See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/270135689

Effects of team building and goal setting on productivity:

A field experiment

Article  in  The Academy of Management Journal · June 1986

DOI: 10.2307/256190

CITATIONS READS

86 4,552

2 authors, including:

Paul Buller
Gonzaga University
44 PUBLICATIONS   1,470 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Paul Buller on 16 November 2015.

The user has requested enhancement of the downloaded file.

* Academy of Management /ournal
19B6. Vol. 29, No. 2. 305-328.

EFFECTS OF TEAM BUILDING

AND GOAL SETTING ON PRODUCTIVITY:
A FIELD EXPERIMENT
PAUL F. BULLER
Utah State University
CECIL H. BELL, JR.
Universily of Washington

A field experiment conducted with 53 hard-rock miners in an under-

ground metal mine examined the effects of two interventions, team
building and goal setting, on miners' productivity and strategy develop-
ment. The study employed a 2 x 2 factorial design with random assign-
ment of participants to the goal setting condition, and it used objective
criteria of performance. Unfortunately, the results were inconclusive.
Although slight improvements occurred in some measures of perfor-
mance and strategy development, it is not clear that these improve-
ments were results of the interventions. More of these improvements
occurred in the goal setting than in the team building condition. Limita-
tions in the research design and factors in the experimental setting that
could not be controlled led to equivocality of these results.

Numerous techniques drawn from behavioral science for improving indi-

vidual and rrganizational performance figure in the contemporary manage-
ment scene. Practicing managers are offered a veritable smorgasbord ranging
from alphabet soup (QWL; OD; MBO: TA; OB Mod; Theories X, Y, Z; QC;
PDM; etc.) to nuts. Naturally, their adherents and practitioners make claims
for the efficacy of these techniques and counterclaims are forthcoming from
critics and skeptics. Three reviews of the empirical literature [Cummings,
Molloy, & Glen, 1975; Huse, 1980; Locke, Feren, McCaleb, Shaw, & Denny,
1980) generally agreed that only a few intervention techniques have been
shown to be effective through well-designed and well-executed research
Investigations. We believe that rigorous evaluation of applied behavioral
science programs is desirable and that field experiments in organizational
contexts are particularly needed. This study describes a field experiment
that examined the effects of two prominent intervention techniques, team
building and goal setting, on individuals' productivity.

TEAM BUILDING
One of the most popular intervention techniques in organization devel-
opment (OD) is team building (French & Bell, 1984; Porras & Berg, 1978). The
The authors wish to thank Gary P. Latham for his helpful comments on an earlier version of
this paper. The suggestions of three anonymous reviewers were also most helpful. This research
was conducted in part under U.S. Bureau of Mines contract number 10387230.
305
306 Academy o/Management Journal June

primary objective of team-building activities is to increase the effectiveness

of work teams. This is usually achieved through improving problem solving
and interpersonal relations in teams. Team building is not a well-defined
concept, however. Organizations employ many variations; the emphases of
programs change over time; and programs are often situation- and problem-
specific. Beer (1976) described four models of team building: a goal-setting
model, an interpersonal relations model, a role clarification model, and the
managerial grid model. Dyer (1977) described a more general model of team
building based on problem solving in which team members identify major
problems, generate relevant information, engage in problem solving and action
planning, and implement and evaluate action plans. We refer to this task-
oriented problem-solving technique for teams as team building/problem
solving. The assumption underlying this type of intervention is that work
teams become more effective by solving their major problems together. The
team-building program studied in this research used team building/problem
solving techniques that are operationally defined in the Methods section.
Research assessing the effectiveness of team building has been inconclu-
sive (DeMeuse & Liebowitz, 1981; Nicholas, 1982; Woodman & Sherwood,
1980). Positive results were found in 29 of the 36 studies (80%) reviewed hy
DeMeuse and Liebowitz, in 2 of the 4 studies reviewed by Nicholas, and in
19 of the 30 studies (63%) reviewed by Woodman and Sherwood. Neverthe-
less, these authors concluded that they could make no firm statements regard-
ing the efficacy of team building with respect to its effects on the perfor-
mance of individuals and work groups. They based their conclusions on the
fact that the research designs of the studies reviewed were generally weak,
and the findings thus lacked internal validity. Moreover, with the exception
of some reported by Nicholas, the studies measured subjective data such as
employees' perceptions and attitudes and usually ignored objective data on
performance. Finally, they often vaguely and imprecisely described inter-
ventions,
BuUer's (1982) review of the literature on team building yielded only
nine studies that implemented clearly defined team-building interventions
and measured effects on objective criteria of performance. Of these nine
studies, six reported positive effects on performance. However, all but one
[Margulies & Raia, 1968) of the successful team-building interventions were
confounded with goal setting or other activities; thus it was impossible to
determine which technique produced the results, Margulies and Raia's study
used a preexperimental design and thus had limited internal validity.
An additional problem with previous research on team building is its
failure to specify and test relationships between independent variables, causal
mechanisms, and criteria of performance. It is simpiy not clear why team
building affects performance, if it does at all. Nicholas (1982) suggested that
team building may lead to improved performance through "modification of
attitudes, values, problem-solving techniques, and interpersonal and group
processes" (1982: 535). A recent model presented by BuUer (1983) suggested
that team building potentially affects a number of individual, group, task.
1986 BuJler and Bel/ 307

and organizational factors that in turn improve performance. This model

predicts that the performance of a team will improve to the extent that it can
identify and solve major debilitating problems in one or more of these areas.
However, existing models are weak in defining specific theoretical relations.
Perhaps Hackman's (1976) model of the effects of workgroups on the perfor-
mance of individuals is the most relevant to this study and the most
parsimonious in describing how team building might work. He argued that
individual performance is a function of an individual's ability, motivation,
effort, and strategies for performance—all moderated by the nature of the
task being performed. To the extent that immediate workgroups enhance or
hinder these mechanisms, they enhance or hinder performance. Certain task
contingencies may affect which of the causal mechanisms is most important
to improved performance.
This research was not an attempt to develop or validate a theory of team
building. The primary purpose of the study was to test team building's
independent effects on performance using objective criteria of performance.
However, the research directly examined the effects of team building on
development of task performance strategies, an individual cognitive mech-
anism. To our knowledge, no previous research has directly tested such a
cognitive effect of team building.

GOAL SETTING
The importance of goal setting for motivating employees' performance
was first explicitly recognized in the management literature by Taylor (1911);
Locke (1968) translated the concept into a contemporary theory of task
performance. According to Locke's theory, if workers accept hard, specific
goals, performance improves, in the most thorough review of goal setting
to date, Locke, Shaw, Saari, and Latham (1981) viewed goal setting as primar-
ily a motivational phenomenon, but acknowledged that it also involves cog-
nitive elements. From a motivational standpoint, goal setting affects the
direction, amplitude [effort), and duration [persistence) of action. From a
cognitive perspective, goal setting is thought to affect strategy development
or ways to accomplish goals. The first three of these mechanisms are rela-
tively direct in their effects on performance. Strategy development is more
indirect. Locke and colleagues (1981) noted that this concept
involves developing strategies or action plans for attaining one's
goals. Although strategy development is motivated by goals, the
mechanism itself is cognitive in essence; it involves skill devel-
opment or creative problem solving (1981: 132).
Goal setting, the subject of extensive experimental research in both labo-
ratories and field settings, has been found to be a robust technique for improv-
ing task performance [Latham & Yukl, 1975; Locke et al,, 1981), Most of this
research has focused on the first three mechanisms by which goals affect
performance. Few studies have directly examined the effects of goal setting
on strategy development, although several studies have inferred such effects.
For example, Bandura and Simon [1977) noted that overweight clients in a
308 Academy of Management /oumal . June

weight clinic who had specific goals for mouthfuls eaten altered their eating
habits. Some individuals ate low calorie foods that did not count in their
quotas; some engaged in more planning than before, Latham and Baldes
[1975) observed similar behavior in a study of drivers of logging trucks.
Some of the drivers who were assigned specific challenging goals for trucks'
weights recommended small changes in their trucks that let them more
accurately judge weight. In general, participants in these studies with hard
specific goals engaged in more creative problem solving than participants
with no goals.
In another relevant study, Terborg (1976) obtained separate measures of
the effects of goal setting on individuals' efforts and the direction of their
behavior on a reading task in a simulated organization. The study operation-
ally defined direction of behavior as the extent to which individuals employed
appropriate behaviors or action plans to accomplish the task. This was essen-
tially a measure of strategy development. The results showed that goal set-
ting was related to both effort and direction of behavior, which in turn were
related to performance. Specifically, individuals who set hard specific goals
on the reading task spent more time on the reading materials [effort) and
were more likely to use learning strategies relevant to their goals (direction
of behavior) tban those who did not set goals.
Finally, scholars have observed that strategy development may be espe-
cially important in complex tasks because if people do not develop strategies
that are relevant to goais, increased motivation caused by goal setting may
not enhance performance (Bavelas & Lee, 1978; Christensen-Szalanski, 1980).
The nature of tasks may therefore be an important variable determining
whether or not goal setting will stimulate development of strategies, lead to
improved performance, or both.
The studies discussed in the preceding paragraphs support conceptual
arguments that both cognitive and motivational mechanisms influence per-
formance of tasks (Naylor, Pritchard, & Ilgen, 1980; Terborg & Miller, 1978).
They provide some evidence that goal setting affects cognitive as well as
motivational determinants of performance. To date, only one reported study
has systematically examined the effects of goal setting on strategy develop-
ment. In a laboratory experiment, Shaw (1983) found that specific hard goals
led to the development of more task strategies than no goals did. The present
study investigated the effects of goal setting on strategy development and
task performance in a field setting.
In summary, our hypotheses were straightforward: team building and
goal setting will each have significant, positive effects on task performance
and strategy development.

METHODS
Context and Setting
The present research was part of a larger organization development (OD)
project designed to demonstrate the effectiveness of OD techniques for improv-
1986 BuJIer and Bell 309

ing safety and productivity in mines. Conducted at a silver mining company,

the program consisted primarily of team-building activities at all levels of
the organization but also involved a variety of other activities. These included
a revision of the company's performance appraisal system to place more
emphasis on safety, an assessment of data processing needs, development of
a program for training supervisors, and the experiment reported in this paper.
Although the other activities potentially confounded the effects of the
experiment, it is important to note that only the supervisory training pro-
gram also directly involved hourly workers. Further, the organization did
not implement this training program until after we had completed our study.
The entire OD project took approximately three years; the field experiment
extended over eight months.
The site for the field experiment was an underground mine producing
silver, lead, and zinc. The mine employed 292 persons with approximately
220 hourly employees involved in underground production operations. The
average hourly employee was 38 years old, had a twelfth grade education,
and had 12 years of work experience. About 60 percent of the workforce
belonged to an international union. The mine was a hard-rock mine in which
ore-bearing rock in vein structures was mined in small working places called
stopes.
As part of the overall OD program, a team-building intervention was
conducted with approximately half of the hourly underground employees
and their supervisors at the mine. Adding the goal-setting induction gener-
ated a 2 X 2 factorial design and thereby allowed the controlled investigation
of both behavioral science interventions in a single field experiment. Thus,
the experiment was a study within a study and was more limited in scope
than the overall OD effort. To our knowledge no one has previously imple-
mented and evaluated either team building or goal setting in the mining
industry. Additionally, few studies of team building have been conducted
with hourly employees as participants.
Study Design
The units of analysis for this study were stopes, small underground
working places in the mine. Twenty of the mine's 23 production stopes were
assigned to the treatment conditions; 3 stopes were excluded because they
were not operational during the course of the experiment. To implement the
team-building condition, the mine superintendent nonrandomly assigned 12
stopes consisting of 36 miners to undergo team building; the 8 remaining
stopes were assigned to the no-team-building condition. The reasons tbe 12
stopes were assigned to team building were: (1) four intact work crews oper-
ated them, each crew consisting of a shift boss (first-line supervisor), stope
miners, and hourly support personnel; (2) they constituted two complete
production levels underground; (3) they included opposite shift crews, with
one crew working the day shift and the other crew working the afternoon
shift on the same level, and the superintendent wanted to address problems
that existed between opposite shifts; (4) the locations of these stopes were
310 Academy o/Managemenf Journal June

important to overall mine production; and (5) the superintendent wanted at

least half of the hourly workforce to get an exposure to team building.
Two weeks after the team-building intervention was completed, the goal-
setting condition was implemented by randomly assigning 10 stopes consist-
ing of 26 miners to the goal-setting condition and 10 stopes consisting of 27
miners to the no-goal-setting condition. Thus, a total of 53 miners, all men, in
20 production stopes were assigned to experimental conditions as shown in
Figure 1, The resulting 2 x 2 factorial design was quasi-experimental because
it lacked complete random assignment to conditions and simultaneous induc-
tion of the two treatments (Cook & Campbell, 1979), Figure 2 shows the
timeline for the experiment.
Two miners worked most stopes, but some stopes had one, three, or four
individuals. Since it was operationally impossible to determine every miner's
individual performance, we used performance of stope teams as the unit of
analysis. Clearly, this situation presents problems in distinguishing between
the performance of individuals and groups.
The stope miners' task. Stope miners are the primary production work-
ers and major source of increases or decreases in productivity in such mines.
Typically, two miners work in each stope for long periods of time, working
both alone and together. Each mines a horizontal cut ahout 100 feet long in
opposite directions from the raise, a vertical opening on the vein. All other
times they help each other in certain phases of the mining cycle. Miners
share equal responsibility for a stope's production. As the rock is mined out,
they pump a sand slurry back in to raise the floor of the stope and the mining
cycle of "drill, blast, and muck"^ begins again. All miners use similar equip-

FIGURE 1
Factorial Design for the Experiment

Goal Setting No Goal Setting

Team Building n = 6 stopes n = 6 stopes

{18 miners) (18 miners)

No Team Building n = 4 stopes n = 4 stopes

(8 miners) (9 miners)
1986 Builerond Bell 311

ment and mining methods. A crew of support people assists the stope miners
by hringing necessary supplies to the work area and helping to remove the
ore. Each level of the mine has from one to six working stopes, and one
support crew serves all stopes on a level.
Contact between Researchers and Participants
Sponsored by the U.S. Bureau of Mines as a consultant for the three-year
OD project, the second author conducted all of the team-building sessions
described later in this section. The first author, also sponsored by the Bureau
of Mines, was a research assistant during one year of the OD project. As a
research assistant, he observed but did not participate in the team-building
sessions with underground crews, conducted the goal-setting intervention,
and collected and analyzed the data reported here.
Independent Variables
The experimental treatments were team building/problem solving and
goal setting. Each variable had two conditions, the presence or absence of the
treatment.
Team building/problem solving. This variable was operationally defined
as a planned series of meetings, facilitated by an outside consultant, with a
team—a group of people having common organizational relationships and
goals. These meetings must be designed specifically to improve the team's
accomplishment of tasks by (1) developing problem-solving procedures and
skills and (2) then solving the team's major problems.
The team-building induction was implemented through a series of six
45-minute meetings over a period of ten weeks. Entire crews, including shift
bosses, attended meetings; groups included from 12 to 20 individuals. The
team-building sessions were task-centered; they focused on surfacing and
solving problems that got in the way of working effectively, efficiently, and
safely. The second author facilitated meetings as an outside consultant and
helped the crews to bring problems related to their task to the surface. These
groups identified task-related problems, generated ideas concerning the causes
of the problems, solved problems and planned actions, and implemented
and evaluated plans.
It should be noted that the team-building manipulation in this study was
a global intervention. By definition, team building works with problems that
teams identify. The specific problems addressed differed somewhat among
the four crews. However, the basic process was the same; all team-building
activities contained elements of the operational definition. They were (1)
planned, (2) facilitated by an outside consultant, [3) engaged in by intact
workgroups, [4) designed to develop the problem-solving capability of the
crews, and [5) intended to solve major problems.
Description of specific team-building activities. Team building for each
crew began with the consultant providing information on the background

' To "muck" means to remove the broken rock.

312 Academy o/Management Journol - June

and purpose of the meetings. It was emphasized that the meetings' primary
purpose was to improve the efficiency and safety of underground operations
and to help make the mine a better place in which to work. Participants were
told that all information gathered in the meetings would be used to generate
constructive ideas for improvements and that their active participation was
voluntary but welcome. The consultant then raised three specific questions:
[1) How can we do the job better? (2) How can we make this a better place to
work? [3) How can we make this a safer place to work? Focusing on these
questions, crew members interacted with their shift bosses, with one anotber,
and with the consultant. The consultant publicly recorded major points from
the discussions on flip charts. He also condensed the problems, issues, and
solutions from each meeting into major themes and fed them back to the
crews via handouts at the beginning of each subsequent meeting. The mine's
management also got a record of major themes for diagnosis of problems and
planning of action.
Four major issues emerged from the team-building sessions: (1) poor
communication between opposite shift crews and between miners and mine
management; (2) poor working conditions underground in terms of both
safety and performance; (3) poor repair and maintenance of equipment; and
(4) pay and bid systems that were perceived as unfair. The crews' discus-
sions generated a variety of suggestions for addressing these problems. The
crews implemented some solutions directly. For example, in an effort to
improve communications between opposite shifts, they scheduled daily meet-
ings between opposite shift crews; none had existed before. The consultant
facilitated several of these meetings between crews. Crews also acted on
some suggestions involving underground working conditions, such as stor-
ing timber in safer, more accessible areas, cleaning up misplaced tools and
materials, improving repair of tracks, and taking better care of drilling
equipment.
During the fifth of the series of team-building meetings, the mine man-
ager and superintendent interacted with the crews on issues that directly
involved management of the mine and were beyond the decision making
scope of the crews. Some of the problems were acted on immediately. For
example, management freed some resources to create additional storage space
underground. Some problems, primarily those dealing with maintenance
and the pay system, were acknowledged as important and placed under study.
Other issues, most notably the need for larger facilities and parking areas,
were postponed for later resolution, since the mine had plans for future
expansion. As management subsequently acted on problems, they made those
actions known to the crews via the meetings and notices posted on a bulletin
board.
Goal setting. This variable was operationally defined as the setting of
specific, difficult, but attainable production goals and the receiving of weekly
feedback on performance in relation to those goals. Performance feedback
was an intergral part of the manipulation because previous research has
1986 BulJerandBeJI 313

found feedback to be a necessary condition if goals are to influence perfor-

mance (Locke et al., 1981).
The goal-setting induction was implemented as follows. Two weeks
prior to actual goal setting with stope miners, those shift bosses with miners
assigned to the goal-setting condition received one hour of training, con-
ducted by the first author, on setting goals and giving feedback on perfor-
mance. Next, the goal-setting induction occurred in a one-hour meeting, also
conducted by the first author, in which stope miners and shift bosses received
graphs depicting the past 18 months of production in their stopes. They were
asked to set specific, difficult, but attainable goals for tons per manshift and
grade of ore^ for the next three months. Goals were set participatively, with
the miners and their bosses establishing mutually agreed upon goals that met
the stated criteria. After goals were set the researchers recorded them on a
feedback sheet and gave copies to each miner and shift boss. The feedback
sheets, which were updated each week, included (1) graphs that displayed
the recent output of that individual's stope in terms of tons per manshift and
grade of ore. and (2) a statement that gave the goals that had been agreed on
for the three-month trial period and described the particular stope's average
output to date. Shift bosses were instructed to present feedback sheets to
miners in the goal-setting condition weekly. Direct observations and post-
experimental interviews confirmed that shift bosses did provide this regular
feedback.
Dependent Variables
Two productivity measures were generated, one for quantity of produc-
tion and one for quality of production. Tons per manshift is a standard
measure of productivity in mines. It measures the amount of rock broken and
removed from a stope per person per eight-hour shift. We used this variable,
which was available from mine records every two weeks, as a measure of
quantity of production. During the eight-month experimental period, 16 mea-
sures of tons per manshift were taken; the first two months of the period
provided baseline data.
Grade of ore is also a standard measure of productivity in mines. It
measures the numbers of ounces of silver and lead per ton of ore taken out of
a stope. This variable, available from mine records monthly, was used as a
measure of the quality of production. Eight measures of grade of ore were
taken during the experimental period, including two measurements of base-
line data.
A third dependent variable was strategy development, operationally
defined as changes in miners' behaviors related to improving grade of ore. A
document known as the grade strategy checklist provided a measure of grades.
It listed seven specific observable behaviors related to grade improvement
that stope miners could control. The mine geologist and the first author
developed the checklist, which contained the following items: controlling
the height and width of the stope. mining with the ore vein, putting in
^ These two measures are defined in the section on dependent variables.
314 Academy of Management Journal June

sufficient ground support, using proper slushing techniques, cleaning out

rock carefully, and accurately locating the brow for a new cut. The mine
geologist, who was blind to experimental condition, completed the checklist
twice for each stope—once before the first goal-setting induction and once
three months later. He rated each behavior according to the following
scheme; good = 15;average = 10; below average = 5. A total of 105 points was
a perfect score. The grade strategy checklist represented an unobtrusive
measure of the extent to which miners were engaging in appropriate be-
haviors to improve grade of ore. It further served as a direct test of the hy-
pothesis that goal setting and team building lead to the development of
strategies.
Measures of goal acceptance and goal difficulty were obtained after the
goal-setting inductions to check for any differences among conditions on
these variables. Following Latham and Steele (1983), we measured goal accep-
tance and goal difficulty with one face-valid question each; participants
responded to these questions on a 5-point Likert-type format.
RESULTS
Productivity
Table 1 summarizes the means for all three productivity variables before
and after the experimental manipulations. Figures 2, 3, and 4 display plots of
the dependent measures over time. ,

TABLE 1
Summary of Means for Tons per Manshift, Grade of Silver,
and Grade of Lead before and after the Interventions

Dependent Experimental Before Interventions* After InlerventioDs''

Variables Conditions Means s.d. Means S.d.
Tons per manshift Team building and
goal setting 14.52 5.63 13.73 5,41
Goal setting 19.32 8.28 23.16 12.51
Team building 13.23 2.28 13,22 3.41
Control group 8,38 3.43 8.91 3,91
Grade of silver Team building and
(ounce of silver goal setting 11.08 1.43 15.24 4.01
per ton of ore) Goal setting 19.95 5.80 20.20 s.as
Team building 10.75 1,06 14,30 0,45
Control group 16.07 0.47 32.42 5.97
Grade of lead Team building and
(percent of lead goal setting 7.36 1.01 7,58 1,05
per ton of ore) Goal setting 6.31 0.74 7,13 2,00
Team building 6.98 0,64 9,33 0.79
Control group 5.92 1.48 4,37 1.30
" Means sbown are averages of tbe dependent measures taken during tbe tbree months pre-
ceding the interventions,
'' Means sbown are tbe averages of the dependent measures taken during the three months
following tbe interventions.
1986 BuIJer and Bell 315

•S S
c -'

.9 £

I
J
;jiqsuBui
316 Academy of Management /ournal June

•B
a

il

(310 }o uoi/i9A[is JO saouno)

JSAJIS }O
1986 BuJler and Bell 317

JO uoj/pea] jo %)
pea|
318 Academy o/Management/ournal June

Note that in Table 1 the largest increase in the mean grade of silver was
in the control group. However, one stope accounted for 88 percent of this
increase. In examining these results, the mine geologist stated that the large
increase in grade in this stope resulted from a change in geological condi-
tions rather than from improved performance of miners. We therefore con-
ducted all of the following analyses with and without the data from this
stope hut found no appreciable differences hetween the results of the two
analyses.
Since complete random assignment to experimental conditions was not
possihle, we followed procedures for analyzing quasi-experimental data sug-
gested by Cook and Camphell (1979). First, we performed an analysis of
variance on the productivity measures for the two-month baseline period
preceding the experimental manipulations to determine if any significant
differences existed among miners in the various conditions. There were no
significant differences hetween conditions on pretest measures of tons per
manshift and grade of lead, but there was a significant main effect for team
huilding on grade of silver (F^-(6 = 4.28, p<.05). In addition, we conducted
an analysis of variance on measures of goal acceptance and goal difficulty to
determine if the experimental conditions differed on these variables. No
significant differences were found.
Next, an analysis of variance with repeated measures was performed for
each of the three productivity variables using measures covering the eight-
month experimental period. These results are given in Table 2. They show
that there was a significant difference hetween stopes in the goal-setting and
no-goal-setting conditions on tons per manshift but not on grades of silver or
lead. There was a marginally significant difference between the team-building
and no-team-building conditions on grade of silver but not on tons per
manshift or grade of lead. There was a marginally significant interaction
between team building and goal setting on tons per manshift. Additionally,
there was a significant time effect of the two interventions on tons per manshift
and a marginally significant time effect on grade of silver. Finally, we found
a marginally significant team x goal x time interaction on grade of silver. We
found no significant differences among any of the conditions on grade of
lead.
The results of the analysis of variance with repeated measures are,
unfortunately, somewhat equivocal. We found no significant difference among
conditions on preexperimental measures of tons per manshift, but found a
significant difference between the goal-setting and no-goal-setting condi-
tions over the course of the experiment. This finding, in conjunction with
the results summarized in Figure 2, provides some support for a positive
effect for goal setting. Yet, we cannot draw an unambiguous conclusion for
two reasons. First, even though the difference before the experiment between
the goal-setting and no-goal setting conditions on tons per manshift was not
significant, it was substantial and could by itself conceivably account for the
results. Second, the small sizes of cells along with considerahle variability in
dependent measures could have produced spurious results.
1986 BulJer and Bell 319

CO i n in Ol

c a

•a -o
S es
13 « 65
^ a
-^ 9

s-5.
(/J 0) to
d c d
09 in

u
< -2 C3
I O Ul in ID •

IN

>

5
>
o
0}

S 3

" ( 3 CO
.? £ DO
Ul «
X X X
320 Academy of Management /oumaJ June

The meaning of the marginally significant difference hetween the team-

building and no-team-building conditions on grade of silver is also unclear.
There was a significant difference between these conditions on grade of
silver hefore the intervention, with miners assigned to the no-team-building
condition apparently outperforming those assigned to the team-building
condition. This initial difference between the two conditions did not change
during the course of the experiment. However, we cannot conclude that the
absence of team huilding led to greater improvement of performance than
the presence of team building. In addition to small cell sizes and high vari-
ability in the data, the lack of random assignment to team-huilding condi-
tions prohibits firm conclusions. It is possible that other factors, such as
history, selection-maturation interactions, or other threats to internal valid-
ity caused the difference found between conditions (Campbell & Stanley,
1963).
Next, an analysis of variance with gain scores was conducted. This
procedure, suggested hy Reichardt (1979), examines mean changes in perfor-
mance between treatment groups from pretest to posttest. It is a useful tech-
nique for analyzing data from nonequivalent group designs hecause, like
analysis of covariance, it helps to control for any pretest differences between
groups. Analysis of gain scores was more appropriate than ANCOVA with
our data since the pretest differences between treatment conditions were
substantial. Unlike the ANCOVA method, the precision of analysis of gain
scores is not related to the size of pretest differences (Reichardt, 1979). The
procedure employed a 2 x 2 analysis of variance with goal setting and team
building as independent variables and scores for changes in productivity as
the dependent variables. Changes were measured hoth as absolutes and as
percentages. We conducted the analysis both with and without data from the
unusual stope in the control group. No significant main or interaction effects
were found, indicating that the treatments had no effect. But again we can-
not draw firm conclusions because, as Reichardt (1979) pointed out, analysis
of gain scores is subject to other problems—bias from selection-maturation
interactions, floor and ceiling effects, and regression effects—threats to valid-
ity that simply cannot be ruled out.
To analyze the effects of treatments further, we assessed changes within
each experimental condition before and after the inductions. Using time as
an independent variahle with two levels—average performance measures both
before and after interventions—we conducted a one-way analysis of variance
on tons per manshift, grade of silver, and grade of lead for each of the four
experimental conditions. The results showed that the goal-setting-only con-
dition had a marginally significant effect on tons per manshift (F^ 3 = 7.1.
p<.08); results for no other conditions approached significance. The com-
bined team building-goal setting condition had a marginally significant effect
on grade of silver (FT,5 = 4.5, p<.09) and so did the team-building-only
condition (F^ 5 = 4.9, p<.OB). The team-building-only condition had a mar-
ginally significant effect on grade of lead (Fj^ = 4.7, p<.08), with no other
conditions approaching significance. Although the results of these within-
1986 Bailer and BeU 321

group analyses are only marginally significant, they provide some evidence
for some treatment effects. Specifically, team building showed a positive
effect on grades of silver and lead, and goal setting appeared to improve tons
per manshift. Again, since the treatment groups were nonequivalent and the
cell sizes small, we cannot entirely rule out factors other than the treatments
as explanations of the results obtained.
In summary, results of analyses of the data on productivity were not
conclusive. The weight of evidence, however, suggests that the treatments
produced little or no effect on performance. Although there is some evidence
consistent with a positive but small effect for goal setting on tons per manshift,
rival explanations of this result could not be ruled out entirely. There was
also some evidence for a positive effect for team building on grade of ore. but
this evidence is open to even more questions.
Strategy Development
An analysis of variance on pretest scores on the grade strategy checklists
showed no significant differences among experimental conditions. An analy-
sis of variance with repeated measures was then performed on pretest and
posttest scores on grade strategy checklists. We found no significant main
effects for either team building or goal setting. There was, however, a main
effect for time (Fjie = 19.7, p<.001), indicating that miners improved grade
strategy across conditions over time. When we submitted scores from grade
strategy checklists to an ANOVA with the four experimental conditions as
separate treatments and time as the independent variable, there was a signifi-
cant effect over time for the team-building-only condition (Fi,5 = 14.3.
p<.01), and marginally significant effects for the goal-setting-only (Fi,3 =
7.7, p<.07). and the combined team building-goal setting condition (Fi,5 =
4.8, p<.08). The control group showed no significant improvement on scores
for grade strategy. This within-group analysis suggests that the major gains
in strategy development over time were achieved in the experimental treat-
ment conditions. However, when an analysis of variance using gain scores
was conducted, results showed no significant differences among conditions.
Tahle 3 summarizes the mean scores on the grade strategy checklist for
experimental conditions before and after the experimental manipulations.
Qualitative Assessment of the Interventions
Although it was not the primary emphasis of this study, we also assessed
participants' reactions to the interventions. After the interventions, the first
author conducted structured interviews with three randomly selected shift
bosses and ten randomly selected hourly employees; these interviews pro-
vided data for this evaluation. In addition, these individuals completed a
brief postexperimental questionnaire pertaining to the specific problems
identified by hourly employees in the team-building sessions.
In general, the structured interviews revealed that miners liked both
interventions. With respect to team building, most miners thought that the
intervention allowed them to express their ideas to management and get
322 Academy of Management Journal June

TABLE 3
Summary of Mean Scores on the Grade Strategy Checklist*
before and after the Interventions
Experimental Conditions Before Interventions After Interventions
Team building and goal setting 55.00 70.00
Goal setting 70.00 85.00
Team building 65.83 80.00
Control group 62.50 66.25
^ Maximum possible score on the grade strategy checklist was 105.

things out in the open. Several employees observed that there seemed to be
better relations hetween management and the union as a result of the
intervention. Specific comments included: "The meetings brought a lot of
stuff out. got problems out into the open." "The guys felt better identifying
problems," and "The meetings helped us to talk better with the opposite
shift." But several hourly employees mentioned a shortcoming that team
building had: nothing seemed to change much as a result of the meetings.
Others noted a lack of interest on the part of the hourly employees and that it
was a nuisance to come to the meetings. Some representative comments
were: "Nothing really happened as a result of the meetings." and "I'd like to
see the people take a bigger part; if more would have talked up. things would
have heen better, more things would have changed."
Those miners involved in the goal-setting intervention most often men-
tioned regular feedback on their productivity and the grades they were obtain-
ing as a valuable part of the program. Representative statements were: "Goal
setting made us more conscious of what we are doing," and "It showed me
how I can make a little better money—set standards for myself." However,
two miners expressed suspicion regarding goal setting: "I thought that the
goals would be used to set day's pay and incentive rates."
As noted in the Methods section, the team-building sessions identified
several specific problems: communication between opposite shift crews;
communication among underground management personnel; politics and
favoritism in pay aud bid systems; equipment and maintenance; and working
and mine conditions. We administered a brief postexperimental question-
naire to determine these shift bosses' and employees' perceptions of the
effects of team building on these problems. The questionnaire simply asked:
"For each of the following prohlems listed, please indicate if the problem is
as a result of the crew meetings"; the response format ranged from 1 =
much worse to 5 = much better. Results indicated that workers perceived
improvement in performance in all of the problem areas. The largest per-
ceived improvement was in communication between opposite shift crews (x =
4.07), followed by politics and favoritism [x = 3.92), communications among
underground management personnel (x = 3.85). equipment and maintenance
(x = 3.46). and working and mine conditions [x = 3.23).
1986 Buller and Bell 323

To summarize, the data obtained through structured interviews indi-

cated that shift bosses and hourly employees generally liked and accepted
the interventions, although we noted some negative reactions. Further, results
from the questionnaire showed that employees perceived team building as
having improved the specific problems it had identified. Obviously, tbere
are problems witb tbe validity and reliability of tbese qualitative findings.
However, tbey provide some additional evidence consistent with positive
outcomes of the interventions.

DISCUSSION
Altbougb our design was an improvement over past researcb in many
respects, the results of this field experiment were not conclusive. Even
wben results are in tbe predicted direction, we cannot say witb certainty
tbat goal setting or team building produced these effects in performance or
strategy development. Tbe sbortcomings of tbis experiment illustrate tbe
many difficulties encountered in doing field researcb.
Tbe main problem in interpreting our results is tbe fact tbat we could
not ensure that experimental and control groups were equivalent. Differ-
ences between groups owing to selection, maturation, or instrumentation
could therefore bave influenced tbe findings. We tried several, but cannot
be certain tbat any statistical procedure we employed accounted for all po-
tential biases in our results. As Cocbran and Rubin (1973) pointed out, "If
randomization is absent, it is virtually impossible in many practical circum-
stances to be convinced tbat tbe estimates of the treatments are in fact
unbiased" (1973: 417). Cook and Campbell (1979) suggested tbat, in interpre-
ting the results of sucb quasi experiments, it is important to consider tbeir
consistency witb previous researcb. Tbe findings of previous research on
goal setting are quite clear; hard, specific goals, if accepted, lead to improved
task performance. Tbe results of tbis experiment, altbough in tbe positive
direction, are not as clear or dramatic as tbose of previous studies.
Several factors relating to tbe researcb design and setting may belp to
explain our relatively weak results. First, tbe small number of stopes in tbe
analysis was a limitation, as was the high degree of variability in the perfor-
mance data. Botb limitations make it difficult to detect systematic variance.
In addition, tbe experiment was conducted with unionized bard-rock miners—
subjects wbo were skeptical of modern management tecbniques. As two
employees remarked in tbe postexperimental interviews, tbere was some
concern that tbe goal-setting intervention was an attempt by management to
boost productivity and tben adjust the incentive pay scales downward. If
tbis sentiment was widespread, it may bave caused some miners to restrict
their output. Although miners in the goal-setting condition indicated a high
level of goal acceptance, we cannot be certain that tbis acceptance was real
because of problems witb the measure of acceptance used in tbe study.
Specifically, our measure of goal acceptance consisted of a single, face-valid
question whose validity bas been questioned because goal-setting partici-
324 Academy of Management /ournaJ June

pants tenci to respond positively to it with little variance in their responses

(Locke et al., 1981). And. even though employees expressed generally posi-
tive reactions to the goal-setting intervention in the structured interviews,
reactive effects of the interview itself may have colored their responses.
Thus, the weak results may simply reflect miners' relatively weak accep-
tance of or commitment to the goals set. Finally, the nature of the miners'
task and working environment may have contrihuted to the minimal results.
Hard-rock mining is a difficult task in an uncertain environment. Although
the performance of miners is a function of their ability and motivation just as
it is in other johs, events beyond their control can influence their output,
perhaps more so than in many other blue-collar jobs. In such an environment,
it is possible that increases in miners' efforts and strategy development did
not translate into improved performance.
Yet, these limitations of the study are also its strengths. Previous research
on goal setting has been conducted primarily in laboratories under con-
trolled conditions. Those field experiments that have been conducted have
generally had nonunionized participants performing relatively simple, rou-
tine tasks. As others have pointed out, problems may arise in applying goal
setting in situations in which tasks are complex, unstructured, or interde-
pendent (Baumler, 1971; Latham & Yukl, 1975; Locke & Latham, 1984). The
present study suggests that the effects of goal setting on performance may be
muted under such conditions. Further research in similar settings is needed
to continue to assess the degree to which goal-setting effects can be
generalized.
Also, previous research examining the effects of team building on perfor-
mance has been inconclusive, largely because of poor research designs and
reliance on subjective measures. The present study, employing a more rigor-
ous design and objective measures, also produced ambiguous results.
Although a modest improvement in grade of ore occurred in the team-building
conditions, we cannot be certain why it occurred. Several factors, in addi-
tion to those cited above, may have contributed to the results obtained. First, the
team-building intervention was more diffuse than the goal-setting interven-
tion. Although increased effectiveness in production was one of the objec-
tives of team building, the sessions did not focus specifically on increasing
tons per manshift and quality of ore. The broad nature of this intervention
could have reduced its impact on these output variables, even thougb miners
perceived improvernents on the specific probiems identified by team building.
Related to this issue is the fact that we evaluated performance for only
three-and-a-half months following the intervention. Because team building
involves rather complex and relatively slow processes of developmental
change (French & Bell, 1984; Porras & Berg, 1978), our analysis may not have
captured the full effects of the intervention. A longitudinal assessment is
now being conducted to examine long-term effects.
Two final problems in interpreting the results on productivity concern
possible selectivity bias and reactive effects of the experiment. The mine
19S6 BuHerandBeJl 325

superintendent made the nonrandom assignment of stopes to the team-

building condition primarily because tbose stopes were on tbe two major
production levels of the mine. And, altbougb notbing suggests tbat miners in
tbe team-building condition differed from tbose in tbe non-team-building
condition, the geological conditions of tbe stopes differed between these
treatment groups. Stopes in tbe non-team-building group were primarily at
high elevations in the mine, wbere tbe rock is harder than at low elevations.
Generally, tbe barder tbe rock, tbe faster miners can drill, blast, and muck.
Tbe non-team-building stopes also apparently had ricber veins. These differ-
ent geological conditions may explain tbe preexperimental differences
between treatment groups. Tbe geological differences between stopes were
impossible to control for and make it difficult to compare effects on perfor-
mance between treatment groups. Under tbese conditions, the results of tbe
analyses witbin treatment groups, whicb examined cbanges in performance
over time, are the most informative.
The reactive effects of tbe experiment may also have caused problems.
As mentioned, employees in tbis experiment were not used to being studied
by people otber than "efficiency experts." Although tbere is no strong
evidence, some miners may have suppressed tbeir output out of suspicion of
the reseaicbers' motives. On tbe other band, the modest performance effects
found may have been due to miners' increased motivation resulting from tbe
so-called Hawthorne effect. Tbe structured interviews and informal observa-
tions revealed tbat many miners felt positive about, and even enjoyed, all tbe
attention from the researchers and mine management. These possible Haw-
thorne effects do not overly concern us because one of tbe causal elements of
most bebavioral science approaches to improving productivity, particularly
an approacb like team building, is increased attention and focus on employ-
ees and tbeir work. In fact, we argue tbat tbe objective of many such interven-
tions is a sustained Hawthorne effect.
Finally, we cannot entirely rule out tbe possibility tbat competition
between groups or diffusion of the treatments between groups affected the
results. Tbe interventions were public in tbe sense tbat employees knew
wbo received team building or goal setting. Altbough there is no evidence to
suggest this, it is possible tbat rivalry could have emerged between tbe baves
and the bave-nots. Diffusion of the interventions also could bave occurred,
but tbis was not very likely, particularly in the team-building condition.
Miners in tbe control group could have set personal goals as a result of tbe
intervention, but there is no evidence tbat tbey did. Only tbose miners in the
goal-setting condition received weekly performance-feedback graphs on tons
per manshift and grade of ore to minimize possible contamination of tbe
no-goal-setting groups. However, all miners received bimonthly information
on tons per mansbift on tbeir paychecks; they did not, however, get feedback
on grade of ore. Tbus, miners in tbe goal-setting condition received more
frequent and specific feedback tban otber miners.
The present research produced some information, consistent with previ-
ous research, regarding the cognitive effects of goal setting. Miners who set
326 Academy of Management Journal

goals and received periodic feedback appeared to develop and apply better
strategies for improving the grade of ore they produced than miners in the
control group. Again, the limitations cited for the productivity variables
make firm conclusions unwarranted. It is also important to note that any
improvements in grade strategy owing to the effects of goal setting were not
translated into increased grade quality. The weak effects of strategy develop-
ment or conditions beyond the miners' control affecting grade of ore may
have been responsible for this lack. The study suggests, however, that cogni-
tive mechanisms as well as motivational mechanisms can affect the relation-
ship between goal setting and task performance. Employees who set specific
hard goals and receive feedback on them may not only work harder than
employees who do not set goals and receive feedback—they may also develop
more appropriate ways to accomplish task goals. Team building also appeared
to have a positive effect on strategy development, consistent witb Hackman's
(1976) model. And, even though strong conclusions cannot be drawn, it is
interesting to note that team building may have influenced the development
of strategies which in turn may have improved quality of grade.
In summary, although the findings of this field experiment are unclear,
the results are important for several reasons. First, the study raises a question
about the extent to which goal-setting effects are generalizable to union
employees working on complex and uncertain tasks. Further research under
similar conditions is warranted. Second, the study suggests that goal setting
and team building may affect cognitive mechanisms such as strategy
development. Future research should continue to examine this mechanism
using more refined measures. Third, the study points to our continuing lack
of understanding of how team building influences performance. We attempted
to operationally define this intervention more clearly and to examine its
effects more systematically than previous research has done, and yet its
effects remain elusive. Further research should attempt to document team-
building activities and possible intervening mechanisms more precisely and
examine their relation to outcome variables. Perhaps path analysis or other
causal analysis techniques (James, Muliak, & Brett, 1982) might provide
useful means for assessing the multiple relationships generated by the team-
building intervention. Finally, declining productivity continues to be a major
problem in underground mines and other organizations, and managers con-
tinue to be beseiged with a large number of quick-fix solutions from behav-
ioral scientists and others. Unfortunately, many approaches to improving
employees' productivity remain untested. The present research, by provid-
ing a systematic evaiuation of two specific techniques for improving the
performance of employees, provides managers with a more objective basis
on which to judge the usefulness of team building and goal setting.

I REFERENCES
Bandura, A.. & Simon, K.M. 1977. The role of proximal intentions in self-regulation of refractory
behavior. Cognitive Therapy and Researcb, 1; 177-193.
1986 BuIJer and Bell 327

Baumler, [.V. 1971, Defined criteria of performance in organizational control. Administrative

Science Quarterly, 16: 340-350.
Bavelas, J.B.. & Lee. E.S, 1978.Effects of goal level on performance: A trade-off of quantity and
quality. Canadian Jaumal of Psychology, 32; 219-240.
Beer. M. 1976. The technology of organization development. In M.D. Dunnette (Ed.), Handhook
of industrial and organizational psychology: 937-993. Chicago: Rand McNally & Co.
BuHer, P.F, 1982. The effects of team building-problem solving and goal setting-performanve
feedback on the productivity of underground metal miners: A field experiment.
Unpublished doctoral dissertation. University of Washington, Seattle.
Buller. P.F. 1983. Team building-problem solving and performance: Conceptual issues and a
theoretical model. In D. F. Ray (Ed.), Applying systematic knowledge to problems af
organization and management: 288-290. Mississippi State University: Southern Manage-
ment Association.
Campbell, D.T.. & Stanley, J.C. 1963. Experimentai and qaasi-experimental designs for
research. Chicago: Rand McNally & Co.
Christensen-Szalanski, M,], 1980. A further examination of the selection of problem-solving
strategies: The effects of deadlines and analytic aptitudes. Organizational Behavior and
Human Performance, 25: 107-122.
Cochran, W,C.. & Rubin. D.B. 1973. Controlling bias in observational studies: A review, Sankhya,
35: 417-446,
Cook, T.D,. & Campbell. D.T, 1979, Quasi-experimentation: Design and analysis issues for
fieid settings. Boston: Houghton-Mifflin.
Cummings, T,C,, Molloy. E.S,. & Glen, R,H. 1974, Intervention strategies for improving produc-
tivity and the quality of work life. Organizational Dynamics, 4: 52-68,
DeMeuse. K.P,. & Liebowitz, S.J. 1981, An empirical analysis of team building research. Group
and Organizational Studies, 6: 357-378.
Dyer, W.G. 1977. Team building: Issues and alternatives. Reading, Mass.: Addison-Wesley.
French. W,D,.&Be!l, CH, 1984, Organization development: Behavioral science interventions
for organization improvement (3rd ed.). Englewood Cliffs, N.J,: Prentice-Hall.
Hackman, ),R. 1976. Group influences on individuals in organizations. In M.D. Dunnette (Ed.),
Handbook of industrial and organizational psychology: 1455-1525. Chicago: Rand
McNally & Co.
Huse, E,F. 1980. Organization development and change (2nd ed.). St. Paul, Minn.: West
Publishing Co.
James, L.R., Muliak, S.A., & Brett, J.M. 1982. Causal analysis: Assumptions, models, and data.
Beverly Hills, Calif.; Sage Publications.
Latham, G,P., & Baldes, J.J, 1975. The "practical significance" of Locke's theory of goal setting.
Journal of Applied Psychology, 60: 122-124.
Latham, G.P,, & Steele, T,P, 1983, The motivational effects of participation versus goal setting on
performance. Academy of Management fournal, 26: 406-417.
Latham, G,P., 8t Yukl, G.A. 1975. A review of research on the application of goal setting in
organizations. Academy af Management /ournal, 18: 824-845.
Locke. E,A, 1968, Toward a theory of task motivation and incentives. Organizational Behavior
and Human Performance, 3: 157-189.
Locke, E,A., Feren, D.B., McCaleb, V,M,, Shaw, K.N,, & Denny, A.I, 1980. The relative effective-
ness of four methods of motivating employee performance. In K,D, Duncan, M.M. Gruneberg,
& D, Wallis (Eds,), Changes in working life: 363-388. New York: John Wiley & Sons.
328 Academy of Management /ournal June
Locke, E.A., St Latham, G.A, 1984. Goal setting; A motivational technique that works. Englewood
Cliffs, N.J,: Prentice-Hall,
Locke, E.A., Shaw, K.N,. Saari. L,M,, & Latham, G,P, 1981, Goal setting and task performance:
1969-1980, Psychological Bulletin, 90: 125-152.
Marguiiesp N,, & Raia, A.P, 1968, People in organizations: A case for team training. Training and
Development Journal, 22(8): 2 - 1 1 .
Naylor, J.C., Pritchard, R.D., & Ilgen, D.R. 1980, A theory of behavior in organizations. New
York: Academic Press.
Nicholas, J.M, 1982, The comparative impact of organization development interventions on
hard criteria measures. Academy o/Afanagement AeWeiv, 7: 531-542.
Porras, J.I,, & Berg, P.O, 1978, Impact of organization development. Academy of Management
Review, 3: 249-266.
Reichardt. G.S. 1979. The statistical analysis of data from nonequivalent group designs. In T,D.
Cook & D,T. Campbell (Eds.), Quasi-experimentation: Design and analysis issues for field
settings: 147-205. Boston: Houghton-Mifflin.
Shaw, K.N, 1983, Tbe relationships among goals, strategies, and task performance. Working
paper. University of Kentucky, Lexington.
Taylor, F.W, [1911] 1967. The principles of scientific management (Reprint) New York: Norton.
Terborg, J.R. 1976, The motivational components of goal setting. Journal af Applied Psychology,
61: 613-621.
Terborg, J.R., & Miller, H.E. 1978. Motivation, behavior, and performance: A closer examination
of goal setting and monetary incentives. Journal of Applied Psychology, 63: 29-39.
Woodman, R.W.. & Sherwood, J.J, 1980, The role of team development in organizational
effectiveness: A critical review. Psychological Bulletin, 88: 166-186.

Paul Buller earned his Ph.D. degree at the University of Washington: he is an assis-
tant professor of management in the Department of Business Administration at Utah
State University, His current research interests include productivity improvement,
organizational change, and strategic management.
Cecil Bell earned his Ph.D, degree at Boston University: he is an associate professor of
management and organization in the Graduate School of Business Administration at
the University of Washington. His current research interest is organization develop-
ment.
View publication stats