of theOffice of Naval Research and monitored by the-Personnel and TrainingResearch ProgramS, PSychological 'Sciences Division, Office-of NavalResearch, under Contract No. N00014-764-0362, Work Unit No. NR 156-051.Reproduction in whole or in part is permitted for anyApurpcise of theUnited States Government. This document has been approved for publicrelease and sale; its distriution is unlimited.
-a-1CATION, OF THIS PAGE (137,en Dor. En:art.,e)
REP RT DOCUMENTATION PAGE READ INSTRUCTIONSBEFORE COMPLETING FORM-------.EPOR ' NUM3LN 2 1 C. - RECArIERT's CATALOG NUMBER
TITLE, nd S.hcfaFuture Performance Trend Indicators: A CurrentValue Approach to Human Resources Accounting:Report III
S. TYPE OF REPORT 6 PERIOD COVERED
Technical Report.
PERFORMING ORG. REPORT NUMnER
7 LS
Patricia A. PecorellaDavid G. Bowers
CONTRACT OR GRANT NU RfaJ
NO0014-76-C-0362
9. PERFORMInG ORGANIZATION NAME AND ADDRESS
Institute for Social ResearchUniversity of MichiganAnn Arbor, Michi an
10. PROGRAM ELEMENT. PROJECT, TASKAREA & WORK UNIT MURDERS
NR 156-051
11. CONTROLLING OFFICE NAME AND ADDRESS
Personnel & Training Research ProgramsOffice of Naval Research (code 458)Arlington, Va 22217
12 REPORT DATE
Januar 1977LI B OF PAGE
794. ZiONITOING AGE CY NA 1 ADORE5S(I!dIiirnC from Contru!ih'Q OiiIcJ IS. SECURITY CL AS P
Unclassified
IS DECLAcSIFICATION DOWNGRADINGSCHEDULE
15, EflcTRIRUIION STATEMENT 1 p
Approved for public release; distribution unlimited
7 ENs-RmuTiort STATEMENT (n1 nhn. m bqfrcI mit'ud in B 20. Ii dIii.rnt lroir.
li. sup L . ENTARY NOTES
Uport)
K EV 'MORO ontlnun on r'rr Id o Iinncoury nn.J IdønIIfy by block n
APST PACT (Cor.tln..i on rvoro 4.1.. Il v...ry end Id..atIiy by Woe( n r)
ultiple regression in a double cross-validated design was used to predict twoperformance measures (total variable expense and ahsence,rate) by multi-monthperiod in five industrial firms. The regressions do cross-validate, andproduce multiple coefficients which display both concurrent and predictiveeffects, peaking 18 months to two years subsequent to the first wave of surveymeasurement. Some evidence of reci'procal causation occurs for absence rate.The human organization characteristics display remarkable constancy across
the:years separating Waves 1 and--2 of survey measureMent. No-evidence
of curvilinearityllaS. found. It:is concluded-that -.the value attributionsteps', necessary for a-current value:human resource aCcounting system,may be safely attempted.
-
SECURITY Ct... ASSIFICATI
TABLE- OF CONTENTS
NTRODUCTION
Relationship Tre d Indicators_to Navy ManpowerProblems'
@
E._PM
1
METHODS . . . . 11
Measures of-Organizational Functioning 12
Measures of Periurmance 8
Analysis Procedures 25
. RESULTS- . . . . . 26
Limitations Upon Likely Relationships 26
Suitability of_the Data Set . 31
Multiple Regression Analyses: Wave I SOO = 34
MultipleRegression Analyses: Wave 2 SOO 0 . . . 43
-Comparisons Between Two Waves of Survey Data-. 51
Two Remaining Issues 58
-DISCUSSION 65
RgFgRENCES
DISTRIBUTION LIST ......... 74
FUTURE PERFORMANCE TREND INDICATORS:A CURRENT VALUE APPROACH T&HUMAN RESOURCES-ACCOUNTING
REPORT 1
NULTIVARIATE PREDICTIONS OFORGANIZATIONAL PERFORMANCE ACROSS TIME
Patricia-A. PecorellaDavid G. Bowers-
This report summarizes Phase I of a two-phase researth effort being
conducted to develop and refine a current-value human .resources accounting
procedure. Designed for-use by organization decision4akers-, the methodology
weuld be geared toward-providing "future performance.trend Indicators.
That is', it would provide estimates of the future productive potential of
today's human organization. Work in this area has been motivated in large ..
part by the frequent occurrenee of seemingly inapOropriate management-actions
.concerning _human resOurces utilization _and by the belief that key-decision
Makers:- lack of-certain-information fosters ineffective practices.
The situation is perhaps most clearly illustrated br-what.mayrbe termed
.the 'contingency parado." A ratlier'substamtial body ofevidence,indicates-
that better cost performance occurs under a more open4- 'participative"
-managementsystem than under a more rigid, "autocratic," tightly directed one.
Likert---1961, 1967; Drexler & Bowers, 1973; Franklin.& Drexler, 1976).
4
When the question is-posed directly to them, senior managers tend--to verify
.th s finding in their experience. Yet, when .confronted with.a need for
higher eMciency, Managements_typically move toward what has b2en shown to
be a less cost effective system---the rigid autocratic one (Likert-, '1967).
The problem here is a management system which be ieves that organizational
effectiveness can be attained--if not guaranteed-by (a) demanding particular
outputs, and (b)-manipulating various aspects of the organization's technical
and record-keeping systems_Seeming-short-rud-gains_do_result- from=these-
-practices: headcount reductions reduce payroll costs; faster equipment
allows faster production. The problem, however., is that- the gain-may_
be spurious, since long-term loss' may-instead be the result.'
Another examOl_e_of, the contingency .paradox is provided _by:L.awrence an'd
Lorsch (1969). They have found-that an organization's structure and func-
tioning should be responsive to file environment in which it operates. More
fluid unpredictable environments require internal -flexibility and an ability
to coordinate creative1y. Yet, in contradiction to accepted theory,
orqanizatjons whose,environments become more fluid and less predictable seem,
in fact, to turn toward rigid, bureaucratic methods for coping with their
uncertainty._
One explanation for these paradoxical practices is that the information
systems servicing managers and key decision-makers are deficient. First;
these systems typically 'provide detailed-readings on outcomes only,
detailed etatements-Of-production for the-previous month. No indication isfi
given as to.what conditions and events' led to the reported outcomes.
Furthermore, there is no guarantee that the combinationrof human organizati-on
functional charatteristics that led to the outcomes even exist an-y -longer,
although an assumption is made that it does. Second,- there exists a
time-lag warp in .conventional management information systems. They focus_
almost exclusively upon short-term outcomes-and provide little of. no data
-upon the relationship to longer range outcomes of the organization.
There is a need for additional information inputs to management decisions
f the inappropriate practices are to be corrected. An improved information
system must have the ability to assess the impact of current management_
procedures upon future effectiveness. That is, we need to recognize that,
with increasing complexity comes greater lag time--that the-effects of
today's human organi.4ation practices are felt further into the futule than
is true in simple'r instances. Such being the case, we need ai information
system that will provide-managers with inputs coneerning the likely impact
(in cost-effective terms) of present conditions upon future outcomes.
The idea of assigning cost-effective values to the human organization
:is not a new one. Brogden and Taylor (1950) proposed "the development of
an overa 1 index of an employee s value to the organization." They went on
to suggest that it be calculated in dollar units, determined on a cost
accounting basis. Recent attempts to gather these additional measueements
are known as Human Resources Accounting (Hermanson 1964). To date three
routes or methods have bo,en conceptualized:
(1) The "Incurred Cot" method -- measuring the amoun s already
invested in the human organization (Brummet, Pyle, & Flamholtz,
1968; Pyle, 1970a, 1970b).
The "Replacement Cost" method estimating the cost of replacing
the organization's hui resources (Flamholtz, 1969).
.The "Present Vale" Method estimating the future-productive..
potential of current human resources (Likert,.1967; Likert,
Norman, 1969; Likert & Bowers, 197
Our research is-co cerned with developing and refining a methodology._
for-Human Resources Accounting of a present value type. This arifftach
is generally recognized as theoretically desirable but operationally
difficult to implement.
Issues and Problems
The ability to fOrecast future productive potentia) depends upon our
possessing-adequate- knowledge-and-measuremnt-capabilitiesin- a-number of--
areas.
First, .we mu-' have identified the-key dimensions of the_human or aniza-
tion and accuired the abilit-- to measure these..key_dim§12aLgcLnIgly_.
Several theories in the psychological literature propose conceptual models
for understanding the functioning of human organizationS. Most of -hem lack
the necessary comprehensiveness, however, focusing instead upon one or two
isolated constructs, such as- 'motivation" "interpersonal relations."
In addition, very few of them focus upon the-causal flow of-events in
izational functioning. Yet theories are needed which describe ho-.. the
key dimensions interrelate across time.
A notable exception to the general lack of causalflow-proposition is
Likert's meta-theory, which places constructs in a causalintervening-end
result sequence (Likert, 1961, 1967; Bower 1976). Briefly, organizational.-
climate and managerial leadership are viewed as_ the major Causal variables,
Ober leadership and-group process as intervening variables, and satisfaction
and performance as end result variables. Figure 1 shows graphically the postu-
lated relationships among these variables. This causal flow of events-takes
place w tki-rt a framework of the organization as a system'of overlapping groups.
--
_
Figure
Relationships between Major Social-Psychological Factors and Outcomes
MAJAGERIALLEADERSHIP
ORGANIZATIONAL GROUPCLPATE PROCESS
PEERLEADERSHIP
SATISFACTIONJlealth, &Personnel
--Performance*
ProductiveEfficiency,FinancialPerformance,-& Quality,
Personnel-performance includes such factors as _urnover, grievance rateand-absence nate.
10
6
(The groups are described as "overlapping" because for all persons below the
very top and above the very bottom of the organization, each -is a member of
two groups simultaneously -he is a subordinate in the group iiiitediately
aboVe and a supervisor in the group -iniediately below.) The dual membership
implicit in this fact serves an integrating or linkage function for the
organization, that is,..it serves to.knit together the functiops, purposes
and needs of the various parts of the system.
Ecn.ally_important_is_the_fact=that =the theory is suPported_by_a=wealth,_.
of empirical evidence--indeed -it represents a crystallization- in coricePtual
form of a large Volume of empirical findings. Its comprehensiveness has
been tested in a variety of civilian settings (e.g., Bowers & Franklin, 1976).
ItS'aPplicability to two military settings has been tested as well (Bowers,
1975 1975b), and its major causal statements have been examined with
cross-time and cross-echelon analyses (Franklin, 1975a; 1975b).
A survey method has been developed by Taylor and Bowers (1972;) for
measuring the key dimensions in Likert's meta-theory with reasonable accuracy
and objectivity. It utilizes a standard, machine-scbred questionnaire
entitled the surLa_of_kmILLILiqal_(y11). The questionnaire has been used
extensively for both diagnostic and information feedback purposes within
organizational development studies. Utilizing Likert's meta-theory and the
survey methodology developed to measure its-principal dimensions,_w-e believe
that the first set of conditions can be met.
Second we must have valid indicators _ of the ormflization's effectiveness':.
Organizations typically employ multiple Criteria to evaluate their performance.
Ultimate criteria artHthose-butcomes directly related to the organization's
production goals, such as volume, cost quality, and efficiency. Penultima e---
criteria are inte-mediate rather than endresult outtomes such as attendance,
human costs, and resource developMent. This notion of performance criteria--
falling into a hierarchy of outcomes has been proposed by other researchers
as well e g., Seashore, 1965).
While most organizations collect performance data pertinent to one or
more of the above criteria there are several potential constraints on the
data's validity. More specifically the vllidity of performance data become
questionable when the following practices occur:
Changing standards or bases differentially from subunit to
subunit or period to period,
(b) maintaining common standards for all subunits, but in
situations in which the work nature or mix has changed
over time drastical y and differentially from subunit to
subunit,
(c )_agglomerating performance information into cost centers which
bear little or no resemblance to the real organizational
operating structure, and
relying upon collection procedures which systematically distort
reported results (Taylor & Bowers, 1972).
.
It is even possible that performance data are'deliberately "fudged"- when
the -control and reward sYstems of an organization encourage supervisory .
and r. -supervisory employees to protect themselves by .reporting inaccurate
Performance. figures. These situationS'also pose problems for :traditional
accounting methods and reports used-to assess- the .short-run profitability.
Nevertheless,_it is important to assess the validity of .the performanCedata
to.be used in developing fUture performance tHend.indicators....
Third we must establish the relationships between ke dimensions
of_the human or anization aldAerformance. Failure to find meaningful,
consistent relationships between functional and performance-properties of
_ the_organization seem to stem from limitations_in.theAata_or_methods,used__
to investigate them. Sometimes the wrong variables are attended to. At
Other times the correct variables are measured poorly, with ad hoc measuresL
of questionable reliability and construct validity. TypiCally, there is
a lack of awareness of time lag or insufficient data to assess the tiM lag
_
operatind. When problems in the quality of the-survey and organizational
data are taken into account and solved, as-we feel they have-beenin the
case of the data sets we propose to use, the relationships can emerge.*
Fourth,_ there must be evidence supporting the durability of changes in
organizational functioning_and effectiveness-. Little research_has been
_conducted: on this-topic. However, a'follow=up study (Seashore 8I-Bowers, 1970)
of a highly successful-organizatiOnat.develOpment prograM sudgested
--- :that changes-inlbusiness, outcomes, as well as in employee attitudes, that
_______resulted_froM_the_formal_change program (1962=1964) had persisted_several-
years-'hence.. While but one- study, the positive resulti are-encouraging.
FinallY,.there must be a Siatistical'technique for-comPuting futUre
perforMance trend indicators. -A procedure, developed-by,Likert and Bowers,
(1973) 1s-the One We propose totest and:refine. ..It_inVOlVes,(1) obtaining
regression:equations-:between-huMan organ zation variables And perforMance
variables; (2)- conVerting gains in One t_ (predicted) gains in the other,
3) removing the standard fromthe performance meatUre, and (4) capitalizing
7- *5ee7Tay1or ancrBowers (1972)-, PeCorella and Bowers (1976a,T1976b ) forzero=order:analyses of civilian Aataand.Bowers,(1973), Franklin and Drexler(1976),-:,Drexler and Franklin (1976) fori'comparableL'Analyses,in military
settings.
and discounting the reSult, based,.upon estimates of lag times obtjed
by research-
Relationshi of Trend Indicators to Nall Man ower Problems
Future Performance Trend Indicators tie in important ways to the work of
Dunnette- et al. _Dunnette et-al., 1973;-Borman 84..-:Punnette, 1974) whiCK
focused-on developing a personnel;status_index-for the Na4, --Like the ideal
product-mhich, theyconceptualized,'-thiS 'present .one mould be:
a single index whose_components remain=retrievable
.-.on a scale Ihich-permits=cross-time comparisons and which--
J.5 evaluative, not merely descriptive
--CePable of providing'estimates
:not just:for-individuals.
sensitive-to major fluctUa ions,-bu _ esiftan to-minor. _
ones
'credible to and_easily interpreted-by a
_and reasonablyresistantito fudgipg .
tisinga pollty._captdring methodology, they identified several major
components of such an index. Three components-stood-out as important potential --
indicators:for the Navr retention rate;discipline- as meas-ured-by
arge audience,
unauthorized .absence--rate -andrate Of-less-tham-honorable distharges) and
readineS meOured by_manning level and maintenance ratings_ _
Our research is attempting to-develop-a= means of forecasting outcomes of --
this type based upon key dimensions of theiiuman organization. Caplan and
Landekich (1974) say-that two steps would be involved jri tUch-a ventUre:
fi rst, estimate the amounts and timing of future benefits; second, estimate
the preSent-Valio? of those future-benefits-. The work to be reportecLhere,,
.focuSes- on the first phase of our.- proj_eCt;:the phase concerned wiift the first
f these tasks. In the subsequent phase of the research, value attribution
will occur: that is dollar conversions will be undertaken. The method
is being tested first in data from civilian sites. Upon successful completion
'of these anabises,----themethod will be tested for-its- general i zabi 1 i ty: to
-Navy". data sets.
11
METHODSa
Phase I of the project Ealled for multivariate analyses of data in the-
Organizational:_DevelOPment Research .Program s data _bank. In this final report-,-
for OhaSei- --data from--fiV\e.industrial.organizations (representingcontinubus\..
.
proCess and-a§semblYjine-_manufacturing)were investigated.. The :data sources,-
_
and.,analySl.s procedures:are-4eseribed below;
Daia_Sources
Between 1966 and 1970 dat' on organizational functioninganckperforMance
were collected from several industrial organizations as part of the Michigan
=
-Inter-Companyl_ongitudinal StudyAICLS_ Out of six potentially useful data-,
sets-from-this--study., five met all-of the eriterianecessary_for inclusion:in
least two waves of comparable organizational_ tune-
tioning data with measures of sufficient internal
consistency;
organizational performance measurements across
==.T
:with each perfbmante-Period-displaying-suffic
internal-consistency;
/!The-objectives,procedureS-,----.and TeSUlts.of. ICLS:-have-been:AeStribed-by
Likert et-al.(1969 ) -and:BOWerS- (1971 1973
12
zero-order relationships between organizational_func-
tioning and performance measures which weredireC-
tionally-appropriate and of_sufficient magnitude to
a.
merit proceeding with multivariate analyses.
Data meeting these criteria were available from a polyvinyl chloride
7-plant (Organization I) _two'assembly-plants of a large,,multi-location
manufacturing company (Organization fIT, a large oil refinerY (Organization
III), an aluminum extrusion mill (Organization IV)_and_three pap& and
cellophane miliS f another multMotation company (Organization_V
Measures of Orsanizational Functionin
ICLS was begun in order to make feasible the systematic investiga ion
of relationships between characteris ics of the human organization-and-_ _ _
performance levels of organi,zational units. The Survey of Organizations
questionnaire (SOO) a mathinerscored,-_standardized instrument, was_developed7
as-an integral part'of-this-research program;--The questionnaire waS heeded
to-collect comparable data-from diverse organizational sites in an economical
-and effitient Manner.- _The firstyersiOn of:fthe SOO-was Completed in 1965.-
While modifications have Since been Made- in the7questionnaire, most of:the
"core_measures remainedconsistent_across_the__ICLS sites.
-*Organiiation V,-a marketing firth, was_eXcluded because its perferMance measu eshad.--heen intricately_tonstructed-for:the:special TurpoSes orthelqCLS
-projedt-: They-.--werethe sourte_Of_suspitiOn toncerning:their qUalitY_then,-and-this-IsuSPicionremains=-7-What_tWteasUres produced'was a relativelylow fi7equency-of..diT'ktion4113-rcOrre.ct coefftdents.
n its current. edition
.,aspects,of the-rwork Setting'.
the SOO inc udes 124 items focusing on various
Six items focus on individual demographic
characteristics Forty-two additional spaces are provided for supplementary
questions tailored to a par icular organization or study. Responses to most
items regarding the work setting are recorded on a five-point extent scale
ranging from (1) "to a very little extent" to (5) "to a very great ekl.ent."
A description of the instrument together with statistical information regarding
_the validity and- reliability of its component elements -is provided_by_laylOr
.and Bowers (1972) ir the questionnaire manual.,
.fiVe:key dimensions of organizational, functioning are measured by the
SOO: Organizational Climate, Supervisory Leadership, Peer Leadership,
Group Process and Satisfaction. Organizational climate refers to the
-OrganizatiOnwide Conditions policies and practices within:which each work
group operatesThese conditions and_practices are created-fOr a:work group
by,other groups, especially those above it in the organizational.hierarchy.
Climate conditions set bounds op what does and what-can go:on' within any work,
Aspects of, climate can help or hinder conditions within g oups, or may
do both at the same time.
-Supervisory leadership comprises inte-rPerSonal and ta= -re ated
behaviors by supervisors as viewed by their subordinates. Peer leadership
comprises analogods interpersonal and task-related behaviors by work-_ .
group members toward each-other. Group process measures those things
_which_characterize the group as a team and whether group members work
together well or poorlY. The way in which group members share informati n,
make decisions, and Solve problems determines,the group's effectiveness and--.
the quality of its outputs. Satisfaction measures whether group
members are satisfied with economiCand related -reWards, the immediate
Supervisor, Vie- organization as- a- systeM,:the -job .aS a whole; compatibility
with fellow 14ork_group members, and present-and- future progress within
the organizatiOn.-
In its:cUrrent.yersion 16..,majerJndexes..from the 500--.measure,-these five
dimensions-of:organizational functioning. .Forthe:purposes of oUr present
research, two climate indexes (Technological Readiness and Lower Level
Influence) -have been eliminated-due to unsatisfactory reliability (alpha
icoefficients displayed:in Organizations- I through-,VAsee PecOrella &
Bowers, 1976), In additiol, Organization-VI had-no .measure of oroup:proceSs
,
Since our multivariate analyses require that all sites have data for all
predictors the group process index was dropped for all 'die, organizations
in our sample. Thus, we are left with 13 key SOO indexes As measures of--
organizational functioning. Brief descriptions of-the key indexes a
---The -500-was:administered at least twice to ,the five--erganizati_ons:
-diStdssed in-this report with the-,time between survey administratiens ranging
-frOm -11 to 24-months, -Table-2 lists-the-dates_oUthe-administrations
Cronbach!s. Ceefficient:Alpha (Bohrnstedt,--1969):.an&Scot Homogeneity-
R Scott,,1960), computed to-assesS the:internal, consistency ,of_the 13
-.major SOO-indexes, were -reperted in to earlier rePorts (Petorella & Bowers,
.1976v,-1976b). fable.,3suMmarizes.the---yesnits of these:tests, in, the:five--
-_-organilations. As, the results- shoW, the.S00 indexeS- di played moderate to
iiigh.internal,consistency,* with-alphas averaging .72 to 94 And HR'S
averaging .58 to
t-:Should be-noted that-statistics on the 500's-internal-eonsistency'werecomputed-usinggroup rather than individuaTdita._ _The-data-were aggregate&beeauSe all later_analy-sesmill also be Conducted-atthe_group level. ..
prganiza tional- Clinia
15
Table 1_7.
CRITICAL INDEXES
Of T E SURVEY OF ORGANIZATION$ ,
Decision Making Practices -- the manner in whichsystem: whether they are made effectively, madebased upon all of the available information.
Communication Flow the extent to which information flows freely in alldirections (upward, downward, and-laterally)-through the organization.
Motivational Conditions -- the extent to which_conditions (people, policies,and procedures) in the organization encourage or discourage effective work.
Human Resources Primacy -- the exteht to which the climate, as reflected inthe organization's practices, is one which asSerts that people are among-the organizat4,n's most important assets.
decisionsf.are Made, n the-rieh --level
.
SU_ervisow
-_--- Supervisory-Support- -- the behavior of-a supervisor toward a subordinate'which serves:to_increase the_suLordinate.'s_feeling_ofpersonal worith.
Superviiory Work facilitation 7behavior-on-the-part. of-supervisors:vhichremoves .obstacleS Which hinder success=ful 'or positively,
. . . _ .
which provides the_means:necessary for Successful. performance.-
:Supervisory Goal Emphasis- behavior which generates enthusiasmpressure for achieving excellent_performance levels.
--
Supervisory Team Buildincr-7,-behaviorwbich-encoUrages,suberdinates to-_--develop mutually satisfying:Interpersonal -relationships.
Peer Leadership
--
Paer_Support "--.behavior.of subordinatet, directenHances-each member's feeling of-personal-worth'.
_Peer Work facilitation' - -,behavior which. remevesjob.
_d toward one another, Which
_Peer Goal Emphasis behavior on'tenthusiasm for doing a good job.
Peer Team Building: behavior of subordinates toard one another whichencourages the development of-close, cooperative working relationship's.
_ _ _
e part o
oadbloc s to doing-a -ood
subordine hich stimulates
isfaction- -- a. measure of lidneral-..satisfaction-made.-up Of itpros tapping.Satifaction with- pay, with the supervfSbr, with Co-Workers (peers)-,- with.theHorganization,- with advancement oppOrtunities,and with:the-job
-16
Tabl
-DATES OF SOO ADMINISTRAT ONS
Time 1 Time 2Number ofMonths-Between
Organization I MaY 1966 May 1967 12
Organization II
Plant 1 October 1969 October 1970 12
Plant 2 October 1969 September 1970 11
Plant 3 December 1969 January 1971 13
Plant 4 February 1970 February 1972 24
Organization III April 1968 June 1969_ 14
,J)rganization IV July 1969: June 1970 ,s11
0 ganiza ion VI April 1966 April 1967 12
1
1
Table
RELIABILITY CF SOO MEASURES:
MEAN AND RANGE OF ALPHA COEFFICIENTS
AND HOMOGENEITY RATIOS*
_.A1Obk.Coefficients .
Mean (Ratige)_
-.Homogeoeity-Ratios,:.
Mean (Range)
Arganization.l.
Organization II
.72
2_87
.51-.86)
.71-.91)
.58.
.67
.26- 85
.38-.86
,Organi2atio ...84 C 67-.94) .65 .41-.84)
.Organization -IV-- 94 .78- 94) .70 .40-.88)
0 -ganizdtion- V ;85 .72-.94) .67 - 85)
*Includes data from Waves 1 and
Measures of Performance. .
7earlier reports (Petorella & Bowers, 1976a; 1976b; Bowers & Pecorella,
1975) two levels ofOrganizational effectiveness criteria-were identified-.
Ultimate criteria those organizational outcomes pertinent to the
ganization pr6duction:qoals :and-include-variables like volume, cost,
quality, and efficiency. Penultimate criteria are intermediate organizational_
outcomes and inClude variableS like attendance human costs -and resource
t --Four organizations (II. III, IV and VI) provided a useabledevelopment.
.general-cost-reasure,referred.to-here-as total variable ex ense
.and.four MOO, proVided:useable measures,of--total-' absence (ABS).
Defiiitions of-these two measures and thenumber-of months covered-by each
are provided in Table 4. ,.
Performance:data. originally provided-by_th organizations corresponded,:7-
to-different- sizes of organizational-units-. Sorry.
.
,performance, some..departmental, and Still- others--group performance. -An early-.--
isSueTwas theappropriate level :Of aggregation of data for -analyses. relating
tne-SOO indexes to:pe. formance measures. :The.choites wer either to __ _ _
aggregate the SOO,data to match the grosseSt units for which performance data--
were available .(this would reduce the N substantially and also-,reduce the
SOO-Variance) Or to-impute. Performance data to the7grouP-Ir!vf?; his-y/061d--
introduce a large number of tied scores, reduce the potential variance in
the performance measures, and thus probably depress the corre etions between
SOO indexes and performanc measures). The decision was_made t impute per
formance data to all.work-groups included in each-cost center.
the o iginal'level -of:aggregation and the N
Table 5 indicates
before and after-imput.vion.--
.*Organilation I provided a- measure of_TVE,bilt the data were:no_ Aiseable for-reasons .disuissed by.Pecorella and_Bowers .(197.6a_
MEASURES OF PERFORMANCE
TitleTotal,lbsence_
Number of employees ahsent_in a month as percentageof total numher ofemplpyees. (High ScorePoor Performance).
Definition
Duration Nov. 1965-Nuv
Title
Definition
Duration-
%'Production.EffiCieney
Actual Manhours workedpercentage of budgeted
Ananhours.(High Score 7Performance)
Poor
dan 1969June 1970
Absence Rate
Number of mandays missedas a percentage of numberof mandays scheduln.u.
(High Score = PoorPerformance)
Sept.. 1969-May,.1970..
Ti ti
Definition
-- Duration
Title
Defipi.tion
Duration
Overtime Labor Costs
-:-Total overtime as percen,,tage of total scheduledwork dayt.-
(High Seore 7 PoorPerformance)
Jan 1968-April 1969
Variance of actual 'pro-duction costsJrom budgetedcosts as a percentage ofbudgeted costs-.(High -.Score 7 poor
Performance)
-July-1969-March 1970
Total Absence
Total days absent as percentage of total scheduled
, work days.
(High Score = Poor
Performance)
Jan, 1968-April 1969
VI Title
De fin i tion
Duration
Total Variable Expense
Largest actual expensefigure from each costcenter, encompassing allexpenses, as a percentageof the budgeted figuresfor the cost centers.(High Score 7 PoorPerformance)
I965-Aug
Total Absence
,Aumber of-employeesabsent as percentage ofthe total number of
-employees.
1968 Nov. 1965-Sept. 1956
-
20
Table
PERFORMANCE DATA - LEVEL OF PtGREGATION
AND N BEFORE AND AFTER IMPUTATION
'gani zation
Before Tmputati on
-. Level of AggregattonAfter ImpUtation
N
P1 ant 38
Department 18 71 (TVE) 118 ABS)
Department orDi vi si on 11 414
IV Department 6 124
VI Cost Centers 150 TVE 193 (TVE)95 (ABS) 131 (ABS)
25
21
Definin erformance eriods One of our preliminary analytic tasks had
been to define the si ze of performance periods, that is , the number of_ months
that a "period" could reasonably be judged to contain_for each organizati
together with internal consistency (alpha) coefficients for the multi rlionth
A nonmetric techniquetalled Smallest Space Analysis SSA) was _used to
identify the performance months to be -coMbined to -form the performance periods
The results of these _analyses have. been dfscussed in previous reports
(Pecore1)0 & 'Bowers , 1976e; 1976b) Figures 2 summarize the findin s
via- diagrams which- portray the way performance- months:- cl ustered-;-. II _the
..fi gures .performance Months-. Were -.ordered .rel ati ve to when- the SOO was fi rst
administered. Thus , the performance month oCc6rring one month .previous to
the first_ SOO _administration.was "Minus-_.one month" (7-1m), the-: one occurring_
--the same month7 as the_ survey was-- To, the .one.- occurring one -month subsequent
'snryey was- +1m, etc, Each- performance month..is represented _in the
fi gures. by --- a dot. Performance months .whi ch the SSA .analyses indicated as being
cl ose together= were circled. Performance . months were requi red to be
seauenti al in -.order 'to b-e-.c1 ustered: into a.- performance_ period. The performante
periods were-label led A through S.
Within each measure .,performance -per ods .were roughTY. _comparable across
sites .in_.-,terms of---theirstime-relatipn to the SOO -admini_stration- Performance
periods -r-anged -. from_ one to- -e even--- month§ in .-..abs 1 ute.i length. Our analyse§'=--
permitted the cal culati on f internal consls tency coeffi cients for- the per7
formance. -periods , Table -6 summarizes the :alpha c-oefficients and _homogeneity
ratios -calculated for the performance periods comprising of more than one month.
_-
Performance,Months
Figure 2
, Total Var1ble pense - Performance Periods for All SiteS
Organiz _ on II
(Plant 1) (Plant 2) (Plant 1
pnizationvI
(plant 2): (Plant-3
0 T )-,TO
+4
+5
+6
+7
+8
+9
+10
+11
+12
+1
+14
+16
+17
+18
+19
+20
+21
+22
=. +23
+24
+25
+26
+22
+28
Figure
Absenteeism Rate - Performance periods for All Sites
PerforinceMonths
0 ganization II
,(S00 ii)40IN;
Table 6-
RELIABILITY OF PERFORMANCE PERIODS:
MEAN AND RANGE OF.ALPHA COEFFICIENTS AND HOMORNEITY RATIOS ,
M an (Range Across Periods
.Organization I
Organizatthr II
Or anilation III
Organization IV
r anization VI
Mean (Range_Acr
.75 (A2='.94)
78 ' (A3-36)
9 (,38.-.69):'
(.90) '
The res lts were quite encouraging. For TVE, the alphas'averaged .89 to
.97 and the homogeneity ratios .70 to .89. For ABS, the alphas averaged
.60 to .90 and the HR's .49 to .90.
. Analysis. Pyocedures
Our research- is concerned wi h -developing-and refining a methodology--
for Human Resources Accounting of a.present value type-. _This report describes-
analyses designed-to-establish the multivar a e relationships-between
characte istics of the human organization and its-organizational- effectiveness.
As such, it describL5 the completion of Phase Tyesearch-activities.
More specifically, performance_measureS for-the included organizations
were converted to -standard_scores based on each organization's score distri-
bution -for a particular period. The separate organizational files were then
-merged into a single maSter fiLL For the analyses in relation to.total
-Variable expense, as for:tho5f1! i4 relationto -absenteeism, the total. Sample
f groups.was split into two sub-samples by randomlyassigning the _groups-
in each organization. Each sub-sample was submitted to multiple regression
procedures predicting performance from Survey _tores. The weights. derived.-.--
-from each sub-sample were then applied to the survey scores from the other sub
samplei.- the performance-scores predicted and these predictions_correlated
with the:actual scores Thls procedure, termed "double cross-validatiOn"
was performed for each perfermanceperiociand-served-as-a rigorous test of
the generalizability and Stability of the _regression equations produced .
t proVides the bWs-for value attribution activities to_be-attempted i- the-
sedond.phase.Pf the.research,
RESULTS
Two research ques ions were the main focus of analyses reported in,
this section:
(1) -How strong is the multivariate relationship bet een the human
--.organization and performance, ar,c1 how stable is it across. sub-
samples-of a given population?
What is the lag time between-human- organization characteristics-
-.and their maximum impact on the organization's performance7
Limitations jpon Likely RelationsITIL
Before examining the-actual -relationShip between the human organization
and -performance, prelimiriary analyses investigated potential.limitations which
characteristics of our data sets might have interjected into the-findings.
At_least two issues presented possible constraints -upon the -relationships
:that might be obtained.
.First, the reliability of the measures might have been su--icien ly,
low that it
inevitably
coefficient
formed-a barrier-to predictive validity. --While it- is not ---
rue that unreliability -presents a -liMit,for the-Nalidity
(Guilford-, 1956, p. .470) much of what has been said .on this
-topic comes from selection testing and seems-off-target to the present
problem A ConSt ain't in the present case.would result if internal con-
tency was highenough tebe acceptable yet far froM_extremely,high and_if..
this interna ly Consistent variance was largely-.absorbed by,common factor
27
variance with the criterion. In the present case, therefore one may
reasonably question whether the observed validity coefficients suffer from
a "ceiling" effect of limited internal- consistency in both the predictor
and criterion measures.
The second constraint had to do with differences that have been observed
in the magnitude of the zero-order survey-to-performance correlations from
one organizational data set to another. (PeCorella &.Bowers,-1976a; 1976b).
-It was our feeling that the differences were related-to -capital--versus-
labor-intebsiveness. Our expectation was .that in capital-intensive Organize-i
-tions, less performance Variance would be tied directly to human olganization
--.characteristics.
To assess the likelihood that unreliabili -y of the measures wonld act
as a constraint on the relationships, a method for estimating the-expected
maximum coefficients of survey with performance measures was employed,-
(Ohiselli & Brown, 1955). The results, presented -in Table-7, show that_the
:highest expected validity-coefficients- range from 69 to .89-for absence ano
from .80 to . 89 for TVE. These coefficients are suffitiently high .to
suggest that-no,ser 'ceiling" effect was imposed by the reliability
.coefficients for the measures.
-,Tablesi8 and-9 summarize the analyses_conducted -to -assess-the effects -
--of-capital-ve us labor-intensiveness Upon-the -±eroorder-relationships.
_I-ThrWratios,-develOPed frem figures in the 1971 Fortune. -500-liSting,--Ware.
used to estimate labor intensiveness:
net..sa)es Including serVice and rental:revenues,- in
the, number-of eMployees;
total assets less.depreciation and depletion
the number o employees;
Table 7
RELIABILITY LIMITS ON
HIGHEST EXPECTED VALIDITY COEFFICIENTS
31Wzation
N,g1lost..Expected Highst.1;xpatc:47.
:Mem 30 (osencilaliity MahAbsen(q Tle Valwat
Alpha Coefficients Cocffcts Alpha CpaffiCiehts Co3ffiint
Yan
Alpha Coefficients
-Organization
I (Polyvinyl-Chloride Plant
II-(Assembly.Line)
Vi _Paper & Cellophane Mills
III (Oil Refinery)
ZERO-ORDER CORRELATIDSIETWEEN SOO_ AND
ABSENCE IN RELATION TO LABOR .INTENSIVENESS
Ratios Estiatng Labor Intensivenesst
Sa1es/5 Employees A ets Employees *
Stockholder Equity/
# Employees***
Mean of Median
Significant
Zero-Order
Correlations
$23,220 23,969: $10,376 .45
36,554 23,599 d13,978 .5
38;169 39 698 -201100
130,774 .142,064 81 069 .17
-lot sales, including service and rent 1 revenues.
_ *.!Total.ii-seti-,le depreciation and depletion.
-***Sum-of capital stockourplus, and retained-earengs.---
4:Figures used for the labor intensiveness ratios were tiken from the 1971:Fortune,500 listing.
qrganization
'1 Assembly-Line)
VI (Paper & Cellophane Mills)
IV (Aluminum)
ii .(0i1 Refinery
*Net sales
Table 9
ZERO.ORDER CORRELATIONS BETgEN SOO_ AND
TVE IN RELATION TO LABOR INTENSIVENESS
Ratios Estimating Labor Intensivenesst
Sales # Employee-* Assetsi# Employees**
3 ,564
38,169.
32 755
130,774
,599
39,698.
142,064
including service and rental revenues.
**Total assets less depreciation and depletion.
°. Sum of capital stock, surplus, and retained earnings .
tFigures used-for the labor intensiveness ratios were taken from the 1971 Fortune 500 listing.
Stockholder Equity
# Employees**
Mean of Median
Significant
Zero-Order
Correlations
978. .45
20,100. ,
34
.27
.81-069 .16
Stockholder equity sum of capital stock surplus,-and
retained ,earnings) in relation to the naiber of employees.
These three labor-intensiveness estimates were computed for each organizatiOnin our data set. Tables-8 and 9 present the average zero-order. correlations
orsurVey with performance data for each of the 'organizations, arrayed`by` labor
intensiveness. The data show that the average correlations between human
organization and performance characteristics are higherin the_ most" labor- -
intensive organizations (q-,.45) and lower in the most capital-intensiye sit-Es_ _
(ru.16). What we must keep in mind, therefore, is th`at_-our !'co,mposite
9 =organization ,,--,proulbly like-organizations in real- life contain ;, sub-segments
.
r -whose performance is more clos-ay related to human-. argarrization 'properties-
and other subsegments where this does noehold
-Suitibilityy the- a a Set
Earlier in the report several requirements regarding the reliability-,
and validity of our measin-es were listed and the data's-satisfaction_of these-
requirements considered. Six organizational data sets were originally_
-_- - -examined (Pecorella- & Bowers, -1976a -.1976b).- we-found- that-Organization.,,
,TVE data were apparently subject to the effects 'of interplaY Of fixed-and
variable productibm.--Costs with corporation7ass4ed-production-quotas. _
While- its_ absenCe data:were- inclUded, -Organization, I:was-dropped -fr66 the--
TVE -analysis--; Organization V's performance measures had-been-intricately !=,
constructed for the-special purposes of-a development project some years ago.
-Their quality was 4uestionable -and-they produced _a relatively 1°y/frequency__
of directionally correct coefficients, Thus, Organization V was also dropped.
: 32
Therefore, reliable data reMaine'd from ten facilities, in_
-"Absence-data from _groups in
--a polivinyl chloride plant Organizati-cin
- _
--four assembly:plants (Organ4zation
--a' largeyoi 1 refinery i(Organi zati on 1,0
- -two paper and cellophane_millS' (Organization V_
Total Variable Expense data from groups in:
-two assembly plants. .(Organization II)
large oil refinery (Org'anization:III)
- -an aluminuM extrusion mill Organization IV
--three paper and cellophane mills (Organization VI),
In addition to having reliable data, the multivariate analyses to be-
cdnducted required a sub-stantial number of cases for each period. Furthermore;.
_
our ability to assess the lag time of the human organization's _impact on
performance depended upon having TVE and absence data-that extended aCrbss
several performance periods. Table 10 -reports the periods for which each
organization had performance data. The listing indicates that giata =were-
available across an extended period of time, although not all organizations'
--had _data for all- periods; --Absenceldata- were available-from- at-liasttfirie out
the _four organizations -for Periods B through'H, and from one.organization
for_Periods _ . _There- were TVE data ,for all: periods A .throilgh
however,-data were,available only from OrganizationVI for :Many, of those
periods. Therefore, in the periocis _where several .organizations providedTVE
data, it -would_ he impOrtant- td inVestigate the -n.r0Pregentat1veness" of
relationships :produced on ,Orgaai zati On- s data .
Overall, -the -data set
analyses.
appeared.suitabl e .for -Tthe:: plan mul ti vari ate
,
1.
TVE _
T-
`.
Organizatton
Organization III':i
"Organization IV4'
Organization VI,
1- ,," , otd *772:P:I. " 4,"4.,.'"Otto.-
,
, Table:10, ,
,
. .
9rganizationa1 ,Sites with:TVE and AbsenceData..,
For -Each,Performance Period1-1
11
,
_ , wfro,
ABCDE
/ /
/
1 1 .11.
ABS A B
,Organization I__
Organization II
Organization HI
,,Organization VI
-1 2
/ ,
1, 1 I / '1- "1 1-
11 J K.
_
11.
1.
J
f ^
.11
-
_
_
Multi le Reiressibn Anal ses: Wave 1 SOO
The analysis design was to split the entire array of groups into two
random sub-sample halves, perform-multiple regressions- onJeach sub-samOle _and_
then double_cross7validate-the_regressions. Our expectatimwas that-acros-§-
performance periods we would find a-pattern like the one in Figure 4.
The hypothetical relationship portrayed in Figure 4 illustrates two
types of effects of the human organization on performance: concurrent and--
predictive. In other words, characteristics of the human organization
were expected to relate to-performance at two periods_ in time. -Concurrent
relations would be folind at the same time-the characteristias., as measured
My the SOO; existed. (The _SOO has been shown to describe a period of up to
Six months prior to the surveyadministration.), Predictive relations would
appear at some future time, probably several months followingithe survey
administration.- Thepredic ive effects were expected to be:the stronger and
-_ would be evidenced-by highe multiple R's in later performance periods than in
earlier periods.
Tables 11 and 12 report the multip e-regresston and cross-validation sta-_
tisLics -for the-two random sub-samples. The 13 key SOO indexes were the predic-
tors of total Nariable expense and_absence. First of all a number of_t
sample:R's were ftdtrately-high and_statistically significant:- the coeffiCients,
for total-variable expense ranged frbm ...24 to .78 and seven out -f 18 of-Ithem_ *
Were sinifiCentbeYOnd th_ 05 leVel For,s_ix out of th_e_nine TVE -performance
perieds tested in this way, and:at-least one sub.--sample had a statistically
significant R (see Table 11)Y' The coefficients for absenCe (ABSYranged frorn
1.1rOnly:Peridds A through:rwe.-e cross7validated,because Periods LI through Sincluded data_fron'OnlY:Organization,V1. -This7Meanhat too feW cases_ were-generally _available,:for=the crossValidation protedures.
45
.
Mu I ti ple
Co rrel 'citi on
Figure 11,
F;POTHETICAL RELATIONSHIP BETWEEN
TJ. HIM ORGANIZATION AND PE FORNANCE
T4.17.0
143 .1+4
_
Table I I-.
1ULlIPLE REGRESSION AND CROSS-VALIDATION STATISTICS
.Sub-Seple 1
R
'SuhSample 2
FOR TWO TVC SUB-SAMPLES: SOO WAVE 1 I IDES AS PREDICTORS
,39 ,39 -.42
92 64 219
.38 .77 .01
.55 42 ,35
,96 63 223
.01 ' .01
Cross-Val. R's
Sub7ump1e I from sub7
sample_2 1eigibts-
Sub-sa91e 2 rffisub
samplefl weig.bts
4;
254
.01
.25
255
.22
:30 .23 ..23
p..p1 p: 6 p..01 p<, 01
.:42: .237 24 .18
p<:07 p<J1 p<.01
.38 .78 .65 .29
219 '100 24.. 24 201
Al .04 .38 .83 .23
.24 .46 .69 73. .37 -..
223 98 27' .27 208
.47 .08 .56 .41 .01
.21 .16 .06
p. 01. p< . I I _too- few, ,cases
-to crots:7
p<.33
validate .
.13 .20 .11
p<.07 p < 05
MULTIPLE REGRESSION STATISTICS FOR7THE EITIRE SAMPLE:-
WAVE,1 500 IEES AS PREDICTORS:O.tABSENCE
Ii
Entire Sample
.53 .43
.01 .01 .18 1 :01 .01 .01
Table 15
EVIDENCE OF LAG TIME
MULTIPLE REGRE SION STATIISTICS ACROSS SEVERAL TIME SPANS (WAVE 1 S
Span 1 Span2 Span3
Mos 7 & 8 sub-
Span 4 Span 5
Mos. 9 - 16 sub-Mos, 17 - 26
upan 6
MOS. 27 - 31
Span 7
Mos, 32 37 sub-
5 mos. prior 6 MOS.- subsequent sequent to sequent to subsequent tu subsequent to sequent to
to SOO Wave 1 to SOO Wave 1 SOO Wave 1 SOO Wave I SOO Wave 1
-(Flods
SOO Wave 1 5_00 Wave 1
(Periods A&B&C) (Feiii5ds D&E) Wilod F) Periods GNI) 'ARUM) TFiriods li&O&P) (griods Q&R& )
TVE
Entire Sample
Mean V
Control Org. (VI)
Mean R .34
.27
,34
.40 .44
.46
.48
48
.41 . 9
SPan q Span 1
Mos. 7 - 12 p
prior to
SOO Wave 1
Triods A&B)
6 mos. prior to
SOO Wave 1
iFrioa C)
Span 2 p n 3/4,±
Mos, 7 - 14 sub-
6 mos sub e uent sequent to SOO
to SO011ave . Wave 1
(PeMds D&E F) (Periods G&H&I&3)
ABS
Entire Sample -
Mean R 8 .43
57
the organizations inCtUding Organi,ation V -prov ded data- for Periods A to
If the R'S far Periods.J to S qere to be taken as representative of- our.
larger samp e af organizations therefore, the mean R'S -for the earlier periods
-for-Organization VI would need-to correspond_to. the mean- R's for the.total
sample for those. same _periods. The-results in Table 15_ indicate-that
Organization VI's data resembled quite closely_the_dataof-our.total sample_
and were thus likely to be representative.
As far as the data extended, the results were strikingly similar for TVE
and AbS. -1Tie-q-VE relationships would 'appear- to peakAil-Spail-5. (Mean.R..,F 48),
17726 months--following the first survey administration-, and_then. begin t
-The data- for_ABS only extended as fa-r_aj Span...3-, but were. rising
-that point."--
While:the-rise and._-fall were not as dramatic aS- Our .hypathetiCal chart
depicted them,- _they were there-and. followed d.pattern -very similar _to:the
.one hypodiesized.- -The relationships varied around a valuesdf 40 --peaking,-
-.at-a somewhat higher Value- eighteen- manths. ta two ..years,after the- Wave 1 SOO
measurement.and two an,Cone-half to . three years after the presumed_onset
of the organizational conditions measured-(i.e. from Spans 0 and. 1)..
-The coefficients-, by their. magnitude :reflect the "smoothineeffect
f-,oUr blocking of the performance measures-Jr-Ito "periods" and ''spans."--
Jhese blackings contain months_in. which the_relationships are- tilich stronger
.-40as well:6S Months in which theY are-much_weaker- or even zero.
relationships appear'ed considerably more-''even" than woUld be
trUe with finer--.slicTh.gs.
43
Multiple Regres-sion AnalYses:- ave 2 SOO
Parallel analyses were conducted to investigate-the relationship.bet een
-performance and the human organization.measures uSing Wave 2-500 data. In. . . .
most cases, the seconcladministration of the,S0.0 followed the-first. by.about
onc year. Since we expected similarity in the social SyStems characteristics
over that -year and thus in the measurements dbtained): we expected-the--
relationships-to Performance to be similar bui slightly- weaker for'early
spans of performante data (i Spans 0-3) . -The'strength- of the r6latioh.-
ships for- these:-Spans relatiVe-to-those-obtainedwith Wave-1 -SOO data woUld::
be- expected tO -decline as the correlations,between the waves.of survey.data...--
i declined..
_-On the . otherliand,-we expected the- relationships during later performance-
spans:(e.g. Spans 4--7) to-be-as.--strong as-or:stronger-than-those for-.Wave 1.
Concurrent-and Predictive effects relativeto the.--second:-survey administration
were expected tdemerge:during theilater Spans,-
TableS. 16 and 17 report the multiple'regressibn andlcross-validation
.
statistiCs for-two random-sub-saMplesAt_siii.rg.-Wave-2-of-ther---S00.
,-r
results corresponded closely to our expeCtattons. Ftrst of All.
-.severalsof-the-JVE----S-ubsample-R's were-.moderatelY--htghalthough few Hof them .
mere.statistically-significant: the coeffiCients for TVE. ranged- from-...2Tto
.73; four out of-18-were.signtficant beyond the -05 leVel-,(see-Jable 1
*As with SOO Wave 1 data only Periods A through I were cross-validatedbecause Periods J through S included data from only Organization VI. Thismeant that too few-cases were generally available for the cross-validationprocedures to be applied.
60
Table 16
MULTiPLE REGRESSION AND CROSS-VALIDATION STATISTICS
FOR TWO TVE SUB-SAMPLES: SOO WAVE 2 INDEXES AS PREDICTORS
Sub-Sample 1
Sub-Sample 2 -
R
.47 .56 .28 .37- ..72 .70
95 65 204 265 206 127 29 29 197.
.07: .06 .27. .01 53 ...16 34 43
Cross-Val, R
.34. .51 31 .25 .36 ,45 .59 ".73
93 65 204 261 203 125 27 27 195
.67 21 .10 .23 01 .01 6 41 .01
Sub-sample llfrom sub-
sample 2 weights
Sub-sample 2 from sub-
sample 1 weights
.23 .09 .18 .04 too few cases .09
5 p<.06- p.19 p.01 1)4.52 pc 05 to cross- p,20
validate
:10 .19
p.37 p<.14 p.26
.14 .12 .30 .08
p.09 pc.01- p.26
Table 17
MULTIPLE REGRESSION AND CROSS-VALIDATION STATISTICS
FOR TWO ABSENCE SUB-SAMPLES: SOO WAVE 2 INDEXES AS PREDICTORS
ub-Sample I
- .61 35 .28 35 .52
111 182 190 50 190 157 159 145 135 135
.01 .01 ..01 .96 33 09 09 01 01 _ .01
.54 .39
110 178
.01
132 132
.34
Cros -Val. R
Sub-sample 1 from sub-
sample 2 weights .22 34 .15 .05 .20 .11 .28. .42 .23
p<.01 p< 01 p.31 p4 50 p.01 p.16 p.01 p.01 p4-.01
Sub-sariple 2 from su
sample 1 weights-
p.25
The tdefficients for ABS were better.. They ranged from .28 to .66
with 13 out of 20 of the correlations significant beyond the .05 level.
For rine out of ten ABS perfOrmance periods at least one sub-sample
R was statistiCally significant e Table 17).
The cross-Validation R)s for TVE-were only Significant (p<;10) in
Periods B, F, and I. Again, the ABS results were stronger. T4ectOss-
validation R's were significant for all but one ABS period- (Period G).
These results suggested that the Waye 2 SOO measurements were,related.to
early perfonmance periods but that the results for TVE were weaker than those
for ABS and weaker than the results obtained for SOO Wave 1 data. The
differences related to the SOO waves were expected. The findings regarding
TVE versus ABS were also not surprising. The stronger, more 'consistent
relationships between-the S011 and ABS were noted in an earlier report
(Pecorella & Bowers, 1976a) and were also found with Wave 1 data in the
previous section. The stronger relationships were never more striking than in
the present findings, however. These findings seem to support the notion that
penultimate criteria, such as absenteeism, are more likely to remain in close
contact with and responsive to aspects of human organization functionin
than are ultimate criteria such as cost performance.
Next, each of the performance measures were submitted to similar
analyses using the entire array of wave 2 data. Table 18 shows the results
for TVE. The R's ranged from .18 to .62; three out of the 19 multiple
correlationsmeresignificant (p.05). Table.19 shows the results forABS.
The Vs. for:_ABS .ranged_from 28 io-,59 Wiih-90% of the toeffitients:
.significantbeyond the. .05.1evel. .
Table 18
MULTIPLE REGRESSION STATISTICS FOR THE ENTIRE SAMPLE:
WAVE 2 SOO INDEXES AS PREDICTORS OF TVE
426 .34 .23 .27
188 130 408 526
'.44 31 06 .01
...21 .37 ..55 ,.46
409 252 56 56 _92'
.-18 .01 .19 57 .43
.62
56 159 87 87
.04 .11 .10 .30
e
,c1.-..4.-1,-,1_11,fik,,, . -, ,_:.-,-,,L-2-,' 7 42.'; -....L, ,- -;,4-, ,:,....",,,...1; :4'..;,,r4 ' '.., t z, ' _,.-,---,,,-7.74..,; &,,.-.. .1441.-414 4/_.....,,.../L' ..4',......g."4=-Ye;u3-1 -V: -,.-4,...-.4.74,041-07,-A=vd,=_-1-_-..yreL.R.,3g1-421rtMtryfrpgioraitinfipp2
r* I:. '_°,7.7e.,,,,-.;,.-,. 1 &-,- 4, , i -,,, 1-t -r,,:,#-# ty:'-#,I1.4..-/ ',"-,1i4.7.",
__ .:,,,r. t-4, A.A,".7:-,..-lv,9-7..-,).-t-.7m1;tv.F4-!41.1.1,7#..171;:*# 'r,.;#4,-='`-r. ----A.",:gitt#4,_AE*4_1,47,,,_:#4.! iiel'tt '- A 7. , .,-_, - ,... 1 . ., A _ ., g %., S 4,74 I gg, I "AL; t A,..11,,-2, ;,... ;, _7: ,441 .±-.",,r,T.,4 ',F. AY `-r:-,,:, ca',I,r, 1 ", :f. #.., ,,,,,,r.t7
- ,j, 7,i, i , 7 ,... ; r -- 7, - r r- , ' '..- r,...;'!. -'-': ^r-'1, ,,"-- ' :-" ,
r ' ; ' ' ' ' '' . 1 ' .1,1.77,, 4 A ,-.-_-,e ,, -, -,,,, ,' , -,. 0(4, ,-- ;,,,f JP,.- 1 4 '''t-''''
'', . ,
' -' . . '7 ' --,--##-...." '" ' *: ' ' '..''' --' -'1'. -- '' ' -.' L ' ` '' '- - ' ' r## ' ' . :,-. ----7 .5 _,-, -, .;-: = .._ - ,--.. ..,.- 7-- ,.;, ,.- .:-.-L,-..:, ,.- ,, , , -- -7. 7." . #7" J'' 0 ' ''''-', ..., "ir,'''.7:-....., , r,-<7..;,7,-;_-..3,..e.,4,---;441,r,,,r,-F,:y,737,7,,,-tt,,44,_ ,
-
,
' .--., -; -, ---`7, . e-,....
, , " - . , -' " ''. ' 7. .i -, , ,r1 q - "-, - -.' -,. i ,1 , I, . .. ,,,1 ' ,-.,- - sr,_ 1 ..., , ...A- Ill , r r . ' t
, . , ' ..-- ' , ' e- , J '1 .1', . .rrjr'1 r .f , A ..1 ,,,,4,,,,,,
' 7 ',- : 1, ' , ,, ,. .,1
- 1.' , 1-- ' -., 7 1. ''',., r,r', ' ''' ' , 1 ' - ,
' I 1. ,- ..,._
- r -' ' 7 , ', 7 , = , -; ,-' TT.- ; '7,4',',',,:. -, ' -_ , 1, -, , '---"-,--r-rf
t.1#-A
,
,
.""*
r
r.
I- -
--fAtp,,,,t " - - , . _
-: - ; - ; ' -
§- ' -
1.=7.ti f
r,irrrtryqr,`,..:r f
,
'
r 7 ; t'f 7,
A _ t A 1-"
Tab109,- CI, E'r I
-,, .
REGI1ES'SION STAiISTICS FOR-THE L'ENTIRESAMPLE::1-,
r
2r.e
, 11
-;:tx,117 7'1
'14'V- =;"
-
_ _ " -_ ..,!I L'T --", ,-.., ,.f, ..-2_,.--`"-:-- , '''' -'. '' -''' .' ,' '
WAVE 2 S00"INDEXES AS PREACTOBS'pF ABSENCE,_ _ ._ r 1 { ,
H';
-, ,7'' --- -,''' t -1- 1.'.-.' '4t'-r b i ;e 1.
2 '" ;-t....1t.!QA;T r
` -, t ;, ',-, 7 ; ,.. -IL
;.,..., . .
Entire Sample,
-=
'
^
4-7-i
, r
V
,ABS.PenodsA , ,r1 =0-, 71-* ## r "
A , - A A, er.,A1Arl
,-E, G '- ;-
1 ' J:4_ _ , "
n
21
59 alL:139,---,4 2_
I
221- 360 309 314 286 ;267" 7,--'-267---Jc°,,A7# - _
y 1 1 r ASA rf-44-
101"'"7=';01' -7., 01 ."
I'UP...,
,
7
"
J ,;
' -
_
t , "Xi
The periods were then blocked'into the saMe multi-period spans-as were
used for wave 1 data. Table 20 shows the mean multiple R for each.of the_
seven TVE Spans and the four ABS _Spans. this case Spans 3 and 4 would
contain concurrent effects and Spans 5, 5, and 7 predictive effects.
Relationships in Spans 0 through 2 would represent "shadow" effects -- that
_
is, relationships resulting from the-carryover ofsocial system properties
from one year to the next.
Once again, the mean'R's for Organization VI alone were also-computed
for-the 1VE periqds,ind-they sered as an indicator of the validity of_
coefficients in Periods J-S for which only Organization Vf provided data.
_ The results in Table.2 -indicate:that:Organization VI,'s data were quite close
to the data for-the total wave 2_sample.
___ For TVE there were sf4ris'of both concurrent and predictive. effects.. .
IThe_multiple R's were in the 20's-Auring Spans -1 and-,2j"shadOw",periods_
in the 30's during'Spans-3 and "concurrent" porio4s and in the 40's
during Spans 5-7 ("predictive" periods). The predictive effects were again
the strongest. it is likely, however,. that our data did not extend far
enough to pick_up the peak relationship.to Wave 2. surVey meaSures'.-
fl-
5.1
-r
^
-1'2,1
,
,
PE-
+1
Enti re- Sample
Mean R
CaltcoUrg,i110
_Mean R
2
Table
7EVIDOCE OF LAG.TIME:
EIKE REGRESSION STATISTICS.ACROSS SEVERAL TIME SPANS WAVE 2 s
_
L
7
,
' 1 c
Scan 1, Span 2-
Mos. 17 21
prior to
SOO Wave 2
Tiods A&B&C)
Mos. 11 16
prior toSO.3.Wave 2
Wriods BE)
a+,+=144.
-Span-3 Span 4 : Span '5 Span. 6 Span 7
. Mos, 9 & 10 Mos. suV Mos. 11._. 15 Mos. 16 = 21 ,
prior to 8 Mos. prior . sequent, to - sosequent to, subsequent to
SGO..Wai/e'2 to'SOO.Wave 2.-.. : -SOO Wave-2 SOO Wave 2 SOO Wave 2-,
Tgriod E) (PeirEds G&H&I) J&K&L&M) . (Piriods N&O&P)
-
.28 ,39 .44 .40 .45;
.44 ,40 A5
0 . Span 1
MO, 20 25. mos. 11 19
pr'.or.to prjor to
wavp 7 0 Wave 2
PTriods AB) : PTriod G)
ntire Sdolpi.e
R
Span 2
Mos; .8. -_13
priorjo
SOO'Wave 2
Triods O&E&F)
I JJ
Span 3/4
I Mori, prior to
inc1 odirg .mon th
of SOO'Wave 2
(Periods G&I-(1&,1)
.7 . . . ....... .
Com arisons Between Two Waves of Sur_y&_Datà
Figures 5 and 6 portray-the relationship between the Survey o_,
Organizations indexes and performance (i.e., TIE and ABS) across time._
The R's obtained for- both waves,1 and 2 of the survey_data are plotted.
The.similarity in shape of the two curves (Wave 1-500 versus performance
-Tand Wave 2SOO versus_those_same -performance period_scores)_-s-uggests-that- -
the social:systems in place at these two points in time were-themselvesAuite
similarc -Despite the fact that a_full yea-r_had jntervened betWeen thetwo
benchmark points n most instances-a year-of_some form of intervention-
activity), comparative stability, not radical change, seems to have occurred.
Much of what we see at the time of wave 2 represents the persistence in
time of a-set of:conditions and properties which existed at the time of
wave-1. This is borne out jn the pattern of inter-Wave_correlations
preSented in Table 21. Here we see the maximuM possible coefficients-
(the squareroot of the ceoss product of the_two alpha_coefficients
compared to_-the_actual inter-wave coefficients. -As the data indicate,:
--amoun -inter-wave correlation is substantiali-althoughit-_does-not
saturate all available variance. There is room-for movement to-evfdence
itself, but there is as well a high degree of inter-wave similarity_ _ _ _
The problems involved in interpreting lagged effects with multiple waves
of discrete predictor variables are illustrated by the graphs in Figure 7.
In section (a) of thiS:figure, we see what our expectation would be for a
behavior segment one month long and a lag of six months in total cycle time.
Mulliple R'
.50
40
.35
.30
.25
Figure 5
. Relationships Between SOO and IVE Across- Time:
Comparisons Betqpn SOO Waves'l and 2
Ia
04.0
Wave 2 SOO
.
Wave 1 SW
44
Sparc) Span 2 Span 3 Spati 4 Span 5 Span 6 Span 7
Wave 1
SOO
Wave 2
SUU
A
Fi gure 5
Relationships Between SOO and TVE Across Time:
Comparisons Between SOO Waves 1 and 2
. 50
rr_
Index-
Decision Making Prictices
-Commun-i cat ion Fl
Motivational Conditions'
Human Resources Primacy
Supervisory_Support
m 507 Work Grou'ps)
(AcrosS Organizations).
Mc,an I_U211a _Ccieffici ents
Wave 1 Wave 2
.86
_
.77 .82
.74
-.86 .88
Maximum Possible
Inter-Wave Coefficient
.83_
.77
97.=-..^,----
87
.81
88
_
Supervisory. Goal Enphasis .78, ., . _
Supervisory Work Facilitation_
"Suoervisory Team builrling .86
Peer Support .86
Peer Goal Enpnasis .80
Peer Work Facilitation _.87
Peer. Tea:r:Buil d ing .86
Satisfaction 78,
_
.93
.88
.91
,-tt=
:Actual
Inter-Wave Coefficient
'
, ,
.61 _
.
.57'-
.62
.71
;
. 51' 7 L71=
.51
734
,O
7-
s Assa,vs 1'4
- -
,
-1sc,.- ;;
5
Figure 7
Interpreting Laggi,i.d Effects with Multiple WaveS of Prqictor :leasurements: A IlypotheticAl Case
,60
3phavinr.
2 3 4 5 6
Months
Onelonth.diiration', single *le (a)-
1 2 3
Behavior 1
6 7 8 9 10 11 .12
Months .
Behavior 2
Err,
Onmonth duration, two single cycles (behaviors unrelated toeach other) (b)_
.;
.0
1 2 3 -6- 7
Montbs
Behaviorl. Behavior
Oneinonth
_
duration, two single cycles-(behaviorsnot identicaq (c)
t', t4
1
:4
I., Wa v e 2r
t
!lave 1
9 10 11 12
Figure 8
Interpreting Lagged Effects Using Multiple Waves of Measurement
A Hypothetical Case of-Reclprocal Causation
Months
Behay'or
One7ronth duration twcy:s.ingle cycles, behaviors -similar but not identical,
Witl.1-:.sorr(2 amount of:.!'raérse causation,
than the curve for that later -segment's own contemporary cycle we say _that
organizational practices are "caused" by= performance in the preceding peri od.
When the reverse occurs we .say that behavior -auses" performance. Irt-
-.instances such as that .diagrammed in Figure- -13i in which-both effects .are
apparent, we -term it "reciprocal causatim " The 'relationShips obtained-
AbsenCe seem to fit thiS latter pattern.
TwO Remaining Issues
As- final footnotesAO a main body of:findings; -it seemed:appropriate-to
examine rather specifically results _concerned with .two issues: what the
.multiple regression results:_are at.the monthly:leyel as opposed to
aggregated -"periods" of_months) and (b) whether there is reaSonabi 0-- if kel ihoed
. of curvilinearity -in: rel lopShi0S- WhiCh- We- haVe treated thus far as linear.-
:.-Reqresoion8 -by Alon h. -The:multipIe regress ons predicting total-.
variable expense and absenteeisM Monthly' ScoreS (again standardized wi
Sites) from -SOO 'indexes were-repeated for-Waves 1 and-2 _of survey data.
The results (presented in Tables 22, 24, and 25):confirm our expectationt._
The coefficients; While statiStiCally -significantWitha freqUency,for
exCeeding chanCe (40:4 67 peT6ent significant beyond the. .05.- level , for
example), are generally . semeWhat loWer than 'those predi ct i np; performance. .
scores for mul ti -month peri ods Thi s is -parti cularly_ true in the 1 ater
months ,. where -.cases _become: .fewer.ahd. where pool ing. months into periods-adds
reliability to the -performance -measures . There are, however', '.Occasi onal
coefficients which attain very high- values, again -as we expected:7U. appears,
therefore, that we can safely disregard monthly performance measures, since
-atialyses:at :thi s. level- appear to -prOvi de .us: with .little that is-not ...obtained
with gr,ate con idence at a more aggregated level .
Table 22
MULTIPLE REGRESSION STATISTICS FOR THE ENT RE SAMPLE:
WAVE 1 SOO INDEXES AS PREDICTORS OF TVE BY MONTH
TVE MONTHS
Entire Sample
6
R 1 .81 .81 .82 .41 .2 9 .35
N 61 61 61 61 127 127 381 442
.01 .01 .01 .01 .06 .69 .01 .01
13 14 15 16 17 18 19 20
27 .29
N 509 509
.01 .01
26 27
.32 .21
156 156
.28 91
.2
513 192
01 01
28 29
.22' 2
156 156
.88 26
.33 23 .24 2
192 415 409 409
.01 -,06 .04 ,01
30 31 32 33
.38 .31 .32 .21
149 119 119 113
.06 .61 53 98
10 11 12
.42 32 .44 .34 .27
381 509 244 251 509
.01 .01 01 .01 01
21 22 23 24 25
.27 .40 .22 . 9 .34
409 161 161 156 156
01 .01 .86 .03 .14
34 35 36 37
33 .34 36 .2
113 113 114 119
.70 44 95
MULTI.PLE REGRESSION STATISTICS .POR THE ENTIRE SAMPLE:
WAVE I SOO INDEXES AS PREDICTORS ,OF ABS MY MONTH
En ti re ample
_
Table 24
MULTIPLE REGRESSION STATISTICS FOR THE ENTIRE SAMPLE.:
WAVE 2 500 INDEXES AS PREDICTORS OF TVE BY MONTH
TVE MONTHS
4 5
Enti re Sample
. 62 .52 .56 ,52 .4
N 57 57 57 57 131 131
P .04 .28 .15 30 03 .56
.34 ..21 0 1 ,24
351 f108. 351 526 294 300 526
.'01 01- .01 03 .02 01 01
13 14 15 16 17 1 19
8
20 21 22 23 24 25
.23 .20 .22
N 527 527 542
01 .06 .02
26 27 28 29 30 31 32 33 34 35 36 37
.31 .30 .16 .18 18 .22 40 35 37 35
191 192 399 392 392 391 165 '_165 158 158
.13 .04 .63 .43 .42. 13 .01 .07 ;05 11
.32 .31 .28 436 .25 34 .35 432. .38 .40 4O
N 159 159 158 158 151 12.2 122 117 117 117 120 122
24 .2 .48 .07' .76 37 .30 54 .20 .12 81 11
8990
Table 25
MULTIPLE REGRESSION STATISTICS FOR THE ENTIRE-SAMPLE:
WfkVE 2 SOO INDEXES AS PREDICTORS OF ABS BY MONTH'
ABS M NTHS
5 6 10 1
Entire Sample
76 .33 35 .41 .34 4 3 .21 1
24 147 147 375 375 432 432 211 211
.47 .23 .17 .01 . .01 .01 .01 .76 .08
13 14 15 16 17 18 19 20 21
.30 .40 4 .40 .26 .49 .35 .69 .75_
314 81 24 274 267 267 Z5.7 41 41
.01 46 15 01 15 01 .0.! 08 01
.36
317
.01
22
=1,==!=1,717.==,
21 .27
309 424
.43 .01
23 24 25
.50 ,
41 41 41
76 33 11
92
63
Curvibi earity It is at least possible-that the-multiple regression-
coe_ficients which-we have obtained Understate the real relationships.which
exist-between SOO indexes_and performance beCause those true relationships
are in some fashion curvilinear. An accepted methodof-testing for cur-
vilinearity is to compare correlation- ratios-with product-moment coefficients::
obtained for-the -same .data set (McNemar, 1969). Multiple classification
analysis has been developed as a Multivariate teehnique analogous to, multiple-
regreSsion, yielding as well partial correlation ratios (Andrews, et al.,
1973). In the pretent instance, one would expect curvilinearity to evidence'
-itself-in Ihe :form 1-F larger multiple coefficients from multiple classification--
.analysis than from multiple regreSsion. To test this three periods -of data
for_TVE and for ABS were submitted,to.multiple classification analysis, uSing
all 13 SOO ind6(es. as-predictors. The-results are presented-in Table 26-,_ .which compares.-multiple p-ediction_coefficients .from.the -two procedures;
As thesV-find ng-S- indicate, there is little evidence of substantial curvi--
linearity present i- the relationships.
64
Table 20
MULTIPLE COEFF CIENTS FROM MULTIPLE CLASSIFICATION ANALYSIS
AND MULTIPLE REGRESSION PROCEDURES, FOR SELECT PERIODS
Goefficients
Multiple.cation
Analysis -- -Regression
65
DISCUSSION
The two general- research questions-posed.at .the outset of the:Results--
section appear to- have_been answered-rather-,conclusively. Multivariate-
relationships of respectable magnitude do-occur, and.they.do pross7-validate.
An estimate of the lag tiffe involyed for organizationsAf the type"focused
upon in .the present study is-at least approximated. Furthermore-, cortain-.
possible concerns seem to have.. been-unwarranted. Unreliability in .both
predictors and criteria does-not:appear td.present.a serious limitation:
Hinternal-consistency_reliability_coefficients-for both- types ofAleasures
-are quite high-, and the-multiple-cot. elation,coefficients between
_not-appear-to encounter-an upper ".barrier..1-1
Second -the possibility that relationships might not occur in some sites
which nevertheless had reliable survey-and performance=veasures was not
realized. Instead, we find that the magnitude of relationships_between
survey and performance masures appears to be constrained by the eXtent t
which the organization is capital intensive.
Third, there is no e idence of significant curvilineariv present in
the relationships between survey predictors and-prformance criteria .
coefficients generatediby a non-linear procedure appear-to be'
almost identicalto those-produced by a linear method.-
,Finally,,00llapsing,performance in_o multi-month .periods-does not appear
to have done drastic damage to the relationships. Indeed, it appears to
have improved the reliability Of our predictions.
9 5
While these_concerns appear to be no longer justified, therefore,
there are other factors which do appear to have reduced the magnitude:of
the obtained-coefficients by removing portions of relevant- variance. One
of these is the imputation process, by which we assigned the performance
scores for a cost center to all of the non-supervisory work groups that
comprise it While in reality the various groups in-a-particular cost
center no doubt contributed differentially to its measured effectiveness,
the_measuring_system_does_not_record-their-differences,--This artificially'
-increases the number of tied performance scores,. thereby reducing variance
in.- the criterion measures. For this reason-,- the- multiple cerrelation
coefficients understate by 'an-unknown amount-the true-relationships whirJ'L
.- exist- between a Work group's human organizational conditiOns and its
performance.
Another factor potentially -reducing our obtained-relationships
standardization process whereby each-work group's performance measure-was.
converted to a standard score in its own- distribu ion for that particular --
period. Not only does this pro-edure remove real variance .that in theory
exists-between or anizations-and which would perhaps enhance our obtained-..
coefficients (there are differences in human organizational characteristics-
among the firms which produce correlated. differences..in. perforMance'-- twt :the
latter is. reMoved) it alsci removes real-variance-among cost-centers-. acro.i:s_.-.
1)eriod
-For all of these reasons, therefore we must-keep in mind that the
Multiple correlation coefficients obtained in the-present study- understate
the true values that- exist- and 'represent a- conservative estimate of their.
rength.
67
this context, the pattern of obtained relationships must be
regarded as particularly reassuring. Statistically significant multiple
correlation coefficients are obtained in proportions far outweighing
chance. Using Wave 1 survey data to predict total variable expense
coefficients were-obtained which range from .27 t .70. Similar predictions
to absenteeism rate yielded a range of coefficients from .20 to .53. For
both measures, predictions using Wave 2 survey data produced ranges varying
only slightly-from these values.
Lag time estimates contain elements that bo h confirm and expand our
expectations. Whilc the rise and fall in obtained relationships anct as
dramatic as our hypothetical chart might have depicted them, they are there.
Peak relationships appear to occur 17 to 26 months after SOO Wave 1 and
tAlo and one-half to three years after the presumed onset of the conditions
measured by that Wave. The results were strikingly similar for absenteeism
rate. In the case of the latter measure (absenteeism ), though not for total
variable expense, there was evidence to-suggest somp_amount of :reciprocal
Causation," that is, improved organizational practices as a response to an
earlier high absenteeism rate.
An ancillary finding is that social .Sy$tem_pns ancy, rather :than .change,
appears to exist. SoCial'system similarity between the two-waves of survey
data was -quite streng despite rather substantial effOrts-which.in each.
.instance coincidentally went on-to attempt to improve:those. systems;.
Pulling -these various findings together, it would appear- that-fiVe
concepts are required to explain the data. Fir-st, there are concurrent
effects-,..significant relationships--to performance-whose time period was
mere or less ..conteraporapeoUs to-the organizational conditions-measured-by.. a.--
particu ar survey wave. Second, there are predictive effec significant
68
-.relationships tu performance in time periods subsequent to the-organizational
conditions measured by a particular survey wave.and whose occurrence
flects the fact of lag time Third there is- the shadow effect, the
occurrence_of simdlarly shaped curves for adjacent survey waves, defining
their relationships to the same performance periods and attributable to
the apparent tendency of social systems to_ remain rather invariant aCross time.
Fourth, there is What we have, termed reciprocal causation- for which evidence
in the present study occurred for the absenteeism measure and Which iriall
likelihood occurs for other outcomes-as well. in.addition-to the postulated
main effects of organizational practice§ causing performance, there is a
normal responsiveness of the social system to-earlier performance particularly-_
toldepressed performance). Fifth, there is outcomp_tloseness, versus rembteness,.
reflecting the place- of the various measures-in an events seq6ence 9 ganiza.'-
tional practices versu .outcomes, penultimate and ultimate).-Finally, it seems appropriate to comment-on the analysis itself. The
double cross-validation design is-, we feel, particularly r_gorous. It. ..r.,....___
helps to assure that the results would generalize to other, stmilar,.
settings and that the findings_ do not simply capitalize upon characterisits
of a_rparticular sample
.In this connection, it should be noted that, while the organization--
included in these analyses _.ci not cover the entire spectrum of American' work
--life and are civilian rather than military, they do resemble the Navy in
many ways:
(1) in varying degr es they are- la ge, comple- and oriented
around expensive .hardware;
the work is, except in administra ive sectors, hot, heavy,
, and-dirty;
9
69
each is a part of a larger entity which depends upo- it
in some measu. e for its performance.
The shortcomi gs of the present analyses would appear to-center around
the absolute magnitude- of the obtained-coefficients. They would appear to-.
explain no.more than 25 percent of the variance in performance among cost
centers. .0f course, perspectives on the meaning of this percentage may
vary: it may be seen as "only 25 percent;" thd.other,hand, to be able
to explain (and presumably affett ) 25percent of performance- variance is
no mean feat.
Still the percentage requires explanation. While-the theory 'From which
we work seems .at least acceptable comprehensive, it is Obviolis that a large
portion Of performance variance remains tobe'!explained. Obviously-, not all
possible predictors are included in the present array, andthe addition of
other variables might improve our ability to predict-.
Despite -this obvious possibility, it is worth reiterating t. he fact that
-several facets of our- procedure deliberately removed-or excluded potentially
relevant criterion variance. There is the very real possibility -- indeed
the likelihood -- that__a much higher portion of performance Variance would
.be -accounted for were those additional portions included-in our criterion.
measqre
On the..basis of the findings,.therefore, we feel tha-t _the basic
requirements --for- constructing luture_performa -e_trend indicators -- a
current value- approach to human-resourtes accoUnting have been met:
(I Key dinensions of the human organization have been
identified and accurate measurements thereof obi:lined.
(2 ) Reliable, valid indicator _of organizational effectiveness
have been obtained and refined.
70
(3_ Relationships.between key dimensions of the human
organization and performance have been established.
At least ancillary evidente supporting the durability.
of changes in organiz.ational functioning has been obtained.
System stabil ty, not erratic fluctuation, seems to be
the..rule.
Accordingly, the research effort will --urn toward two lines of
-neteSsary:extension-:-
The analyses just reported will be replicated as far as
possible; using Navy survey and-performance da
in-fiand, from earlier studies).
already
For the preSent civilian _data sets-, a start-wi l be made in
the value attribution phase. This will invOlve converting .
inter-wave.survey changes (modest though they may be) into
changes in dollar-value of future performance. Capitalized
and discountedthese changes then will represent gains and
losses in the current value of the human organization... .
.
71
_REFERENCES
Andrews, F.M., Morgan, J.N., Sonquist, J.A., & Klem, L. Multi lec_l_alLificat:yzAnalsis. Ann Arbor, Michigan: Institute for SocialResearch, University of Michigan, 1973.
Bohrnstedt, G.W. A quick method for determining the reliability andvalidity of multiple-item scales. American SociololijaLl_gyil1969, 34, 542-548.
Borman, W.C. & Dunnette, M.D. Selection Of components to comprise a NavalPersonnel Status Index (NPSI) and a s nategy for investigating theirrelative importance. Report to the Office of Naval-Research, 1974.
Bowers, D.G. SyLt_Ems of o anization: Mana'ernent of the human resource.Ann Arbor, Michigan: The University of Michigan Press, 1976.
Bowers, D.G. Navy manpower: Value ractices, and human resourcisre_irements . Ann Arbor, Mic igen: Institute for Social Research,1975a.
Bowe D.C. Work values and preferences of officers a d enlisteds in theU.S. Army. Report to the U.S. Army Research lnsti u e for theBehavioral and Social Sciences, 1975b.
Bowers, D.G. 0.0 . techniques and their results in 23 organizations: TheMhclugan ICL Study. The Journal of A lied Behavioral Science, 1973,9, (1
ers, D.G. Development techniques and organizational change An overviewof results from the Michigan Inter-Company Longitudinal Study. AnnArbor, Michigan: Institute for Social Research, 1971.
Bowers, D.G. & Franklin, J.L. Survey-guided development: Data basedorganizational change. Ann Arbor, Michigan: Institute for SocialResearch, 1976.
Bowers, D.G. & Pecorella,-P.A. A current value approach to human resou cesaccounting. AccountialEE, 1975, 45 (2), 25-40.
Brogden, H.E. & Taylor, E.K. The dollar criterion: Applying the costaccounting concept to criterion construction. Personnel1950, 3, 133-154.
Brummet, R.L., Pyle, W.C., & Flamhol z, E.G. Accounting for human resources.Michigan Business Review, March 1968, 20-25.
Caplan, E.H. & Landekich, S. Human resource accounting: Past, present andfuture. National Association of Accountants, New York, New York, 1974..
-Dr xler, J.A. Jr., & Bowers, D.G. Navy retention-rates and human resourcesmanagement. Institute for Social esearch, May, 1973.
10 1
72
Drexler, J.A., Jr. & Franklin, J.L. Relations between social psychologicafactors and performance criteria measures in multi-organization research:Key methodolooical issues. Report to Navy Personnel Research andDevelopment Center, San Diego, California, 1976.
Dunnette, M D., Milkovich, G.T., & Motowidlo, S.J. Possib e approachefor development of a Naval Personnei Status Index (NPS1). Report to
the Office of Ndval Research, July, 1973.
Flamholtz, E.G. The theory and uasureTnt of an individual's value toan organization. Doctoral disser ation, the University of Michigan,Ann Arbor, Michigan: University Microfilms, 1969, No. 70-14, 519.
Franklin, J.L. Down the organization: Influence processes across levelsof hierarchy.- Administrative ifps.-1A.111,_,. 1975a, 20, 153-164.
Franklin, J.L. Relations among four social-pSYChefogicar-asPeCts oforganizations.- Administrative Sin_c_t_P_OrterlY, 1975b-, 20.,- 420-433.
Franklin, J.L. & Drexler, J.A. Jr.- Influences.of organizational conditionsand practices on reenlistment, Operational readiness,-..and satisfactionin the Navy. Report to Navy Personnel Research and Development Center,San Diego, California, July, 1976.
Ghiselli, E.E. & Brown, C.W.New York: '_McGraw-Hill, 1955.
-Guilford, J .P Fumlamental statistics ifLEEd.tica:ti on. New York :
McGraw-Hill, 1956.
Hermanson, R.H. Accounting_fornuman assets. EaSt Lansing, Michigan:
Bureau of Business-and Economic-Research, 1964.
Likert R. TILejlElp_2trii_ration. New York.. McGraw-Hill, 1967.
Likert, R. NeIALpitIerns-ofmanaggffent. New York: McGraw-Hill, 1961.
Likert,--R.-kBowers, D.G. improving the -accuracy of P/L reports byestimating the change in dollar value .of the human organization,cli,i-LaDIJATian_fttnDi, March 1973,.15-24.
Likert, R. & Bowers,- D.G. -Organizational .theory_and human resource -.accounting.,
1969, 24 (0,-585-592..
Likert, R., Bowers, &-Norman, R.M. Howto increase a firm's lead tiin recognizing and dealing with- problems.-ofmanaging its-humanor anization,. Michigan Business-_Review, January 1969, 12-17:
McNemar, Q. Ps cholo ical statistics. New York: John Wiley & Sons,1969 (4th edition
Pecorella, P.A. & Bowers, D.G. Future performance trend indicators: A
current value approach to human resources accounting. Technical Reportthe Office of Naval Research, September, 1976a.
Pecorella, P.A. & Bowers, D.G. Future performance trend indicators:current value-approach to human resources accountingTechnical Report to the Office of Naval Research, November, 1976b.
73
Pyle, W.C. Human resources accounting Financial Anal sts Journal,Sept.-Oct. 1970a, 69-78.
Pyle, W.C. Monitoring Lumen resources--on line. Michi an Business Review,1970b, 22 (4), 19-32.
Scott, W.A. Measures of test homogeneity. Educationll and PsycholpgicalMeasurement, 1960, 20, 751-757.
Seashore, S.E. Criteria of organizational effectiveness. tagiLaulBusiness Review, 1965, 17 (4), 26-30.
Seashore, S.E. & Bowers, D.G. Durability of organizational change.American Psychologist, 1970, 25 (3), 227-233.
Taylor, J.C. & Bowers, D.G. Survey of 0r9anizations. Ann Arbor, Michigan:Institute for Social Research, 1972.
74
Distributionjist
Navi
4 Dr. Marshall J. Farr, DirectorPersonnel and Training Research
ProgramsOffice of Naval Research(code 458)Arlington, VA 22217
ONR Branch Office495 Summer S-LBoston, Mass. 02210ATM: Dr. James Lester
1 ONR Branch Office1030 E. Green St.PasadPna, CA 91101
ATTN: Dr. Eugene Gloye
1 Dr. M.A. Bertin,Scientific Director.Office of Naval ResearchScientific Liaison Group TokyoAmerican EmbassyAPO.San Francisco 96503
ONR Branch Office536 Clark St.Chicago, IL 60605ATTN: Dr. Charles E. DaVis
Office of Naval ResearchCode 102Arlington, VA 22217
Office of Naval ResearchCode 200Arlington, VA 22217
-Office of Naval ResearchCode 201Arlington VA 22217
Office of Naval ResearchCode 400Arlington, VA .22217
-Office-of Naval-llesearch.Code. 401.
Arlington,- VA 2221.7
12 Manager, Program in Man:D.,PrR&D'-
Office of Naval Research(Code 450)Arlington, VA 22217.
6 Commanding OfficerNaval Research LaboratoryCode 2627Washington, DC 20390
I IleadMaripoWer-Training andReserve Group (0P-964W;
Rocim 4A538, PentagonWashington, -DC 20350
Manpower Analysis and SyemsDevelopment Branch (Op-121)
Room 1606, Arlington AnnexWashington DC 20370
1 Human Resources Program FanagerNaval Material Comnand 0344Room 1044,- Crystal Plaza 052221-Jefferson Davis Higlway-.Arlington, VA 20360
Special Assistant for EnlistedForce-Analysis
Bureau of Naval Personnel (Pars 2xRoom 2628, Arlington AnnexWashington, D.C. 20370
Director, Human Resource ManagementNaval Amphibious SchoolNaval Amphibious Base, Litt e Creek,Norfolk, VA 23521-
Special Assistant. for-.Re7.en-ion..Analysis (Pers712)
Room 2403,.Arlington AnrexWashington-, DC 20370
1 LCDR Charles J.:Theisen, Jr., MSC, USN4024Naval-Air Development Ce7terWarminster, PA 18974
Commanding Officer ---U.S. Naval. Amphibiow S:ho 1Coronado-, CA-. 92155.-
1
1
CDR-Paul D. Nelson, MSC, USNNaval Medican R&D Command(Code 44)National NaVal Medical CenterBethesda, MD 20014.
Commanding Officer--Naval,Health Research CenSan Diego,- CA 92152Attn:....Library
75
1 Chairman, Leadership & Law Dept.Div. of Professional DevelopmentU.S. Naval AcademyAnnapolis, MD 21420
-Scientific Advisor lo the'Chiefof_Naval PerFonnel (Pars Of)
Naval Bureau ol Personnel-,
Room 4410, Arlington AnnexWashington, DC 20370
Dr. Jack-R. Borsting'Provost & Academic DeanU.S. Naval POstgraduate SchoolMonterey, CA 93940
Mr.-Maurice CallahanNODAC (Code 2).Dept. of t.he NavyBldg. 2, Washington Navy.Yard.(Anacostia)Washington, DC- 20374_
Office of Civilian PersonnelCode-342/02 WAPWashington-, DC 20390Attn: Dr. Richard J. Niehaus.
Office of CiVilianTersonnelCode. 263
Washington, DC_ 20390.
Superintendent (Code 1424)Naval Pbstgraduate-SchoolMonterey, CA 93940
Dr. H.M. West:IIIDeputy ADCNO fer Civilian Planningand Programming (Acting) -.
.
RooM:2625,-.Arlington Annex:.Washington-,.DC. 20370
1 Chief of.Naval Tecr,nici'Naval Air Station Mamp7:aMillington, TN 354Attn: Dr. Norren Ka-r
Prtncipal_Civilian--Adv'..T*frrEducation and Trainir:
Naval Training Command, -AAPensacola, FL -32508Attn: Dr. Wil iam L.
Dr: -Alfred F. .Smode, ,ctcTraining Analysis -&.Eveaticr. OrDepartmentof the. !;_avy...Orlando, FL 32813-.
Chief .of Naval.- Educatic7 and
Training -Support (01A:PenSacola,- FL-- 32509 --
Capt. H.J. Connery,'USN-Navy,Medical .R&D Commanf-NNMC, Bethesda, MD_ 200:4-
Navy:Personnel R&D_ CentrCode 01San Diego, CA 92152
5 A.A. Sjoholm, Head, Tecloical SupportNavy Personnel R&D_CentarCode 201San Oiego, CA 921 2
2 Navy Personnel R&D Centartode--310 .
San Diego, CA 92152Attn:..Dr. Martin _F. Wiskoffl
--Dr.- Robert.Morrisbn.
-NaVY PersOnnel R&D CenCode 301.
San Diego, CA- 92152
Navy Personnel R&D Cenr_rSan Diego, CA 92 52Attn:. Library
Navy Personnel R&D CenSan Diego, CA 92152Attn: Dr. J.D. Fletch:r
76
Officer-in-ChargeNavy Occupational Development &Analysis Center (NODAC)
Building 150, Washington Navy Yard(Anacostia)Washington, DC 20374
Dr.- Worth Scaniand.Chief of Naval Educa ion & TraininNAS, Pensacola- FL 32508
,3111/
1 -Technical Director_LIS...._Army.Research_institutik- for_
the- Behavioral-1 Social Sciences1300 Wilson.- Blvd.
Arlington, VA 22209
Armed Forces Staff CollegeNorfolk, VIIk_ 23511
Attn: Library
1- Commandant .
U.S.--Army Infantry SOhoolFort-Benning, GA 31905
Attn: ATSH-I-V-IT
1 CommandantU.S. Army Institute of
AdministrationAttn: EAFort Benjamin Harrison, IN 46216
Dr. Ralph DusekU.S. Army Research Institute1300 Wilson Blvd.Arlington, VA 22209
Dr. Joseph WardU.S. Army Research Institute1300 Wilson Blvd.Arlington, VA 22209
Dr. Ralph CanterU.S. Army Research Institute1300 Wilson Blvd.Arlington, VA 22209
Dr. Milton S. Katz, ChiefIndividual Training & PerformanceEvaluation Technical Area
U.S. Army Research Institute1300 Wi 1 son Bl vd.
Arlington, VA 22209
1 .0CDR, U5AADMINCEN-Bld. #1, A310Attn:- AT21-0ED LibraryFt. Benjathin Harrison,, IN- 45216
Air Force
1 Personnel Analysis DivisHQ USAF/DPXXA .
_Washington, DC. -20330
1 ReSearch BranchAFMPC/DPMYP
.Randolph .AFB,. TX 78148_
AFHRL/AS- and
--..-Wright-Patterson AFB
-Ohio -.45433 :
I or. Ross L. Morgan (AFHRLASR)--Wright--PattersOn AFB
Ohio 45433
Dr. Marty Rockway (AFHRL TT)Lowry AFB-- .
Colorado 80230
1 Dr. Alfred-R.-FreglyAFOSR/NL,- Bldg. 4410Bolling AFB, DC 20332
AFHRL/PEDStop #63Lackland AFB TX 78236
Marine Coros
1 Director, Office of Manpower
UtilizationHQ,_Marine Corps (Code MPU)
Quantito, -VA -22134
1 Dr. A.L. SlafkoskyScientific Advisor (Code RD-1)HQ, U.S. Marine CorpsWashington, DC: 20380
Other. DOD
1 br. Robert- Young
-Advanced-Research Projects Agency1400 Wilson Blvd. .
Arlington,.VA.. 22209
Mr. Frederick W. Suffa-Chief, Recruiting and RetentionEvaluation
Office of the Assistant Secretary.of Defense, M&RA .
Room 3D970, Pentagon-Washington, DC- 20301
12 -Defense Documentation CenterCameron Station, Bldg. 5Alexandria, VA 22314Attn: TC
Military Astistant for Human-Resources
Office.of the-Director Of_ DefenSeResearch & Engineering
.Room 3D129, PentagonWashington, DC , 20301
Director, Managemen_ Informa ionSystems Office
OSO,__M&RA:
Room .3B917,'.PentagonWashington, TiC 20301
Dr. William Gorham, Director-Personnel R&D 'CenterU.S. _Civil:Service CommiSsion:1900 E. St. NWWashington, DC 20415
Dr.--Joseph L. Young, Director.Memory & COgnitive-ProcessesNational-Science FoundationWashington,- DC 20550
Dr. H. Wallace Sinaiko,- Director, 1-
Manpower Research & Advisory Services.Sm'ithsonian Institution__801 N. Pitt--St.
-Alexandria, VA .22314
--Robert .W.'-Stump
Nettenal--..Instito e _of EducationWashington,-.DC-- 20208
1
Dr. Gerald V. BarrettUniversity of AkronDept. of. Psychology_Akron, OH_ _44325
Dr. Bernard M. BassUniversity of- RochesterGraduate School, of-Mane-gementRochester,- NY 14527
Dr. Philip G.--.Bernard
B-:15-825-Shady Grove _Ftd-Rockvillei-- MD 20850
. Dr.- .A. Charnes
University of TexasAu-stin--TX 78712
Dr. Kenneth E. ClarkCollege-of Arts & SciencesUniverSity of RochesterRiver campus.StationRochester, NY 14627_
Dr. John J. Collins-Essex Corporation6305 Caminito EstrelladoSan Diego, CA 92120
Dr. Joseph E. ChampouxSchool of Business & A&nnistrationUniversity of New MexicoAlbuquerque, NM 87131
Prof. W.W. CooperGraduate School of BusinessAdministration
Harvard UniversityBoston, MA -02163
pr. Robert -Dubin-
University .of CaliforniaGraduate-School -of AdMinis ration-Irvine,,CA-.. 92664
1 Dr. Marvin D.,DunnetteDept. _of Psychol ogy
University of MinnesotaMinneapol s , MN 55455
ERIC Faci 1 ity-Acqui si tions
4833 Rugby Ave.Bethesda, MD 20014
Major 1.N. Evon4._,:Canadian Forces PersonnelApplied Research Unit
1107 Avenue Rd.Toronto, Ontario, CANADA
Dr. Richard L. Ferguson'The American College Tes
ProgramP.O. Box 168Iowa City, IA 52240
78.
HumRRO/Western Di v is i on
27857 Berwick Dr.Carmel, CA 93921Attn: Dr. Robert VinOerg
Dr. Ezra S. KrendelWharton School, DH/CCUniversity of PennsylvaniaPhiladelphia, PA 19174
1 Mr. W.E. Lassi ter-
Data SolutiOnS -Corp:Suite 211, 6849:Old Dominion:Dr.McLean , 'VA-- 22101
Dr. Victor FieldsDept. of Psychology-Montgomery CollegeRockville, MD 20850
Dr. Edwin A. FleishmanAdvanced Research ResourcesOrganization
Si 1 ver Spring, MD 20910
1 Dr.. Robert Glaser CO41irectorUniversity of- PittSbUrgh3939:_0'Hara St.Pittsburgh, PA.- 15218'
1 Dr.. Gloria L.. Grace
System DevelOOMent :Corporation2500 Colorado A-Ve.Santa MoniOa-:, CA 90406-,
Dr., Richard S. HatchDecision Systenis :Assoc-. ,
5640 Nicholson L-an eRockville, MD 20852
=
Dr. Ernest J. McCormickDepartment of Psychological Science7.Purdue UniversityLafayette, IN 47907
Dr. Robert R.' MackieHuman-Factors Research, Inc.6780 Corton Dr. /Santa Barbara Res-earch ParkGoleta, CA 93017
_Dr. Leo_MundayHoughtbn Mifflin Co.P.O. Box- 1970Iowa City, IA 52240
1 Richard T. Mowday_College of Business Administration,University of Nebraska, LincolnLincoln, NE 68588 _
Human-Resources ResearchOrganization
400 Plaza Bldg.Pace Blvd., at Fairfield Dr.Pensacola, FL 32505
Dr. Lyman W. Porter, DeanGraduate School of AdministrationUniversi ty of Cal iforni aIrvine, CA 92717
Dr.:: Di ane M. RaMsey-Klee
FW He's earc : & SyStethr Desi
947 Ridgemont Dr.Malibu, CA790265.
_
R. Dir._ M. RauchP II 4Bundesministerium der VerteidigungPostfach 16153 Bonn 1, GERMANY
Dr. Joseph W. RigneyUniversity of So. CaliforniaBehavioral Technology Laboratori3717 South GrandLos Angeles, CA 90007
Dr. Andrew M. RoseAmerican Institutes for Research1055 Thomas Jefferson St., NWWashington, DC 20007
Dr. Benjamin SchneiderDept. of PsychologyUniversity of MarylandCollege Park, MD 20742
Dr. Lyle SchoenfeldtSchool of ManaementRensselaer Polytechnic InsTroy, NY 12181
Dr. David J. WeissDept. of PsychologyN660 Elliott HallUniversity of MinnesoMinneapolis, MN 5545
Dr. Keith WescourDept. of PsychologyStanford UniversityStanford, CA 94305
HumRRO/Western Division27857 Berwick Dr.Carmel, CA 93921Attn: Library