The Relationship of Group Cohesion to Group Performance - A Research Integration Attempt - Laurel W...

8/3/2019 The Relationship of Group Cohesion to Group Performance - A Research Integration Attempt - Laurel W Oliver - July

1/28

Technical Report 807 r LE: cop"0)S The Relationship of Group Cohesion to Groujp

Performance: A Research Integration Attempt0)Laurel W. Oliver

AR t Scientific Coordination Office, LondonIBasic Research Office

DTICSP 2 2'198

Research Institute for the Behavioral and Social SciencesJuly 1988

Approved for public rqelaase; distribution unlimitod.

9 06


2/28

U.S. ARMY RESEARCH INSTITUTEIFRTHE BEHAVIORAL AND SOCIAL SCIENCES

A Field Operating Agency under the Jurisdiction of theDeputy Chief of Staff for Personnel

WM. DARRYL HENDERSONEDGAR M. JOHNSON COL, INTechnical Director Commanding

Technical review byWilliam W. Haythorn

NOTICES

9 s lB ON : rimar istr' tion this port as o maby P ie ad ss: r

FINAL DISPOSITION: This report may be destroyed when it is no longer needed. Pleaise do notreturn it to the U.S. Ariny Research Institute for the Behavioral and Social Sciences.position, unless so designated by other authorized documents,

Msl


3/28

UNCASSIFIEDSECUR'TY CLASSIFICATION OP THIS~ PAGE (,Men Date Itatered)REPORT DOCUMENTATION PAGE READ INSTRUCTIONSBEFORE COMPLETING FORM1. REPORT OUMBER 2. O0VT ACCESSION NO. 3. RECIPIENT'S CATALOG NUMBSR

ARI Technical Report 8074. TITLE (end Subtitle) S. TYPE OF REPORT & PERIOD COVERED

FinalTHE RELATIONSHIP OF GROUP COHESION TO GROUP July 1986-July 1987 PERFORMANCE: A RESEARCH INTEGRATION ATTEMPT 6. PERFORMING ORG. REPORT NUMBER7. AUTHOR(,) S. CONTRACT OR GRANT NUMBER(#)

Laurel W. OliverS. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT, PROJECT, TASKU.S. Army Research Institute for the Behavioral AREA & WORK UNIT NUMBERSand Social Sciences5001 Eisenhower Avenue, Alexandria, VA 22333-5600 2QI61102B74FI1I. CONTROLLING OFFICE NAME AN D ADDRESS 12. r.EPORT DATE

July 198813. NUMBER O PAGES2814. MONITORING AGENCY NAME & ADDRESS(If dlffe,ent from Controlling Olfice) IS. SECURITY CLASS. (of this report)

Unclassified1Sa. UDECLASSIFICATiON/DOWNGRADING

SCHEDULE

I1. DISTRIBUTION STATEMENT (of this Roport)

Approved for public release; distribution unlimited.

17. DISTRIBUTION STA1'EMENT (of the abstract entered In Mloack 20, It dilferent from Report)

IS. SUPPLEMENTARY NOTES

IS. KEY WORDS (Continue an reverse side it neceewy, and Identify by block number)Research integrationMeta-analysisGroup cohesionGroup performance

20& A8SrRAfrc~tt~u* ma evere., shb #1noee~eaW sa fdenft~fp.by block niumber)-' The Army's increasing interest in group cohesion has led to increased

research eff rt in that area. This report describes an effort to integratethe cohesion~performance research that employed real world groups using ameta-analytic approach. Tukey's (1977) stem and leaf display was used todisplay the data. The median effect size (product-moment correlation coef-ficient) fo r th e 14 codable studies was .36, and the unweighted mean r was.42. When study effect sizes were weighted by th e number of (Continued)

DD JOR " 47,3 wornoN or I NOV el, is OBSOLETE UNCLASSIFIEDSECURITY CLASSIFICATION OF THIS PAGE (Wler Data Entered)i


4/28

UNCLASSIFIEDSECURITY CLASSIFICATION OF THIS PAGE(mmao Data Enterad)

AR I Technical Report 80720. Abstract (Continued)groups involved, the mean became .32. Rosenthal and Rubin's (1982) BinomialEffect Size Display (BESD) demonstrated that a correlation of .32 increasessuccess rate (high performance) from 34 percent to 66 percent when cohesionis increased from low (below median) to high (above median). Althoughthese findings are problematical because of the very small number of codablestudies, as well ap the conceptual and methodological problems associatedwith the cohesionlperformnance research, higher levels of cohesion would seemto be very desirable for real world groups such as Army units.

Aa Bass For

NTIS GRA&IOTIC 'TAB 0lUnttuUAouflQad 13

Aval lnbiiity radfa5

UINC'LASSIFIED)SECURITY CLASSIFICATION OF' THIS PAGEt'Whon Dot* Er~tford)


5/28

Technical Report 807

The Relationship of Group Cohesion to GroupPerformance: A Research Integration Attempt

Laurel W. Oliver

ARI Scientific Coordination Office, LondonMilton S. Katz, Chief

Basic Research OfficeMichael Kaplan, Director

U.S. ARMY RESEARCH INSTITUTh FO R THE BEHAVIORAL AND SOCIAL SCIENCES5001 Eisenhower Avenue, Alexandria, Virginia 22333-5600

Office, Deputy Chief of Staff for PersonnelDepartment of the Army

July 1988

Army Project Number Performance and Training2Q161102874F

Approved for public release; distribution unlimited.lti.i ...


6/28

FOREWORD

Over the past several years, the phenomenon of group co-hesion has become increasingly important to the Army. Thisinterest has led to increased research effort in the areas ofcohesion and performance. Although there have been many reviewsand summaries of the cohesion research, there has been to date noquantitative integration of that body of research. This reportdescribes an effort to apply a meta-analytic approach to theintegration of the cohesion-performance literature that employedreal world groups and contained empirical data. As such, thereport models some innovative techniques fo r research integrationand also provides support fo r continuing the present research oncohesion and performance.

EDG RM&JOHNIN 4;4Technical Director

vU


7/28

THE RELATIONSHIP OF GROUP COHESION TO GROUP PERFORMANCE: A

RESEARCH INTEGRATION ATTEMPT

EEUTIVE SUMMARY

Requirement:The phenomenon of group cohesion is of great interest to theArmy since it is viewed as a group characteristic that is to alarge extent under the control of the commander and is also posi-tively linked to unit performance. Although much of the cohesionresearch has involved laboratory studies conducted on smallgroups organized for the purpose of the research, some studieshave involved real world groups such as military units, indus-trial work groups, and sports teams. The purpose of the presenteffort was to integrate the research literature investigating therelationship of cohesion and performance in real world groups byusing a meta-analytic (quantitative) approach.

Procedure:A search was made for research reports that met the criteria

of real groups, empirical data, and cohesion-performance rela-tionships. The 14 codable documents so identified were codel andthe effect sizes (product-moment correlation coefficients)analyzed. If more than one cohesion-performance measure was re-ported, a study effect size was calculated by averaging the cor-relations. Tukey's (1977) stem and leaf display (an exploratorydata analysis technique) was used to summarize thi data. Rosen-thal and Rubin's (1982) Binomial Effect Size Dis Lay (BESD) wasemployed to demonstrate the effect of the mean etfect size onsuccess rate.Findings:

The product-moment correlations between cohesion and per-formance for the 14 studies ranged from -. 04 to .90. The mediancorrelation was .36, and the unweighted mean r was .41. Whenstudy effect sizes were weighted by the number of groups in-volved, the mean became .33. Rosenthal and Rubin's BESD demon-strated that a correlation of .33 increases success rate (highperformance) from 34 percent to 66 percent when cohesion risesfrom low (below the median) to high (above the median). Assumingthat cohesion is related to performance at about this level (cor-relation of .33), higher levels of cohesion would seem to be verydesirable for real world groups such as Army units. However, theresults are problematical due to th e very small number of codable

vii

11111 111 'l Id I I


8/28

studies and th e conceptual and methodological problems associatedwith th e cohesion-performance research. Also, th e phenomenon maybe cyclical in nature, with high performance leading to greatercohesion, as well as higher levels of cohesion enhancingperformance.

Utilization of Findings:The findings reported above provide support fo r currentefforts to enhance performance in Army groups by increasing th ecohesion of those groups. The research also models a meta-analytic approach to research integration.

viii


9/28

THE RELATIONSHIP OF GROUP COHESION TO GROUP PERFORMANCE: ARESEARCH INTEGRATION ATTEMPT

CONTENTS

PageINTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . 1

Background . . . .. . . . . . . . . . . . . . . . 1Problem .......... . . . . . . . ....... 2Purpose of Researc ................ . .. 2

METHOD *.. * *.. . . . . . . .s -.. . . . e.. e. . .. . . 2Identifying Pertinent Studies ................. 2Coding Procedurezs.e.............. .3Calculation of Effect Sie............... 3Data Analysis . . . . . . . . . . . . . . . . . . . . . . 3

RESULTS *. o. *.. . . a -.. . . . . . . v. . . . . . . 10DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . 12CONCLUDING REMARKS ......... ........ . . .......... 14REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . 15APPENDIX A. STUDIES INCLUDED IN TH E META-ANALYSIS .... 19

LIST OF TABLESTable i. Summary of studies investigating cohesion-performance relationship in real world groups 4

2. Stem and leaf display of cohesion-performancecorrelations . . ......... . . . . ... . . ............... 113. Binomial Effect Size Display (BESD) forcohesion-performance product-momentcorrelation of .32 . . . . . . . . . . . . . . . . 13

ix


10/28

The Relationship of Group Cohesion to Group Performance:A Research Integration Attempt 1

INTRODUCTIONBackground

Over the past several years, the phenomenon of groupcohesion has become of increasing interest and importance to theArmy (Griffi th, 1987). Cohesion is viewed as a groupcharacteristic that is, to a large extent, under the control ofthe commander and also positively linked to unit performance.This renewed interest in cohesion has also led to increasedresearch effort on the topic.

Previous research on cohesion (or "group cohesiveness") hasbeen quite extensive. Most of it has involved laboratory studiesof small groups brought together fo r the purpose of studying thephenomenon and its characteristics. But a number of studies wereconducted with "real" groups such as military units, industrialwork groups, and sports teams. In some instances, the cohesionstudies investigated the performance variable. Sometimesperformance was contrived (as for a group given th e task ofconstructing a Tinker Toy-like structure), but sometimes itinvolved real-world tasks such as building bridges or winninggames. Although there have been numerous reviews or summaries ofthe cohesion research (e.g., Bass, 1981; Cartwright, 1968; Hare,1976; Ivancevich, Szilzgyi, & Wallace, 1977; Lott & Lott, 1965),there has to date been no quantitative integration of theliterature.

During the last decade, research ieviewers have madeincreasing use of quantitative techniques in integratingresearch. A classic article by Glass (1976) introduced theconcept of "meta-analysis"--the analysis of analyses. The meta-analytic approach to research integration calls for th econversion of research results from a set of studies to a commonmetric. This common metric, or "effect size," can then becombined across studies to derive generalizations about th eentixe sample of studies. Since the Glass article, many articlesand a number of books (Cooper, 1984; Glass, McGaw, & Smith, 1981;Smith, Glass, & M iller, 1980; Hunter, Schmidt, & Jackson, 1982;Light & Pillemer, 1984; Mullen & Rosenthal, 1985; Rosenthal,1984) have been written on meta-analytic approaches to researchintegration.

'The author wishes to express her appreciation to Dr. HarrisCooper and Dr. Lois Northrop for their helpful comments.

I I I I


11/28

ProblemAs noted above, there has been considerable research on

cohesion. Yet th e precise nature of the relationship of cohesionto performance has not been established. While it is generallyaccepted that cohesion usually enhances performance, there issome evidence that under certain circumstances cohesion impairsperformance (Etzioni, 1975). There is also some concern thatgroups assembled for th e purpose of an experiment may notfunction similarly to real life groups. 2Purpose of the Research

The purpose of this research effort was to integrate th eempirical literature on cohesion which involved real work groupsin order to explore th e relationship between cohesion andperformance. The research questions to be answered were:

(1) Is there a relationship between group cohesion and groupperformance?(2) If there is a relationship between group cohesion andgroup performance, what variables moderate this relationship?A quantitative approach was to be used in th e researchintegration effort in order to illustrate th e application ofmeta-analytic procedures and techniques. The present study is anexample of the type of research integration needed in applied

psychology (Oliver, in press).METHOD

Identifying Pertinent StudiesComputer searches were conducted on a variety of relevantdatabases such as DTIC, ERIC, SOCIAL SCISEARCH, SOCIOLOGICALABSTRACTS, and PsychINFO. Previous ARI searches on "cohesion"and "team performance" were updated. Several thousand titles andabstracts were scanned, and promising documents were obtained.

zOver the years, there has been spirited debate about "ecologicalvalidity." Berkowitz and Donnerstein (1982) have argued it is notnecessary to use real people in real world conditions to insureexternal validity. Gordon, Slade, and Schmitt (1986), however,reviewed 321 studies which involved students and nonstudentsunder identical conditions and found that comparisons oftenresulted in significant differences between the two types ofsubjects.

2


12/28

Relevant review articles were searched for appropriatereferences, and the reference lists of articles were alsoinspected fo r pertinent studies that had not otherwise beenidentified. The purpose of the search was to identify documentswhich reported empirical investigations of the relationshipbetween group cohesion and group performance and which involvedreal groups, not groups assembled fo r the purpose of the study.Of 26 research reports which met the above criteria of realgroups, empirical data, and cohesion-performance relationships,

14 documents proved to be codable. The remaining documents wereuncodable due to data reporting deficiencies.Coding Procedure

A coding form was developed which included the followingvariables in addition to the APA-style reference: definition ofcohesion, type of cohesion measure, type of sample, total numberof subjects, number of groups, number of subjects per group,performance measures, effect sizes for each performance measure,and mean effect size fo r the study. Table 1 summarizes the codedstudies.Calculation of Effect Sizes

In most cases, the relationship between cohesion andperformance was reported as a correlation coefficient, usuallyPearson's r. Nunnally (1978) has stated that phi, point-biserialE, and rho "are all the same" and equivalent to the "regular"product-moment coefficient (p. 132). Thus, in two cases(Goodacre, 19511 Rosen, 1969) in which the rank order correlationwas reported, the value for rho was used. One author (Van Zelst,1952) reported pre-post means and standard deviations whichpermitted the calculation of Cohen's d (Cohen, 1977, p. 21),which was then converted to the corresponding r using the formulain Cohen (1977, p. 23). Probability levels for-the chi squareresults of another study (Goodacre, 1953) were transformed intothe equivalent r using the formula in Glass et al. (1981, p.150).Data Analysis

The unit of analysis was the study. When more than oneeffect size was calculated because the researcher had employedmultiple performance measures, the unweighted effect sizes wereaveraged to obtain a mean effect size for the study. In averagingthe effect sizes (correlation coefficients), either within oracross studies, the r to z transformation was not used. Glass etal. (1981) do not considor this procedure necessary, nor do theyrecommend the alternative procedure of squaring the coefficientsand computing the square root of the average (of the squaredcoefficients).

3


13/28

0 C0I.2 CPa0.

a ' .- -4.b.

Ali, IILf ~:~he -_'I.4

a4

~.,.4 V3)

8 .. 6

41 -4

4..4 >v-4 404 .-* 4

b.~ ~ ~bd U6 -4

44- Uc

U3

C44


14/28

144 -4 U-- _15;- F%*%. iAb 18

o~~ 68 -. 0

%1.4.04 c 4, 4O4 "-

.1 - 4

.9 9

C4 ~~9 mw Q4El ~ '4 El l

44 a pq 5


15/28

8 0i044 1 C 0

a. b

- '.2

.- ,

-40

b V'

-44' 44

OF


16/28

44.

.1j4r4 ~ -4N4~~. II s4

144 4prIis"1444A

M.44m 4

-W I-3P4

?

r~ .04' B U -4Xi'Al '('!0qi~b. 44

04

7r4


17/28

3 .4 Aj.D~~% g0j44t3 I. ~~~-4~-

"0 ti0

-44DA~ ~[4.1

~.iin8 44

8


18/28

IAII

"r4 N M

4

'41Ii "li"4iii, p

,I9


19/28

RESULTSTable 1 contains a summary of the 14 coded studies. Foreach study, the sample, the cohesion measure, and 6.he performancemeasures are described. The "Findings" column contains data onthe cohesion-performance relationship. If more than oneperformance measure was involved, the effect size is shown for

each measure. The mean effect size for each study is alsoreported in this column. Other information of interest is givenunder "Comments," and variables other than cohesion andperformance are noted here.The stem and leaf display in Table 2 illustrates the use ofthis exploratory data analysis technique to summarize data(Tukey, 1977). In the table, the intervals to the left of thevertical bar serve as "the stem," and the correlationcoefficients to the right are the "leaves". 3 The display providesa visual picture of the distribution of the data. From such adisplay, one can observe where the values seem to be centered,how widely the values are spread, if the data are skewed in one

direction or the other, if the data separate into groups, ifcertain values or ranges of values are unexpectedlyoverrepresented (or underrepresented), etc.

"JNormally, the stem is the first digit in the coefficient (e.g.,.7, and the leaves the second (e.g., 3, 9). With the smallnumber of data points here, however, it was more meaningful togroup the numbers into intervals of .2 and to display the entirecoefficient in the leaf position.

10

i I a i i


20/28

Table 2Stem and Leaf Display of Cohesion-Performance Correlations

Interval Study r (mean) Summary statistics

.8-.9 .9 Maximum r .90

.6-.7 .69 .73 .79 Third quartile (Q3) .69

.4-.5 .42 .58 Median (02) .36

.2-.3 .24 .24 .32 .36 .37 First quartile (QI) .240-.1 .10 .12 Minimum r -.04( -) .04

Mean (r) .42SDr .28Weighted mean .32

A scan of the display indicates that the study effectsizes (correlation coefficients) appear to be more or lessnormally distributed with the lower values clusteredsomewhat more closely than the higher values. The summarystatistics found on the right of the table confirm theseobservations. The values range from -. 04 to .90, with amedian of .365. The first quartile value is .24, and thethird quartile value is .69. Also included in the summarystatistics are the unweighted mean effect size (.42), itsstandard deviation (.28), and the weighted mean of the studyeffect sizes (.32). (The weighted mean was obtained byaveraging the study effect sizes after each had beenweighted by the number of groups in the study.)

In addition to the variables of cohesion andperformance, other variables were sometimes investigated.These variables included leadership styles, psychologicaltraits, attitudes, group characteristics, and demographicdata. No related variable was reported in a sufficientnumber of research investigations to warrant quantification.

11


21/28

DISCUSSIONThis research integration attempt did not encompp.ss enoughstudies to lead to unequivocal conclusions concerning therelationship between cohesion and performance, nor did it allowan exploration of the quantitative relationship of moderatorvariables to the cohesion-performance main effect. This research

effort did, however, illustrate the procedure of quantitativeresearch integration.The effect sizes calculated for the results of the studiesincluded in this research integration were product momentcorrelations. The procedures described above demonstrated how aresearcher can convert different types of statistics to a commonmetric that can be combined within and across studies. Theeffect sizes are data which can be analyzed using essentially thesame techniques one would use for data from primary sources. Theanalyses conducted on this data set were simple descriptivestatistics; measures of central tendency and dispersion.Because of the limited number of studies, additional procedures

such as testing for homogeneity (Hedges & Olkin, 1985; Rosenthal,1984) or correcting fo r sampling error (Hunter, et al., 1982)were not followed.The approach used here also demonstrated how one may obtainsomewhat different results by using either weighted or unweightedeffect sizes. The principle underlying the weighting of effectsizes is that as the size of the sample increases, sampling errordecreases. Therefore, it is argued, one should weight moreheavily those studies which use larger samples. Following theadvice of Wolf (1986), both weighted and unweighted values arereported here.Although .42 (or .32) seems a modest correlation, Cohen(1977) has suggested that correlations of .3 are "medium" effectsizes while those of .5 or greater are "large" effect sizes.Cohen's guidelines are arbitrary, but they are often quoted anddo provide a measuring stick of sorts. Another comparison whichcan be made is with the results of the meta-analysis conducted bySpector (1986). Spector investigated the relationship betweenemployee-perceived control (autonomy and participation) andvarious outcome variables (e.g., job satisfaction, commitment,performance, and turnover). The effect sizes reported were themean correlation weighted by sample size and the mean correlationadjusted fo r attenuation in both control and (where possible)outcome variables. The data reported for the 24 samples

involving the control-performance relationship were: mean r =.20; adjusted r a .25. While these correlations are lower thanthose found for the cohesion-performance relationship in thisresearch, the results are not dissimilar.Another way of interpreting a correlation coefficient is byusing the binomial effect size display (BESD) developed byRosenthal and Rubin (1982). (Also see Rosenthal, 1984, pp. 129-AR

12p O il I


22/28

132.) The BESD answers the question of what the effect is of apredictor variable (selection device, organizationalintervention, treatment, and the like) on the success rate (e.g.,retention rate, improvement rate, survival rate) attributable tothe predictor. For illustrative purposes, it is assumed thatincreasing cohesion enhances performance.To demonstrate the BESD procedure, the weighted meancorrelation of .32 is used. Following the example of Wolf (1986,p. 32), cohesion and performance are classified into "high" and"low" categories by a median split.

Table 3Binomial Effect Size Display (BESD) for Cohesion-PerformanceProduct-Mcment Correlation of .32Performance

Cohesion High Low TotalAbove median level 66 34 100Below median level 34 66 100

Total 100 100 200

Table 3 depicts the effect of a correlation of .32 onsuccess rate (high performance). As can be seen in the table, 66percent of the high performing units would be above the medianand only 34 percent below the median on cohesion. This resultsuggests that changing the cohesion level from low to high isassociated with increasing the number of high performers fromabout 34 to 66 percent. 4 Assuming that cohesion is truly relatedto performance at about this level, higher levels of cohesionwould seem to be definitely desirable for real world groups suchas Army units.To the extent that this small sample is representative ofthe population of empirical studies based on real groups, amoderately strong positive relationship between cohesion andperformance appears to exist. However, the correlational data donot permit the inference of causality. Although it is generally

accepted that higher levels of cohesion lead to more effectiveperformance, it may be that the reverse is also true--i.e.,higher performance enhances group cohesion. Winning teams may be

"4Using the less conservative value of .42 (unweighted meancorrelation), the success rate would rise from 29 to 71 percent.

13


23/28

cohesive because they win, rather than they win because they arecohesive. Or, perhaps, a cyclical relationship is involved, withhigh levels of cohesion leading to more effective performancewhich in turn increases group cohesion. It should also beemphasized that this research focused only on the simplerelationship of cohesion and performance. Since real worldsettings are complex, this relationship is undoubtedly influencedby a number of other variables. Before drawing final conclusionsabout the relationship of cohesion and performance, it would beimportant to identify those moderator variables and delineatetheir separate and joint effects on the cohesion-performancelink.

CONCLUDING REMARKSIt is emphasized that the set of studies upon which thisresearch integration effort is based is very small. In addition,there are a number of conceptual and methodological problemsassociated with the cohesion-performance research which make theresults of this analysis problematical. These problems arediscussed elsewhere (Oliver, 1987). However, the researchreported here does illustrate the use of exploratory dataanalysis and a quantitative approach to research integration. Theresults also suggest that there is a positive relationshipbetween cohesion and performance that may have practicalimplications for Army units and other real world groups.

14


24/28

REFERENCES

York: Free Press.Berkowitz, L., & Donnerstein, E. (1982). External validity is

more than skin deep: Some answers to criticisms of laboratoryexperiments. American Psychologist, 37, 245-257.Cartwrigift, D. (1968). The nature of group cohesiveness. In D.Cartwright & A. Zander (Eds.)., Group dynamics:. Research andtheory (pp. 91-109) (3rd ed.), New York: Harper & Row.Cohen, J. (1977). Statistical power analysis for the behavioralsciences. (rev. ed.). New York: Academic Press.Cooper, H. M. (1984). The integrative research review: asystematic approach. Beverly Hills, CA: Sage.Etzioni, A. (1975). A comparative analysis of complexorganizations: On power, involvement, and their correlations(rev. ed.), New'York: Freeq Press. ..Fiedler , F. E. (1967). A theory of leadership effectivenesu.New York: McGraw-Hill ....Glass, G. V. (1976). Primary, secondary, and meta-analysis of

research. Educational Researcher, 5, 3-8.Glass, G. V., McGaw, B., & Smith, M. L.(1981). Meta-analysis insocial research. Beverly Hills, CA: Sage.Gordon, M. E., Slade, L. A., & Schmitt, N. (1986). The scienceof the sophomore revisited: From conjecture to empiricism.Academy of Management Review, 11, 191-207.Griffith, J. (April, 1987). Group Cohesion, trainingPerformance, Social support and the Army's new unitreplacement system. Washington, DC: Department of MilitaryPs-ychiatry, Walter Reed Army Institute of Research.Hare, A. P. (1976). Handbook of small group research. (2nded.). New York: Free Press.Hedges, L. V. & Olkin, I. (1985). Statistical methods for mtas-

analysis. Orlando, FL : Academit~c Press.Hunter, J. E., Schmitt, F. L., & Jackson, G. B. (1982). Meta-analysis: Cumulating research findings across studies.Beverly Hills, CA: Sage.


25/28

Ivancevich, J. M.. Szilagi, A. D. Jr., & Wallace, M. J. Jr.(1977). Orqanizational behavior and performance. SantaMonica, CA: Goodyear Publishing.Light, R. J., & Pillemer, D. B. (1984). Summing up: The scienceof reviewing research. Cambridge, MA: Harvard UniversityPress.Lott, A. J., & Lott, B. E. (1965). Group cohesiveness asinterpersonal attraction: A review of relationships withantecedent and consequent variables. Psychological Bulletin,64, 259-309.Martens, R., & Peterson, J. A. (1971). Group cohesiveness as adeterminant of success and member satisfaction in teamperformance. International Review of Sport Psychology, 6, S-59.Mott, P. E. (1972). The characteristics of effectiveorganizations. New York: Harper & Row.Mullen, B., & Rosenthal, R. (1985). BASIC meta-analysis:Procedures and programs. Hillsdale, NJ: Lawrence Erlbaum.Nunnally, J. C. (1978). Psychometric theory (2nd ed.). NewYork: McGraw-Hill.Oliver, L. W. (in press). Research integration forpsychologists: An overview of approaches. Journal of AppliedSocial Psychology.Oliver, L. W. (In Preparation). Cohesion research: Conceptualand methodoloaical issues. (ARI Technical Report). Alexan-

dria, VA: US Army Research Institute fo r the Behavioral andsocial Sciences.Rosenthal, R. (1984). Meta-analytic procedures for social

research. Beverly Hills, CA: Sage.Rosenthal, R., & Rubin, D. B. (1982). A simple, general purposedisplay of magnitude of experimental effect. Journal ofEducational Psychology, 74, 166-169.Smith, M. L., Glass, G. V., & Miller, T. I. (1980). The benefitsof psychotherapy. Baltimore, MD: Johns Hopkins UniversityPress.Spector, P.E. (1986). Perceived control by employees: A meta-analysis of studies concerning autonomy and participation atwork. Human Relations, 39, 1005-1016.Stogdill, R. M. (1965). Manual for group dimensionsdescriptions. Columbus, OH: Bureau of Business Research,Ohio State University.

16


26/28

Tukey, J. W. (1977). Exploratory data analysis. Reading, MA:Addison-Wesley.Wolf, F. M. (1986). Meta-analysis: Quantitative methods fu rresearch synthesis. Beverly Hills, CA: Sage.

17


27/28

APPENDIX ASTUDIES INCLUDED IN THE META-ANALYSIS

Blades, J. W. (1986). Rules for leadership: Improving unitperformance. Washington, DC: vational Defense UniversityPress.Downey, R. G., Duffy, P. J., & Shiflett, S. (1975). Criterionperformance measures of leadership and unit effectiveness insmall combat units. (ARI Research Memorandum 75-9).Alexandria, VA: US Army Research Institute for the Behavioraland Social Sciences.Goodacre, D. M. III. (1951). The use of a sociometric test as apredictor of combat unit effectiveness. Sociometry, 14, 148-

152.Goodacre, D. M. III. (1953). Group ch.racteristics of good andpoor performing combat units. Sociometry, 16 , 168-179.Griffith, J. (1987, April ). Group cohesion, trainingReplacement System. Paper presented at the meeting of theEastern Psychological Association, Arlington, VA.Hemphill, J. K., & Sechrest, L. (1952). A comparison of threecriteria of air crew effectiveness in combat over Korea.Journal of Applied Psychology, 36, 323-327.Keller, R. T. (1987). Predictors of the performance of projectgroups in R&D organizdtions. Academy of Management Journal,29, 715-726.Manning, F. J., & Trotter, R. (1980, April). Cohesion andpeacetime performance by selected combat units. Paper

presented at the VII Corps Battal..on Commanders' Conference,Nuernberg, West Germany.Melnick, M. J., & Chemers, M. M. (1974). The effects of groupsocial structure on the success of basketball teams. TheResearch Quarterly, 45, 1-8.Moos, R. H. (1986). Military Environment Inventory manual.Palo Alto, CA: Consulting Psychologists Press.Nelson, P. D., & Berry, N. H. (1968). Cohesion in Marinerecruit platoons. (Report No. 66-26). Washington, DC: Navy

Medical Neuropsychiatric Unit, Bureau of Medicine and Surgery,Department of Navy.1

I%I ~~19


28/28

Rosen, N. A. (1969). Leadership change and work-group dynamics.Ithaca, NY: Cornell University Press.Schriesheim, J. F. (1980). The social context of leader-subordinate relations: An investigation of the effects ofgroup cohesiveness. Journal of Applied Psychology, 65 , 183-194.Van Zelst, R. H. (1952). Sociometrically selected work teamsincrease production. Personnel Psychology, 5, 175-188.

Date post:	06-Apr-2018
Category:	Documents
Upload:	tomfriis
View:	217 times
Download:	0 times

The Relationship of Group Cohesion to Group Performance - A Research Integration Attempt - Laurel W...

Documents