Educational Evaluation and

Policy Analysis http://eepa.aera.net

Principals' Instructional Leadership and School Performance: Implications for Policy

Development
Ronald H. Heck
EDUCATIONAL EVALUATION AND POLICY ANALYSIS 1992 14: 21
DOI: 10.3102/01623737014001021

The online version of this article can be found at:

http://epa.sagepub.com/content/14/1/21

Published on behalf of

American Educational Research Association

and

http://www.sagepublications.com

Additional services and information for Educational Evaluation and Policy Analysis can be found at:

Email Alerts: http://eepa.aera.net/alerts

Subscriptions: http://eepa.aera.net/subscriptions

Reprints: http://www.aera.net/reprints

Permissions: http://www.aera.net/permissions

Citations: http://epa.sagepub.com/content/14/1/21.refs.html

>> Version of Record - Jan 1, 1992

What is This?

Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014

 Educational Evaluation and Policy Analysis
Spring, 1992, Vol. 14 No. 1, pp. 21-34

Principals' Instructional Leadership and School Performance:

Implications for Policy Development

Ronald H. Heck
University of Hawaii at Manoa

Currently, many districts and states are pursuing reforms that focus on holding principals
accountable for school performance. While effective schools research has established that strong
principal leadership affects school academic achievement at least indirectly, this relationship is
more complex than originally thought. Personnel decisions about principal effectiveness made
by educational policymakers or lay-controlled school site councils should be made only after
careful consideration of research on the relationship between principal instructional leadership
and school outcomes. This article presents data from a study to determine whether principal
instructional leadership is predictive of school outcomes and discusses the implications of the
research for developing school improvement strategies.

For the past decade, policymakers have pur- ited participation in their children's
sued a variety of educational reforms to in- education.
crease student learning. Some have focused Increasingly, district administrators, in-
on the design and delivery of curriculum and cluding principals, are being held account-
instruction, including course content, stan- able for schools' performance. For example,
dards and expectations, and teaching tech- the Oklahoma Board of Education has de-
niques. Others have been directed at altering cided to oversee at-risk elementary schools,
school organization, creating greater ac- closing down troubled schools briefly and
countability, and enhancing school-based opening them with new staffs and principals
leadership. The public's demands for educa- ("Oklahoma Board Votes," 1991). In New
tional accountability have advanced the use Jersey, the education department moved to
of achievement data to evaluate instructional seize control of the Paterson district, granting
efforts, because of concerns about poor edu- the new superintendent sweeping powers to
cational outcomes in many schools and the reorganize the district and replace inferior
perception that America is declining as an principals ("New Jersey Officials," 1991). In
economic power. The effective schools re- Chicago, local site councils composed mainly
search has been a driving force behind politi- of parents and community representatives
cal efforts to improve public education, sug- control the hiring and firing of principals
(Wong & Rollow, 1990).
gesting that improved student outcomes can
be attained through strategic school organi- Ideally, personnel decisions made by edu-
zation and strong principal leadership. The cational policymakers or by lay-controlled
assumption is that improved student site councils that may affect a principal's job
security should be made only after careful
achievement is possible despite a variety of
consideration of the research on the relation-
environmental constraints such as low par-
ship between principal leadership and school
ent socioeconomic status, diverse language outcomes. Although effective schools re-
backgrounds of students, and parents' lim- search has established that strong principal
21
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
Heck
leadership influences school academic • What are the most important instruc-
achievement at least indirectly (Andrews & tional leadership predictors of school
Soder, 1987; Bossert, Dwyer, Rowan, & Lee, achievement?
1982; Hallinger & Murphy, 1987), this rela- • To what extent do the instructional lead-
tionship is more complex than originally ership profiles of new principals in consis-
thought. Previous research still has not tently high- and low-achieving schools tend to
clearly defined instructional leadership or mirror the norms for principal behavior in
provided enough empirical evidence to dem- each type of school?
onstrate that principals who increase the
amount and quality of time they devote to These questions are examined using data
instructional leadership produce higher aca- from a sample of 31 elementary and 25 high
demic performance in their schools (Deal, schools in California.
1987; Lee, 1987). Only rarely has instruc-
tional leadership been defined in studies as Conceptual Approach
specific policies, practices, and behaviors ini- Past research on principal instructional
tiated by the principal (Hallinger & Murphy, leadership has been problematic as a basis for
1987). Therefore, the types of activities that accurately modeling the relationship be-
are important for consideration as an effec- tween a school's environmental and social
tive instructional leader remain unclear. contexts and the dimensions of instructional
Wimpelberg, Teddlie, and Stringfield leadership that may be associated with higher
(1989) note that future research on principals or lower school academic performance
must attend not just to general characteristics (Heck, Larsen, & Marcoulides, 1990; Wim-
of behavior or attitudes (e.g., "has a vi- pelberg et al., 1989). That the context of
sion"), but also to the specific job tasks of the schooling shapes students' classroom learn-
role. It is important to determine how princi- ing experiences (Barr & Dreeban, 1983;
pals allocate their time in more or less effec- Dreeban & Barr, 1988; Oakes, 1989)—in-
tive schools and within varying contextual cluding the allocation of resources to those
situations. Before valid and reliable evalua- classrooms (Bossert, 1988; Heck & Mar-
tion of principal performance can be imple- coulides, 1989)—is well established. Teddlie,
mented, instructional leadership must be Kirby, and Stringfield (1989), for example,
linked to improved school performance. identified differences in classroom teaching
Once these relationships are understood, practices in effective versus ineffective
one may evaluate not only the process of schools. These classroom differences appear
principal leadership, but the products as well related in part to principal behavior. What is
(Pitner & Hocevar, 1987). less clear, however, is the specific manner in
This article examines the case for principal which principals as school leaders may also
effects on school achievement outcomes and contribute to the linkage among school con-
suggests guidelines for the design of school text, school variables, and student achieve-
improvement strategies. The theoretical ap- ment. The link between principal behavior
proach taken rests on the belief that leaders and school outcomes is at best indirect
are only sometimes "masters of their fates." (Boyan, 1988; Heck et al., 1990). Activities
For educational reforms to be fully imple- such as decision making, communicating
mented at the school level, principals must goals and strategies to others, gatekeeping
not only be capable of providing strong lead- with parents and community interests, and
ership when required but must also under-
monitoring classroom work activities have
stand how the larger environment shapes
trickle-down effects that nurture student-
their organizational and individual interac-
level performance (Heck et al., 1990; Teddlie
tions and the relationship between those in-
etal., 1989).
teractions and resulting school outcomes.
The research questions addressed in this Consequently, principal instructional lead-
study focus on two aspects of the link be- ership is assumed to depend on the princi-
tween principal instructional leadership and pal's own beliefs and value preferences and
school performance: on organizational and political variables asso-
22
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
 Principals' Instructional Leadership

ciated with the school and community con- studies of secondary school principals, few
text (e.g., district size, level of schooling, findings from studies of elementary schools
students' socioeconomic and language back- have been validated at the secondary level,
grounds, pressures from the district office, and there are specific differences in the con-
community, and staff; Bossert et al., 1982; text of secondary schools. Because of the
Heck & Marcoulides, 1989; Wimpelberg et multiple and competing goals of secondary
al., 1989). Thus, contextual factors may schools (Farrar, Neufeld, & Miles, 1984), and
sometimes constrain and shape the princi- conditions limiting the joint work among
pal's exercise of instructional leadership teachers, some researchers argue that nei-
(Hallinger & Murphy, 1987; Heck & Mar- ther administrative leaders nor teacher
coulides, 1989). Because much of the early leaders could exert much influence on in-
research on school effects typically centered struction (Cusick, 1983). Other researchers,
on a narrow sample of schools (viz., urban however, tend to support the view that princi-
elementary schools in large districts; Hal- pals at both organizational levels may draw
linger & Murphy, 1987; Rowan, Bossert, & upon a similar repertoire of activities in car-
Dwyer, 1983), the influence of a wide spec- rying out the functions of their role effec-
trum of organizational conditions on the prin- tively (Dwyer, Lee, Rowan, & Bossert, 1983;
cipal's instructional leadership behavior re- Lee, 1987).
mains largely unknown. Another major problem with conducting
School level is one prominent, but poorly research on school processes has been that
understood, contextual variable that has measures of school demographic composi-
been postulated to influence the instructional tion, organization, school effects such as cli-
leadership role of the principal considerably mate, and achievement are all correlated
(Firestone & Herriott, 1982; Hallinger & (Rowan et al., 1983). Because of the com-
Murphy, 1987). A number of studies (e.g., plexity of these various interrelationships, it
Hallinger & Murphy, 1987; Herriott & is important to control for the effects of
Firestone, 1984; Sizer, 1984) suggest that "hard-to-change" demographic composition
there may be specific differences between the variables when attempting to estimate the
secondary and elementary contexts of influence of school processes on achieve-
schooling with respect to size, structure, link- ment. Failure to provide adequate controls
ages to parents, student characteristics, cur- may yield results that are, in actuality, spu-
ricular organization, faculty characteristics, rious (Rowan et al., 1983).
and principal leadership. As Murphy (1988) As Murphy (1988) concludes, a major rea-
argues, the extent to which instructional son for the problem is that researchers have
leadership styles of secondary school princi- provided limited strategies for studying in-
pals will mirror the strong, direct role often structional leadership. For example, studies
exhibited by effective elementary school have often examined correlates of principal
principals is uncertain. This assumption leadership and student achievement in isola-
about differences in the principal's role is tion. Such univariate analyses, however, of-
probably correct, because the requirements ten are difficult to interpret, because they fail
and demands of leadership are confounded to consider intercorrelations among predic-
and compounded by school size, curriculum tor variables or increases in Type I errors that
complexity, and the scale of administrative can result (Tatsuoka & Silver, 1988). Conse-
obligations (Little & Bird, 1987). Such con- quently, the use of multivariate models is a
textual differences may result in variation in necessary activity in building or testing the-
principals' instructional leadership behavior ory, yet this type of data analysis has been
and, in turn, in how this leadership may affect limited (Willower, 1987).
school outcomes. A final aspect of principal leadership with
Hallinger and Murphy (1987) also caution important implications for policy and prac-
not to generalize findings from previous re- tice is the extent to which new principals may
search on elementary contexts to the second- be able to make changes that will lead to
ary level because there is a lack of systematic school improvement, especially in poorly
23
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
Heck

performing schools. Ogawa (1991), for exam- than expected given the socioeconomic sta-
ple, in reviewing the literature on leadership tus, ethnicity, and language dominance of
succession, concludes that successors have a their students (Hallinger & Murphy, 1986).l
tendency to adjust to existing organizational Criteria for participation in the study in-
norms of behavior. In his case study of princi- cluded consistent performance above or be-
pal succession, Ogawa suggests that the fac- low a school's comparison band on both the
ulty members, although hopeful of positive third- and sixth-grade scores in reading and
school change before the replacement began, math (for elementary) or the twelfth-grade
were worried that the new principal would scores in reading, math, and language (for
interfere with their classroom practices. At high school). Eighty-five elementary and 33
the end of the first year, however, teachers high schools met the criterion of consistent
agreed that most of the changes made by the school academic performance above or below
principal were insignificant and did not inter- that expected within their comparative band
fere with their teaching. Similarly, in a review over the 3-year period.
of literature on leadership succession and so- The requirement that the principal and at
cialization, Hart (1991) argues that organiza- least 4 teachers had to return the question-
tions protect against the intrusion of new naire resulted in a sample of 56 elementary
members through formal and informal social and high schools (with a mean of 6 respon-
mechanisms. As Louis and Miles (1991) note dents per school). Additionally, for the pre-
in a study of urban high schools, however, liminary analysis, which focused on the in-
those principals whose schools made the structional leadership variables that were
most improvement in performance focused most predictive of school achievement, the
on changes in teachers' classroom behavior principal and teachers had to be present in
(e.g., classroom management, more individ- the school for the entire 3 years of the study
ualized help for students, increased time on (to control for the possible changes in the
task) and the school as an organization (in- behavior of new principals with respect to
creased orderliness and improved relation- solving school problems). This additional cri-
ships among teachers), had broad support for terion resulted in a sample of 23 elementary
the specific change effort, and sustained the (15 high achieving) and 17 high schools (7
work for longer periods of time. high achieving) for the first analysis.
A second sample of 8 elementary (4 high
Study Methods achieving) and 8 high schools (4 high achiev-
Sample ing) having new principals was also devel-
The sample included all public elementary oped. This sample (although limited) was
and high schools in California that had used to explore the question of whether new
achieved above or below their "comparison principals use patterns of instructional lead-
band" test scores for 3 consecutive years ership that are similar to or different from the
(1984-1986) as measured by the California more experienced principals in schools of like
Assessment Program (CAP). To establish a achievement characteristics.
school's comparison band of similar schools,
differences in students' socioeconomic status Instrumentation and Procedure
and language background are standardized. A questionnaire was administered to the
Results from the CAP test place each principal and a random selection of 6
school's scores in a comparison band that teachers in each elementary and high school
compares its scores with those obtained by whose academic achievement was stable over
other California schools composed of stu- the 3-year period. The questionnaire was de-
dents from similar social backgrounds. signed, consistent with previous research
School achievement may then be compared (e.g., Andrews & Soder, 1987; Hallinger &
among schools, after the effects of several Murphy, 1987), to measure 22 strategic inter-
"hard-to-change" contextual variables have actions (job tasks) between principals and
been controlled. Schools that score above teachers in terms of the principal's instruc-
their comparison band are performing better tional leadership role in (a) governing the

24
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
 Principals' Instructional Leadership

school, (b) developing school climate, and utive years) elementary and high schools and
(c) organizing and monitoring the school's low-achieving (below their comparison band)
instructional program. The three task do- elementary and high schools. This was done
mains of instructional leadership were devel- to investigate the extent to which principal
oped on the basis of Boyan's (1988) sugges- instructional leadership differs by achieve-
tion that current theoretical models of ment and school level. Exploratory discrimi-
administrator effects indicate that principals nant analyses were then conducted to iden-
can influence the governance and work struc- tify an optimal subset of the original 22
tures of the school and Bossert et al.'s (1982) instructional leadership tasks that would best
theoretical model of instructional leadership, classify the sample into the desired groups.
which suggests that principals can influence Mean scores on the instructional leader-
the school's climate and instructional organi- ship activities aggregated to the school level
zation. The internal consistency coefficients were used as the classifying variables. The
for the domains of leadership were high (i.e., means and standard deviations for each
with alpha coefficients ranging from .8 to .9). group on the eight instructional leadership
To assess the degree of implementation of variables retained in the analysis are pres-
each activity (e.g., ensures that instructional ented in Table 1.
goals are clearly communicated to everyone, The underlying structure to the classifica-
communicates high expectations for student tion was examined first. Only one significant
academic performance to staff), participants discriminant function (canonical correlation
were asked the following: "Please indicate = .83) was calculated for the final set of
the degree to which each of the following instructional leadership predictors used in
activities is implemented by the principal (or the analysis, with x2 (24) = 53.03, p < .001.
designee) at your school." Each of the vari- The canonical correlation coefficient indi-
ables was constructed as a 5-point Likert- cated that the variables in the analysis have
type scale with responses ranging from never strong predictive power (i.e., the function
to always. The questionnaire took approx- accounted for 81% of the variance in the
imately 10 to 15 minutes to complete. classification of the groups).
The standardized canonical discriminant
An Analysis of Principal function coefficients are presented in Table
Instructional Leadership 2. The standardized function coefficients
A discriminant analysis was performed to permit the comparison of the magnitude of
determine which principal instructional lead- each predictor, regardless of how it was mea-
ership interactions would best classify low- sured, in classifying the groups by the princi-
and high-performing elementary and high pal's instructional leadership behavior. Note
schools. The goal of such analyses is to find a that high scores on making regular classroom
linear combination of variables that maxi- visitations (.55), encouraging the discussion
mizes the differences among groups in the of instructional issues (.34), reducing admin-
sample and therefore allows efficient group istrative interferences in the classroom (.33),
membership prediction. The achievement and emphasizing test results for program im-
levels of elementary and high schools may provement (.26) appear to dominate the clas-
then be predicted according to these optimal sification of these schools. From Table 1 it can
characteristics of their principals' instruc- be seen, for example, that principals in low-
tional leadership profiles. achieving elementary schools are less likely
(M = 3.4) to make regular classroom visita-
Important Leadership Predictors tions than principals in high-achieving ele-
of School Achievement mentary schools (M = 4.2). The means fur-
The primary sample (where principals ther indicate no differences between
were present at least the 3 years of the study) principals in high- and low-achieving second-
was first divided into four groups of schools ary schools on this variable (both Ms = 3.1).
(N = 40): high-achieving (above their com- So, while this variable separates high-achiev-
parison band in all CAP subtests for 3 consec- ing and low-achieving elementary schools, it
25
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
Heck

TABLE 1
Average Ratings of Principal Instructional Leadership by Level of Student Achievement

Reported extent of implementation of behavior 3

Low achieving High achieving
Elementary High Elementary High
(n = 8) (n == 10) (n = 15) (n == 7)
Leadership task M SD M SD M SD M SD
The principal:
Makes regular classroom visits 3.4 (0.4) 3.1 (0.3) 4.2 (0.5) 3.1 (0.7)
Promotes discussion of 3.5 (0.5) 3.2 (0.4) 4.3 (0-4) 3.4 (0.6)
instructional issues
Minimizes classroom 4.3 (0-6) 4.1 (0-4) 4.1 (0.4) 4.2 (0.3)
interruptions so teachers can
teach
Emphasizes test results for 4.1 (0.6) 3.6 (0.4) 4.5 (0.5) 3.9 (0-5)
program improvement
Participates in discussion about 3.6 (0.5) 3.4 (0-3) 4.3 (0-5) 3.6 (0.5)
how instruction affects student
achievement
Ensures systematic monitoring 3.9 (0.5) 3.5 (0.3) 4.5 (0.4) 3.7 (0.7)
of student progress
Communicates instructional 4.4 (0.3) 3.9 (0.4) 4.7 (0-4) 4.0 (0.6)
goals to others
Protects faculty from external 3.4 (0.6) 3.2 (0.3) 4.2 (0.4) 3.7 (0-6)
pressures
1
1 = low, 5 = high.

also helps separate the secondary and ele- among the predictors that are most highly
mentary contexts, suggesting that high associated with the discriminant function are
school principals, in general, spend less time also displayed in Table 2. These coefficients
in classrooms observing instruction than provide another means of assessing how well
their elementary school counterparts. the variables in the model predict group
The correlations (structure coefficients) membership. Variables that correlate highly

TABLE 2
Standardized Canonical Discriminant Function Coefficients and Structure Coefficients for Classifying
Schools by Achievement and Level

Standardized
discriminant
function Structure
Leadership task coefficient coefficient
Makes regular class visits .55 .77
Promotes discussion of instructional issues .34 .74
Minimizes class interruptions .33 .11
Emphasizes test results .26 .62
Participates in discussion about how instruction affects achievement .11 .60
Ensures systematic monitoring of student progress .06 .74
Communicates instructional goals .03 .84
Protects faculty from external pressures .01 .49

26
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
 Principals' Instructional Leadership

with the discriminant function best define the schools (a more significant error in classifica-
underlying dimension. Note that all coeffi- tion). In fact, if one collapses the four pre-
cients in the set are high, except for reducing dicted groups into only two predicted groups
administrative interruptions in classrooms. (by high and low achievement), 90% (36
Caution should be exercised in interpreting of 40) of the schools would be correctly
these coefficients, however, because the cor- classified.
relations alone do not indicate how heavily a A plot of the group centroids (i.e., the
variable contributes to the discriminant func- means of the discriminant scores) for each
tion after adjusting for the remaining vari- group in Figure 1 emphasizes the utility of
ables. For purposes of analysis, correlations the instruction leadership variables in classi-
above .3 may be considered sufficient in in- fying schools into the four groups. As shown
terpreting the discriminant function (Ta- in Figure 1, the plot nicely depicts the multi-
bachnick & Fidell, 1983). variate "space" between the groups, showing
Using the discriminant dimension, high or the relative distance each lies from the "0"
low school achievement as a function of the point (i.e., mean) of the discriminant func-
set of discriminatory variables may be pre- tion. The plot reveals that low secondary
dicted. The results of the analysis to classify (-1.36) and high elementary (1.48) schools
high- and low-achieving schools according to differ the most in terms of the instructional
principal instructional leadership behavior leadership behavior of their principals.
are presented in Table 3. Percentage correct This plot may suggest that the profile of
and incorrect is used as a measure of the instructional leadership in high-achieving ele-
accuracy of the variables in predicting mem- mentary schools is somewhat different from
bership in one of the four groups under con- the other groups of schools in terms of the
sideration. For purposes of this analysis, 70% implementation of these specific instructional
accuracy was used as the criterion of accept- leadership tasks. For example, the means on
able group membership prediction (against the instructional leadership activities in Table 1
25% group prediction by chance). The classi- are generally higher for this group of schools
fication run resulted in 73% of the schools in than the means of the other groups. The results
the sample being correctly classified. Table 3 imply, therefore, some contextual differences
also indicates that high-achieving high in principal leadership that are associated with
schools appear to be the most difficult group school level, but these appear to be less impor-
to classify correctly (below 50%) according tant than the differences associated with the
to the reported instructional leadership be- performance level of the school.
havior of the principal. That is, two high-
achieving high schools had the same leader- New Principals' Instructional
ship pattern as high-achieving elementary Leadership and Achievement
schools (not a significant error), and two had A separate analysis was also performed on
the same pattern as low-achieving high the remaining 16 schools (8 elementary and 8

TABLE 3
Schools Correctly Classified as High and Low Achieving by Principal Instructional Leadership

Predicted group
LE LS HE HS
Actual group No. % No. % No. % No. %
Low elementary (LE) 6 75 1 12.5 1 12.5 0 0
Low secondary (LS) 3 30 6 60 0 0 1 10
High elementary (HE) 0 0 0 0 14 93.7 1 6.7
High secondary (HS) 0 0 2 28.6 2 28.6 3 43
Percentage of grouped cases correctly classified: ; 73%
Note. LS = low secondary; HS = high secondary; LE = low elementary; HE = high elementary.

27
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
Heck
LS US LE HE
(-1.36) (-.91) (-.25) (1.48)

1 • P +-\ 1 • 1
- 2 - 1 0 1 2
FIGURE 1. Instructional leadership implementation in high- and low-achieving elementary and high schools

high schools) that had first- or second-year than what principals themselves say they do.
principals. These schools were also classified In addition, Ebmeier notes that teachers
by achievement to explore the tentative hy- tend to be the most conservative in rating the
pothesis that new principals (especially in frequency of implementation of various prin-
consistently low-achieving schools) would cipal activities. Because teachers have the
employ a different type of school instruc- opportunity to observe more closely what
tional leadership that should eventually lead principals do on a day-to-day basis than su-
to increased school productivity. The hy- pervisors of principals, teachers may be able
pothesis tested was that it should not be pos- to give more valuable (and reliable) informa-
sible to accurately classify these particular tion. This is in sharp contrast to the current
high- and low-achieving schools with new type of principal evaluation, which is gener-
principals on the basis of their principals' ally controlled by supervisors in most dis-
instructional leadership profiles (i.e., results tricts (Educational Research Service, 1985).
would be similar to the 25% chance). Some variation was noted, however, in indi-
Somewhat surprisingly, however, these vidual teachers' observations of the principal
schools were classified with 94% accuracy within each school; of course, this variation
(not tabled), the only error being in one high- disappears when the data are aggregated to
achieving elementary school where the new the school level.
principal's instructional leadership profile Another important issue is that this was an
was similar to principals in low-achieving ele- outlier study, as opposed to a random sample
mentary schools. All of the eight consistently of schools, and therefore identified the best
low-achieving schools (four elementary and and the worst among a few thousand elemen-
four high schools) were correctly classified by tary and high schools. Thus, there are much
performance and level according to the pro- missing data from the "middle" range of
file of the new principal's instructional lead- schools. Further research should be directed
ership behavior. Possible explanations of this toward replicating this study with average
finding are that these latter principals may schools as a third group. The limited number
not have realized that their schools were con- of schools with stable achievement patterns
sistently low achieving in all academic areas, available for study, once important contex-
or they realized the problem but were rela- tual variables are controlled, supports
tively powerless to change their patterns of Rowan et al.'s (1983) contention that the sta-
strategic instructional leadership interactions bility of prolonged school achievement scores
with teachers, students, and the community. is problematic in attempting longitudinal
Several cautions about the ability to gener- studies. Once environmental factors are con-
alize from this study are warranted. The issue trolled, however, the results do demonstrate
of variability in how others view the princi- that principals in elementary and high
pal's role-related activities is an important schools that are extraordinarily high achiev-
one. Views concerning the behaviors the ing, as measured by consistent academic
principal exhibits in the school can differ, achievement in a variety of curricular areas,
sometimes substantially, across groups of re- may be substantially different from their
spondents (Ebmeier, 1991; Heck et al., counterparts in consistently low-achieving
1990). In general, however, Ebmeier sug- schools in terms of the type and effectiveness
gests teachers and supervisors to be in much of instructional leadership they provide.
closer agreement about what principals do These differences tend to be perpetuated de-
28
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
 Principals' Instructional Leadership

spite principal turnover within this particular The results of this study and others that
group of outlier schools. have focused on the principal as well as the
classroom behavior of teachers provide
Implications for Increasing School needed empirical support for the belief that
Academic Performance school variables, including principal instruc-
This section draws policy implications from tional leadership, are predictive of the
the research findings on the relationship be- school's academic outcomes. The extension
tween principal instructional leadership and of a model of instructional leadership to
school performance. The purpose of this other countries also provides empirical sup-
study was to explore differences in the re- port to a growing database of cross-cultural
ported allocation of instructional leadership studies highlighting the importance of the
tasks between principals in elementary and principal's role in facilitating academic im-
high schools with stable achievement pat- provement in the school. The importance of
terns. More specifically, the goal was to de- these findings is that many of these school
termine whether high or low school perfor- process variables may be manipulated by the
mance in elementary and high schools could school principal. The instructional leader-
be predicted from a set of principal instruc- ship model presented in this study provides a
tional leadership tasks. view taken at one moment in time of the simul-
As Witte and Walsh (1990) argue, given taneous contribution of several important prin-
the complex nature of educational systems cipal instructional leadership activities to the
and difficulties researching them, any results overall prediction of school performance, after
must be limited in helping to resolve major equalizing the effects of "hard-to-change" con-
issues surrounding the effective schools de- textual variables such as socioeconomic status
bate. After controlling for some important and language background.
contextual factors (viz., socioeconomic sta- Several policy implications follow from the
tus, language backgrounds of students), the finding that at least some leadership activities
results indicate an ability to predict of the principal appear to be good predictors
achievement outcomes based on teachers' of school performance. If we hold the envi-
and principals' perceptions of the instruc- ronment constant and measure the princi-
tional leadership activities in the school. pal's instructional leadership at various
This relationship appears to hold for both points in time, the resulting profile of the
high schools and elementary schools in the school's instructional leadership provides op-
sample. tions for changing the desired direction of the
The findings presented in this study sup- organization. An analysis of the instructional
port similar efforts to establish links between leadership profile of the principal in a partic-
what principals do as instructional leaders ular school can provide policymakers with
and school outcomes. For example, in subse- information about the congruency of this
quent analyses of data sets from other coun- profile with desired school goals, instruc-
tries that did not use the outlier strategy of tional strategies, climate, and outcomes.
selecting schools, Heck, Marcoulides, and As previous research has indicated (Blank,
Lang (1991) were able to classify correctly by 1987; Crowson & Porter-Gehrie, 1980; Eb-
achievement 77% of the sample schools ac- meier, 1991; Leithwood & Montgomery,
cording to a similar model of principal in- 1982), typical principals ignore the instruc-
structional leadership, although some of the tional planning strategies of teachers in all
specific leadership predictors differed. Heck but exceptional circumstances. Principals are
(1991) also was able to classify correctly 77% often seen as more effective in dealing with
of a random sample of high- and low-achiev- external constituents and receive low marks
ing secondary schools in Singapore according from teachers for their ability to monitor the
to the school's climate, teacher expectations, school's instructional program (Blank,
and the instructional leadership profile of the 1987). The present analysis indicates, how-
principal. ever, that three instructional leadership pre-
29
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
Heck
dictors within the domain of instructional or- goals, communicate those goals to others,
ganization are important in classifying the and translate the goals into effective instruc-
groups of schools. The three that tend to tional strategies to improve outcomes.
dominate (after adjusting for predictor inter- Contextual differences that may affect the
correlations) are the amount of time principals exercise of instructional leadership, however,
spend directly observing classroom practices, should also be noted in developing school
promoting discussion about instructional is- improvement strategies. Some differences in
sues, and emphasizing the use of test results leadership appear to occur across the context
for program improvement. of the school's level. The greatest differences
Policy implications resulting from this find- in the profiles of instructional leadership
ing include having districts emphasize a view tasks according to the set of predictors in this
of the principalship that promotes a strong study are between principals in high-achiev-
administrative role in the area of instruction ing elementary schools and low-achieving
and is evaluation based (Glasman & Nevo, high schools (see Figure 1). Consistent with
1988). Increasing principals' expertise as Virgilio, Teddlie, and Oescher's (1991) study
clinical supervisors, as well as the amount of of teaching practices in differentially effec-
time principals allocate to this activity, ap- tive elementary and junior high schools, it is
pears to be one policy choice that pays divi- apparent from the results that secondary
dends in terms of school performance, espe- school principals do not allocate the same
cially within the elementary context. amount of time to most of the instructional
Although previous research has indicated leadership tasks in the model, whether in
that typical principals allocate very little of high- or low-achieving schools. These differ-
their time to activities aimed at improving ences related to school level need to be studied
teachers' teaching skills, or to providing further. Principals in high-achieving elemen-
teachers with the desired quantity and qual- tary schools appear to devote substantially
ity of feedback on their teaching after class- more time to the implementation of instruc-
room observation (e.g., Ebmeier, 1991),
tional leadership activities than principals in
principals in very high-achieving elementary
any other setting.
schools in this study are perceived as allocat-
ing more time to these activities. This result is Another important finding from the study,
consistent with Heck et al. (1991), who found reported cautiously because of the limited
that the amount of time principals spent in sample of schools available, is that new ele-
classrooms and the quality of their feedback mentary and high school principals did not
to teachers were important predictors of employ a type of instructional leadership that
school achievement in a different cultural differed from the pattern of principals in
context. School personnel need to under- other low- or high-achieving schools. The
stand attributes of effective teaching prac- sample of schools having new principals was
tices and develop strategies to best monitor classified with almost perfect accuracy using
classrooms (Rowan et al., 1983; Teddlie et the same classificatory set of instructional
al., 1989). leadership variables. There are several possi-
The results also underscore the need for ble interpretations of this result that chal-
districts to emphasize the principal's role as lenge researchers to map clearly the inter-
communicator (Andrews & Soder, 1987). relationship among the environmental
Several of the predictors that are highly cor- context, school culture, and important school
related with the classification of schools in processes, such as the exercise of principal
the sample are informing others about the leadership. For example, the finding may
instructional goals of the school, promoting suggest that good schools tend to get good
the discussion of instructional issues, and fo- principals, while poor schools get poor prin-
cusing in particular on how instruction affects cipals. That is, district administrators, school
student achievement. The sum of these vari- staff, and parents may know how to select
ables suggests that effective principals may principals for high-achieving schools. An-
be better able to develop and clarify school other possibility is that because a school is
30
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
 Principals' Instructional Leadership
high achieving, the principal's actions may be reading and math over a 2-year period was
perceived as having positive leadership quali- due to contextual characteristics (e.g., socio-
ties. A final suggestion may be that poor economic status, language background) and
schools are difficult to change for new princi- educational attitudes of the community and
pals because of the necessity of reshaping the staff (e.g., curricular goals, academic expec-
existing normative structure. tations). As Oakes (1989) notes, contextual
The implication, then, is that the achieve- indicators such as academic expectations, ac-
ment context of the school may influence the cess to knowledge through curriculum, and
leadership actions of new principals. This professional teaching conditions at the school
finding is consistent with research on leader are enabling factors that may influence out-
succession and socialization, suggesting that come levels.
new leaders in organizations may encounter Policy implications of these findings sug-
problems with trying to change existing gest the difficulty of creating perform-
norms of behavior (Hart, 1991; Ogawa, ance changes in consistently low-achieving
1991). Instead, they tend to adjust to the schools, where several types of variables may
existing norms. In an earlier study of princi- be working in concert to constrain the princi-
pal succession, for example, Rowan and pal's instructional leadership efforts. This re-
Denk (1984) found that changes in principals search indicates the need for policymakers to
produced primarily short-term achievement develop change strategies that will be sup-
gains, which were also moderated by com- ported at all levels of the organization to en-
munity socioeconomic status. If that finding hance the reshaping of organizational norms
is confirmed by further research, the prob- (e.g., Firestone & Corbett, 1988; Louis &
lems of reforming low-achieving schools may Miles, 1991; Sarason, 1982).
extend beyond merely changing principals,
and it may be too much to assume that one Concluding Remarks
person can reshape these schools in a lasting As researchers have noted (Bossert et al.,
manner (e.g., Louis & Miles, 1991; Pitner & 1982; Hallinger & Murphy, 1987; Heck et al.,
Hocevar, 1987). The research base on princi- 1990), the instructional leadership role of the
pal leadership does not yet support the view principal is one key element in a conceptual
that such leaders can actually transform framework that recognizes the importance of
poorly achieving schools. As Hart (1991) sug- the school's social and environmental milieu
gests, with demands for more creative leader- in determining student achievement. In a
ship from principals and for school improve- sense, the identification of potential con-
ment, researchers need to expand inquiry straints on the exercise of principal leader-
into deliberate strategies to promote in- ship only serves to make establishing mean-
creased outcomes during succession when ingful systems of principal accountability
expectations for change are high. more difficult, because of the complex rela-
Ellett and Logan (1990) further argue that tionships among the school's context, school
the school's staff and surrounding commu- variables that can be reasonably manipu-
nity can play a significant role in shaping the lated, and intended achievement outcomes.
school's culture and norms, which ultimately Nevertheless, because of their ability to ob-
may influence student outcomes. Certainly, serve on a day-to-day basis, teachers' percep-
the ability to classify correctly almost every tions of the principal's implementation of in-
high- and low-achieving school with a new structional leadership activities may be useful
principal (although admittedly from a lim- in helping to determine the principal's effec-
ited sample) provides preliminary support of tiveness in performing the role, both for for-
this hypothesis. The influence of school con- mative and summative evaluation.
text and existing norms of behavior on out- Currently, however, about 85% of districts
comes is also supported by Heck and Mayor in the United States use summative ratings
(1991), who found that between 50% and by a sole supervisor as the only source of
60% of the variance in school achievement in input (Educational Research Service, 1985).
31
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
Heck

It is therefore difficult to identify specific the principal. Educational Administration

performance-based instructional leadership Quarterly, 18(3), 34-64.
tasks that may need improvement. In the Boyan, N. (1988). Describing and explaining ad-
present study, the extent of implementation ministrator behavior. InN. Boyan (Ed.), Hand-
(i.e., time allocation) of some instructional book of research on educational administration
leadership tasks within the three mediating (pp. 77-98). New York: Longman.
variables identified by Bossert et al. (1982) Crowson, R., & Porter-Gehrie, C. (1980). The
and Heck et al. (1990) as important domains discretionary behavior of principals in large-city
of principal influence (governance, climate, schools. Educational Administration Quarterly,
76(1), 45-69.
and instructional organization) was associ-
Cusick, P. (1983). The egalitarian ideal and the
ated with higher or lower school academic
American high school: Studies of three schools.
performance. In establishing at least limited
New York: Longman.
principal accountability (e.g., for school out-
Deal, T. E. (1987). Effective school principals:
comes), formative evaluation that is perfor- Counselors, engineers, pawnbrokers, poets
mance based could be an enhancement to the . . . or instructional leaders? In W. Greenfield
school's outcomes. Recent demands for ac- (Ed.), Instructional leadership: Concepts, is-
countability and increased public control of sues, and controversies (pp. 230-246). Boston:
school policy-making underscore both the Allyn & Bacon.
need and the challenge to develop and imple- Dreeban, R., & Barr, R. (1988). Classroom com-
ment objective systems for evaluating the position and the design of instruction. Sociology
principal's contribution to overall school of Education, 67(3), 1-11.
effectiveness. Dwyer, D., Lee, G., Rowan, B., & Bossert, S.
(1983). Five principals in action: Preparation in
Notes instructional management. San Francisco: Far
An earlier version of this paper was presented at West Laboratory for Educational Research and
the 1991 annual meeting of the AERA. The au- Development.
thor wishes to acknowledge Terry J. Larsen, who Ebmeier, H. (1991). The development of an instru-
provided the data used in the study, and to thank ment for client-based principal formative evalua-
several anonymous reviewers and the editor of the tion. Paper presented at the annual meeting of
journal for helpful comments on the manuscript. the American Educational Research Associa-
1
This information is determined from Califor- tion, Chicago.
nia Assessment Program data on student demo- Educational Research Service. (1985). Evaluating
graphics. Socioeconomic status is developed from administrator performance. Arlington, VA: Au-
parent occupational status and percentage of fam- thor.
ilies receiving Aid to Families with Dependent Ellett, C , & Logan, C. (1990). Analysis of school
Children (AFDC). Language background is de- level learning environments: Organizational
termined by the percentage of students who are coupling, robustness, and effectiveness. Paper
limited English speakers. presented at the annual meeting of the Ameri-
can Educational Research Association, Boston.
References Farrar, E., Neufeld, B., & Miles, M. (1984). Ef-
Andrews, R., & Soder, R. (1987). Principal in- fective schools programs in high schools: Social
structional leadership and school achievement. promotion or movement by merit. Phi Delta
Instructional Leadership, 44, 9-11. Kappan, 65, 701-706.
Barr, R., & Dreeban, R. (1983). How schools Firestone, W., & Corbett, H. (1988). Planned
work. Chicago: University of Chicago Press. organizational change. In N. Boyan (Ed.),
Blank, R. (1987). The role of the principal as Handbook of research on educational adminis-
leader: Analysis of variation in leadership in tration (pp. 321-340). New York: Longman.
urban high schools. Journal of Educational Re- Firestone, W., & Herriott, R. (1982). Prescrip-
search, 82, 69-80. tions for effective elementary schools don't fit
Bossert, S. (1988). School effects. In N. Boyan secondary schools. Educational Leadership, 40,
(Ed.), The handbook of research on educational 51-53.
administration. New York: Longman. Glasman, N. S., & Nevo, D. (1988). Evaluation in
Bossert, S., Dwyer, D., Rowan, B., & Lee, G. decision making: The case of school administra-
(1982). The instructional management role of tion. New York: Kluwer Academic Publishers.
32
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
 Principals' Instructional Leadership

Hallinger, P., & Murphy, J. (1986). The social Lessons from urban high schools. School Effec-
context of effective schools. American Journal tiveness and School Improvement, 2(2), 75-96.
of Education, 94, 329-355. Murphy, J. (1988). Methodological, measure-
Hallinger, P., & Murphy, J. (1987). Instructional ment, and conceptual problems in the study of
leadership in the school context. In W. Green- instructional leadership. Educational Evalua-
field (Ed.), Instructional leadership: Concepts, tion and Policy Analysis, 10(2), 117-134.
issues, and controversies (pp. 179-203). Boston: New Jersey officials seize control of "bankrupt"
Allyn & Bacon. Paterson schools. (1991, September 12). Edu-
Hart, A. (1991). Leader succession and socializa- cation Week, p. 12.
tion: A synthesis. Review of Educational Re- Oakes, J. (1989). What educational indicators?
search, 67(4), 451-474. The case for assessing the school context. Edu-
Heck, R. (1991). The effects of school context and cational Evaluation and Policy Analysis, 77(2),
principal leadership on school climate and 181-199.
school achievement. Paper submitted for pre- Ogawa, R. (1991). Enchantment, disenchant-
sentation at the annual meeting of the Ameri- ment, and accommodation: How a faculty made
can Educational Research Association, San sense of the succession of its principal. Educa-
Francisco. tional Administration Quarterly, 27(1), 30-60.
Heck, R., Larsen, T., & Marcoulides, G. (1990, Oklahoma board votes to oversee 8 at-risk
April). Principal leadership and school achieve- schools. (1991, September 12). Education
ment: Validation of a causal model. Paper pre- Week, p. 12.
sented at the annual meeting of the American Pitner, N., & Hocevar, D. (1987). An empirical
Educational Research Association, Boston. comparison of two-factor versus multifactor
theories of principal leadership: Implications
Heck, R., & Marcoulides, G. (1989). Examining
for the evaluation of school principals. Journal
the generalizability of administrative allocation
of Personnel Evaluation in Education, 7(1),
decisions. The Urban Review, 27(1), 51-62.
93-109.
Heck, R., Marcoulides, G., & Lang, P. (1991).
Rowan, B., Bossert, S., & Dwyer, D. (1983).
Principal leadership and school achievement:
Research on effective schools: A cautionary
The application of discriminant techniques.
note. Educational Researcher, 12(4),
School Effectiveness and School Improvement,
24-31.
2(2), 3-21.
Rowan, B., & Denk, C. (1984). Management suc-
Heck, R., & Mayor, R. (1991). The effects of
cession, school socioeconomic status, and basic
school context and constituent attitudes on
skills achievement. American Educational Re-
school achievement. Paper submitted for pre-
search Journal, 21, 517-538.
sentation at the annual meeting of the Ameri-
Sarason, S. (1982). The culture of the school and
can Educational Research Association, San
the problem of change. Boston: Allyn & Bacon.
Francisco.
Sizer, T. (1984). Horace's compromise: The di-
Herriott, R., & Firestone, W. (1984). Two images
lemma of the American high school. Boston:
of schools as organizations: A refinement and Houghton Mifflin.
elaboration. Educational Administration Quar- Tabachnick, B., & Fidell, L. (1983). Using multi-
terly, 20(4), 41-57. variate statistics. New York: Harper & Row.
Lee, J. (1987). Instructional leadership in a junior Tatsuoka, M., & Silver, P. (1988). Quantitative
high school: Managing realities and creating research methods in educational administra-
opportunities. In W. Greenfield (Ed.), Instruc- tion. In N. Boyan (Ed.), Handbook of research
tional leadership: Concepts, issues, and contro- on educational administration (pp. 677-702).
versies (pp. 77-99). Boston: Allyn & Bacon. New York: Longman.
Leithwood, K., & Montgomery, D. (1982). The Teddlie, C., Kirby, P., & Stringfield, S. (1989).
role of the elementary school principal in pro- Effective versus ineffective schools: Observable
gram development. Review of Educational Re- differences in the classroom. American Journal
search, 52, 309-339. of Education, 97, 221-236.
Little, J., & Bird, T. (1987). Instructional leader- Virgilio, I., Teddlie, C., & Oescher, J. (1991).
ship close to the classroom in secondary Variance and context differences in teaching at
schools. In W. Greenfield (Ed.), Instructional differentially effective schools. School Effec-
leadership: Concepts, issues, and controversies tiveness and School Improvement, 2(2), 152-
(pp. 118-138). Boston: Allyn & Bacon. 168.
Louis, K., & Miles, M. (1991). Managing reform: Willower, D. (1987). Inquiry into educational ad-
33
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014
Heck
ministration: The last twenty-five years and the the recent Chicago school reform. Administra-
next. Journal of Educational Administration, tor's Notebook, 34(5-6), 1-10.
25(1), 12-27.
Wimpelberg, R., Teddlie, C , & Stringfield, S.
(1989). Sensitivity to context: The past and fu- Author
ture of effective schools research. Educational RONALD H. HECK, Associate Professor of
Administration Quarterly, 25(1), 82-107. Education, Department of Educational Adminis-
Witte, J., & Walsh, D. (1990). A systematic test tration, College of Education, University of Ha-
of the effective schools model. Educational waii at Manoa, 1776 University Avenue, Hon-
Evaluation and Policy Analysis, 72(2), 188-212. olulu, Hawaii 96822. Specializations: role of
Wong, K., & Rollow, S. (1990). A case study of principal, policy and politics of education.

34
Downloaded from http://eepa.aera.net at Northeastern University on November 26, 2014