The current issue and full text archive of this journal is available at

www.emeraldinsight.com/0951-354X.htm

Impact of school
Exploring the impact of school leadership
leadership on pupil outcomes
Results from a study of academically improved
and effective schools in England 83
Pam Sammons Received June 2010
Department of Education, University of Oxford, Oxford, UK Accepted July 2010

Qing Gu and Christopher Day

School of Education, University of Nottingham, Nottingham, UK, and
James Ko
The Hong Kong Institute of Education, Hong Kong, PR China

Abstract
Purpose – This study aims to explore the impact of school leadership, particularly that of the
principal (head teacher), on school improvement in England.
Design/methodology/approach – The study adopts a mixed-methods (MM), longitudinal design
to investigate the leadership of a sample of academically effective and improving schools identified
from analyses of national assessment and examination data sets. In addition, questionnaire surveys of
principals and key staff and 20 qualitative school case studies were conducted. The paper reports
results from the questionnaire analyses and changes in measures of school performance over three
school years using data from 378 primary and 362 secondary schools. Confirmatory factor analysis
and structural equation models (SEM) test direct and indirect effects of school leadership and school
and classroom processes in predicting changes in schools’ academic results.
Findings – The analyses identified robust underlying dimensions of leadership and school and
classroom processes and highlighted strategies and actions that school principals and staff had
adopted to raise pupil attainment. The SEM models reveal both direct and indirect effects of leadership
on a range of school and classroom processes that in turn predicted changes (improvements) in
schools’ academic performance.
Originality/value – This original empirical study presents new results on the leadership of a large
sample of effective and improving primary and secondary schools in England. A dynamic model is
presented predicting changes in schools’ academic performance over three years and identifying direct
effects of leadership on school and classroom processes and indirect effects on improvements in
schools’ academic results.
Keywords Schools, Leadership, Principals, Organizational performance, England
Paper type Research paper

1. Introduction
This paper presents results from the Leadership and Pupil Outcomes study, a
three-year research project (2006-2009) commissioned by the Department for Children,
Schools and Families (DCSF) in England. It provides an account of the mixed methods International Journal of Educational
Management
Vol. 25 No. 1, 2011
The research was commissioned by the DCSF and the National College for School Leadership pp. 83-101
q Emerald Group Publishing Limited
and Children Services. The authors are very grateful to all the schools, teachers and pupils who 0951-354X
participated in the research. DOI 10.1108/09513541111100134
IJEM (MM) design used to explore the relationships between school leadership, in particular
25,1 that of the head teacher, and changes in school performance identified by using
measures of pupil attainment collected across three school years. Attainment data were
analysed to identify academically successful schools and heads and key staff surveyed
to investigate underlying dimensions of leadership. Structural equation models (SEM)
that link measures of leadership practices, school processes and changes in school
84 conditions to changes in pupil attainment are discussed.

1.1 Aims
The overall aim of the study was to identify and map empirically grounded direct and
indirect causal and associative relationships between school leadership and pupil
outcomes.
The study sought:
.
to investigate the relationships between variation in schools’ performance
(measured by pupils’ national assessments and public examination results) and
variation in the types qualities, strategies, skills and contexts of school
leadership; and
.
to explore the relative strengths of the direct and indirect influences of such
features of school leadership upon school organisation and processes and as
predictors of improvements in pupils’ outcomes.

2. Mixed methods design

Large-scale studies of complex social and educational phenomena increasingly adopt a
mixed-methods (MM) approach for their research design (Tashakkori and Teddlie,
2003; Day et al., 2008; Sammons, 2010). This research brought together an experienced
team with complementary areas of expertise, both qualitative and quantitative, from
different perspectives and in different contexts (UK and international).
Figure 1 illustrates the different phases and strands of the research and their
sequencing. The choice of a MM design was influenced by a review of the leadership

Figure 1.
Research design:
integrating evidence about
effective/improved schools
literature (Leithwood et al., 2006) that informed the design, the development of the Impact of school
questionnaire surveys, and the first rounds of case study interviews. leadership
The use of MM increased the possibilities of identifying various patterns of
association and possible causal connections between variation in different indicators of
school performance and measures of school processes and the way these are linked
with different features of leadership practices. The sequencing of the study facilitated
the integration of evidence and attempts at synthesis and meta-inferences. 85
The quantitative strand of the project involved three main components:
(1) An analysis of national assessment and examination data sets on primary and
secondary school performance was used to identify schools that were effective
and improving over a three year period (2003-2005). The analyses were based
on relevant published data and key indicators, including both “value added”
measures of pupil progress based on multilevel statistical analyses and
important accountability indicators such as the percentage of pupils achieving
performance benchmarks at Key Stage 2 (age 11), or at Key Stage 4 in public
GCSE examinations (age 16).
(2) A survey of heads and key staff in a stratified random sample of such effective
and improving schools was used to explore features of school organisation and
processes, including leadership. The survey questions asked heads and key
staff to report on the extent of change in different features of school activity and
practice over the last three years. This period coincided with the years for which
the analyses of pupil attainment data had taken place.
(3) A second follow-up survey of heads and key staff one year later to explore in
more detail particular strategies and actions that were perceived to relate to
improvement (informed by the interim results of the first survey and case study
findings).

The qualitative strand used 20 in-depth case studies of a subset of these improving and
effective schools involving three visits a year over two years with detailed interviews of
heads and a range of key staff and stakeholders. The case studies represented schools in
different sectors and contexts, including different levels of advantage and disadvantage,
and ethnic diversity. Interviews with heads and key staff prompted them to speak about
those issues that are most significant to them in relation to the research aims and
objectives and aspects identified as important in the literature review. Interviews with
other colleagues in the school provided insights into their perceptions of the nature and
impact of the practice and effectiveness of school (and departmental) leadership and its
distribution. Surveys of pupils in the case study schools provided additional data on
pupils’ perceptions of teaching and learning, leadership and school culture

2.1 The sampling strategy: identifying effective and improved schools

Analyses of national value added indicators and pupil attainment data for three years
(2003-2005) for all primary and secondary schools in England identified schools that
were more academically effective and that had shown sustained improvement for
further study in a questionnaire survey (Gu et al., 2008).
Approximately a third of primary (34 per cent) and of secondary (37 per cent)
schools in England for which national data were available were classified as more
effective/improved in terms of changes in pupil attainment. It should be noted that
IJEM English schools have recorded significant rises in national assessment and
25,1 examination results during the period 1996 to 2008 that reflect a strong policy drive
to raise standards and support the improvement of weaker schools (Sammons, 2008).
The research sampled schools within this group, allowing the project to focus on the
leadership features and processes of such successful schools, and the results can be
interpreted in terms of processes and approaches that characterise academically more
86 effective and improving schools in England.
2.1.2 School improvement groups. Three sub-groups of schools were identified
based on analyses of attainment and value added trends over three years:
(1) those improving from low to moderate or low to high in attainment and identified
as highly effective in national value added analyses (the “low start” group);
(2) those improving from moderate to higher moderate or high in attainment and
identified as highly effective in value added (the “moderate start” group); and
(3) those with stable high attainment that are highly effective in value added (the
“high start” group).

2.1.3 Questionnaire survey to investigate leadership and school processes.

A questionnaire survey was conducted for head teachers and key staff (two per
school at primary, five per school at secondary level). All secondary schools identified
as more effective/improved were surveyed along with a stratified random sample of
effective/improved primary schools. The survey included specific items that focussed
on heads’ and key staff’s perceptions of change in different areas of school work, and in
other kinds of pupil outcomes (non-academic areas such as engagement, motivation,
behaviour and attendance) over the last three years (for further details of the sample
and survey, see Gu et al., 2008).
The key staff survey closely mirrored that of the heads so that comparisons could
be made between the two. The response rate (Table I) was somewhat higher for heads
of secondary schools which were followed up in more detail to ensure roughly equal
numbers of responses from schools in each sector. Although not high, the response rate
is typical of that achieved by surveys of schools in England in recent years.

Sample size surveyed (n) Questionnaires returned (n) Response rate (%)

Heads
Primary 1,550 378a 24
Secondary 1,140 362a 32
Key staffb at school level
Primary 1,550 409c 26
Secondary 1,140 393c 34
Key staffa at questionnaire level
Primary 3,100 608d 20
Secondary 5,700 1,167d 20
Table I. Notes: aQuestionnaires returned by heads. bKey staff include members of the senior leadership team
Wave 1 survey and middle managers (e.g. Key Stage Leaders). cSchools with returned key staff questionnaires.
d
response rate Returned key staff questionnaires
3. Analysis strategies for the survey data Impact of school
Heads’ and key staff’s responses were compared to explore the extent of similarity in leadership
perceptions. In addition, we examined the data according to a range of contextual
variables of policy interest. These included school sector (primary or secondary) and
socio-economic context (based on free school meal (FSM) band), comparing schools
with more disadvantaged pupil intakes (FSM Bands 3 and 4) and those with less
disadvantaged intakes (FSM Bands 1 and 2). We also examined other potential sources 87
of influence including the head’s total years of experience and the head’s time in their
current post.
A particular focus was to explore similarities and differences in survey responses
between the three school improvement groups identified in the sample. Associations
between the improvement groupings and a range of influences relating to heads’ years
of experience in total and in their current schools, the number of heads in the last ten
years; school education sector and school socio-economic contexts and survey results
were explored.

3.1 Identifying underlying dimensions relevant to leadership and school improvement

Both exploratory and confirmatory factor analysis (CFA) were adopted to explore the
underlying structure of the head teacher and key staff survey data and test the extent
to which the features of leadership and practice identified as important in the literature
review could be confirmed using data for the sample of English effective and improved
schools.
These quantitatively derived dimensions were then related to hypothesised models
of the proposed links between different features of leadership practice and measures of
change in pupil attainment outcomes over three years using structural equation
modelling (SEM).
In developing the SEM models we were influenced by previous research, especially
that of Silins and Mulford (2004) in their LOSLO study of leadership and organisational
learning. Their study also modelled relationships with academic attainment at a
particular time point, measured by completion of high school (Mulford, 2008). Our
study of leadership and pupil outcomes is the first use of SEM to predict change in
school performance (measured by pupil attainment) over time. These models can be
seen as dynamic rather than cross sectional (in contrast to the approach in the LOSLO
study) because they indicate the factors that predict, directly and indirectly,
improvement in school performance measured by pupils’ academic results.

4. Quantitative results: exploring differences by school improvement group

Statistically significant differences between the three school improvement groups were
identified across a range of survey items relating to heads’ years of experience in total,
the number of heads in post in the school over the last ten years, and other aspects
including school education sector and school socio-economic context.
In both education sectors the “high start” group of schools were relatively more
likely to serve less disadvantage communities, defined in terms of the percentage of
pupils on free school meals (FSM). Schools with low disadvantage were classed as FSM
Bands 1 and 2. By contrast, the “low start” improvement group of schools were
relatively more likely to serve disadvantaged communities measured by the higher
percentage of pupils on FSM (FSM Bands 3 and 4).
IJEM Nearly two thirds (n ¼ 105, 65.6 per cent) of primary schools in the “low start”
25,1 group, compared with under one in ten (n ¼ 10, 8 per cent) of the “high start” group
being in high disadvantage contexts (FSM Bands 3 and 4). Similarly over half (n ¼ 84,
50.3 per cent) of secondary schools in the “low start” group compared with only around
one in 20 of schools in the “high start” group (n ¼ 6, 5.2 per cent) were classed as
serving similarly disadvantage contexts. Although 71 per cent of schools responding to
88 the survey were in relatively low disadvantage contexts (FSM Bands 1 and 2) they
accounted for only around half (49.7 per cent) of schools in the “low start” group.
These results point to the importance of school socio-economic context in
understanding differences in school performance results and their improvement
trajectories. In England more rapid improvement may be likely for schools in
disadvantaged contexts with a history of low attainment as reviews of the evidence on the
improvement of poorly performing schools and those in special measures has indicated,
due to the sustained policy emphasis they have been given in the last 15 years as part of a
drive to raise standards and enhance accountability (Muijs et al., 2004; Sammons, 2008).
For the secondary sample, head teachers with less total experience of headship were
more likely to lead high-disadvantage schools, but this pattern was not identified for
primary heads. There were significant differences between the three school improvement
groups in the total years of experience of the head. For both the primary and secondary
samples, less experienced heads were proportionately more likely to be in post in schools
from the “low start” group, whereas schools in the “high start” group were relatively
more likely to have an experienced head in post. In total, 47 per cent of heads of the “low
start” group of primary schools had seven or fewer years of experience as a head, in
contrast to only 25 per cent of those in the “high start” group. At the secondary level, 62
per cent of heads of the “low start” group had the same amount of experience (seven
years or below as a head), compared with 49 per cent of those in the “high start” group.
In both sectors, statistically significant differences were also found between the
three school improvement groups in terms of number of heads that had been in post
over the last decade. Schools in the High Start group were much less likely to have
experienced several changes of head than those in the Low Start group, for example.

4.1 Perceptions of school improvement over the last three years

The survey explored heads’ and key staff’s perceptions of the extent of change in their
school over the last three years in terms of a range of important topics, including school
climate, culture, leadership distribution, teaching, pupil attendance, behaviour and
motivation. For both the primary and the secondary sample there were significant
differences in responses related to the three school improvement groups.
4.1.1 Changes in disciplinary climate. Heads in the “low start” group were more likely
to report sustained improvement in all aspects of pupil behaviour than those in the “high
start” group (it is likely that behaviour in such “high start” schools was already good and
stable). Also, primary heads reported relatively less change in pupil behaviour over the
last three years than their secondary counterparts. The most marked difference was in
relation to the item on “pupils missing class”. At the primary level, 28 per cent of heads in
the “low start” group (n ¼ 40) indicated improvement in this area compared with only 7
per cent of heads in the “high start” group. However, a striking 79 per cent (n ¼ 131) of
secondary heads in the “low start” group noted improvement in this aspect of pupil
behaviour over the last three years compared with only 29 per cent of those in the “high
start” group. This indicates that pupils missing class, rather than general pupil absence, Impact of school
had been a particular feature of low-attaining secondary schools, and improvement in leadership
this is likely to be associated with improved pupil motivation and to have had an impact
upon pupils’ opportunities to learn, and hence their attainment.
Key staff perceptions supported those of the heads in the survey in terms of
agreement about the improvement of pupils’ outcomes in both academic attainment
and non-academic areas. Significant differences were found between staff responses 89
for all three school improvement groups when asked about the extent of reported
change in the discipline climate of the school in last three years. Those from schools in
the “low start” group were the most likely to report that it is “better now” or “much
better now”, in line with the results for heads.
4.1.2 Changes in overall school conditions. There was strong correspondence in
views on the extent of change in school conditions between heads and key staff. For
both sectors, the majority of heads noted considerable improvements in a range of
areas, including:
.
the “commitment and enthusiasm of staff”;
.
achieving “an orderly and secure working environment”;
.
“enhanced local reputation”; and
.
“improved pupil behaviour and discipline as a result of a whole school approach”
(see Tables II and III).

Secondary heads also reported positive change in “more pupils going into
further/higher education” (n ¼ 230, 64 per cent). They were relatively more likely to
indicate “improvements in homework policies and practices” (n ¼ 287, 79 per cent)
than primary heads (n ¼ 262, 70 per cent), although for both sectors this was an area in
which much improvement was reported.
Around 60 per cent of key staff reported “some” or “a lot” of change when asked
about “enhanced local reputation of the school”. Secondary staff were more likely to
report a lot of positive change than their primary counterparts. Between 50 per cent
and 70 per cent of key staff reported some or a lot of change in terms of “enhanced
commitment and enthusiasm of staff”, “changes in the homework policies and
practice”, and “improvements in terms of achieving an orderly and secured working
environment” over past three years. These patterns of change are in line with school
effectiveness and improvement research that points to the importance of the
behavioural and school climate as key characteristics of effectiveness (Teddlie and
Reynolds, 2000) and the results of previous case studies of improving and turnaround
schools (Haydn, 2001; Harris et al., 2006).
For both the primary and the secondary sample, heads in the “low start” group were
most likely to indicate substantial improvement across their school over the past three
years in most areas, while no change was more likely to be reported by those in the
“high start” group (see Tables II and III).
4.1.3 Changes in curriculum, pedagogy and assessment. School improvement group
was related to the extent of reported change in almost all the aspects of and curriculum,
pedagogy and assessment for secondary schools but fewer associations were found for
the primary sample.
 90
25,1
IJEM

Table II.

change in school
conditions (primary)
Heads’ perceptions of
Extent of change
No
change A little Some A lot Total
Indicator of improvement n % n % n % n % n %

Enhanced commitment and enthusiasm of staff 54 14.4 50 13.3 121 32.2 151 40.2 376 100
Promoted an orderly and secure working environment 79 21.0 47 12.5 85 22.5 166 44.0 377 100
Enhanced local reputation 64 17.1 65 17.3 125 33.3 121 32.3 375 100
Improved pupil behaviour and discipline as a result of a whole-school approach 78 20.7 55 14.6 107 28.5 136 36.2 376 100
 Extent of change
No
change A little Some A lot Total
Indicator of improvement n % n % n % n % n %

Enhanced commitment and enthusiasm of staff 32 8.9 49 13.6 150 41.6 130 36.0 361 100
Promoted an orderly and secure working environment 41 11.4 58 16.1 137 38.0 125 34.6 361 100
Enhanced local reputation 44 12.2 53 14.7 107 29.7 156 43.3 360 100
Improved pupil behaviour and discipline as a result of a whole-school approach 44 12.2 70 19.4 127 35.3 119 33.1 360 100

conditions (secondary)
leadership
Impact of school

change in school
Heads’ perceptions of
Table III.
91
IJEM Primary heads in the “low start” group were somewhat more likely than primary heads
in other improvement groups to report a moderate or a substantial amount of change in
25,1 the following items:
.
“using coaching and mentoring to improve the quality of teaching”;
.
“encouraging staff to use data in their work”; and
92
.
“encouraging staff to use data in planning for individual pupil needs”.
For secondary heads, relatively more change was reported by those in the “low start”
group for “using coaching and mentoring to improve the quality of teaching”. Half of
the secondary heads in the “low start” group (n ¼ 85, 52 per cent) reported a lot of
change in this aspect, compared with 40 per cent of those in the “high start” group.
In the secondary sector the school improvement group was also associated with the
amount of reported change in relation to a number of items on classroom observation
and assessment. Heads in the “low start” group were relatively more likely to report a
substantial degree of change relating to “regularly observing classroom activities” and
“working with teachers to improve their teaching after observing classroom activities”.
Substantial change in “regularly observing classroom activities” was indicated by 65
per cent of secondary heads of the “low start” group, compared with 55 per cent of
those in the “high start” group. Relatively more change in leadership practice was also
reported by heads in this improvement group for items related to “incorporating
research evidence into their decision making to inform practice” and “using pupil
achievement data to make most decisions about school improvement”.
All heads in secondary schools reported a considerable change in the extent of
encouraging the use of data by teachers, but this was a particular emphasis in “low
start” secondary schools. In England there has been a strong policy focus on the use of
performance data to support school improvement (Sammons, 2008) and these results
are likely to reflect the wide availability of such data in English schools.

4.2 Actions identified as most important in promoting school improvement

The surveys requested details of the three actions/strategies identified by heads as
most influential in improving pupils’ academic outcomes in their school during the last
three years. These written data were analysed to establish which strategies and actions
were viewed as most important and most frequently adopted by schools.
The written results were categorised based on similarities and differences in the
actions and areas cited. In some cases actions could be allocated into more than one
category, for example “Use of performance data to set high expectations” was
considered to fall into both the category “Encouraging the use of data and research”
and the category “Demonstrating high expectations for staff”.
The most frequently cited actions/strategies by primary heads were:
.
improved assessment procedures (28.1 per cent);
.
encouraging the use of data and research (27.9 per cent);
.
teaching policies and programmes (26.0 per cent);
.
strategic allocation of resources (20.4 per cent);
.
changes to pupil target setting (20.2 per cent);
.
providing and allocating resources (19.4 per cent); and
.
promoting leadership development and CPD (15.9 per cent).
Similarly, actions/strategies most frequently cited by secondary heads were: Impact of school
.
encouraging the use of data and research (34.0 per cent); leadership
.
teaching policies and programmes (27.7 per cent);
.
improving school culture (21.1 per cent);
.
providing and allocating resources (19.5 per cent);
.
improved assessment procedures (18.6 per cent); 93
.
monitoring of departments and teachers (15.9 per cent); and
.
promoting leadership development and CPD (15.1 per cent).

The responses were further grouped according to the five major categories of leadership
practices identified in the initial literature survey. Overall, for primary heads actions and
strategies relating to the category “Improving teaching practices” were cited most
commonly (a total of 359, representing 28.4 per cent of the 1,263 actions/strategies listed),
followed by those that promoted their school’s “Academic press/emphasis” (cited 251
times or 19.9 per cent of the total). Responses relating to the category “Redesigning the
organisation” reached 209 (16.6 per cent) followed by those concerned with “Setting
directions” (122, 9.7 per cent) and “Developing people” (119, 9.4 per cent).
The pattern of results for secondary heads was similar to that for primary schools.
Actions related to “Improving teaching practices” received the largest number of
mentions by secondary heads (258 out of the total 1,168 responses made, 22.1 per cent). A
focus on actions related to promoting the school’s “Academic press/emphasis” featured in
188 responses (16.1 per cent). Actions relevant to “Redesigning the organisation” were
noted 179 times (15.3 per cent of total actions/strategies cited) followed by those connected
to “Setting directions” (115, 11.5 per cent) and “Developing people” (98, 8.4 per cent).
These results link closely with the role of the head teacher as an instructional leader
as discussed by Hallinger (2005), who argued that leadership must be conceptualised
as a process of mutual influence whereby instructional leaders influence the quality of
school outcomes through shaping the school mission and the alignment of school
structures and culture to promote a focus on continuous improvement and high
expectations centred on raising the quality of teaching and learning.
Examples of written comments related to the key strategies reported by heads that
they had used to raise attainment are illustrated below: they show the emphasis given
to using data and research and on developing teachers’ capacities:
Giving specific data to individuals and teams to help inform planning and target setting.
Pupil tracking, target setting and mentoring scheme.
Improve/change curriculum on offer at Key Stage 4.
Managing some teachers’ learning.
Building a learning network.
Focus on the role of the Middle Leader.
Develop of a culture of research and innovation.
Development of a learning toolkit for staff.
IJEM 5. Investigating the relationships between leadership, school processes and
25,1 changes in pupil outcomes: developing the structural equation model (SEM)
Exploratory factor analysis (EFA) followed by confirmatory factor analysis (CFA) was
used to test the existence of hypothesised underlying dimensions identified as
important in the literature review and measured by various items in the survey.
A five-factor model of leadership practice was identified for the secondary head sample
94 (see Table IV) with a relatively high internal consistency reliability of 0.950 (see
Appendix 1, Table AI): the model for primary head teachers was very similar therefore
only the model for secondary heads is described.
One example of the findings in relation to the CFA model of change in school
conditions is illustrated before discussion of the overall SEM results for secondary
schools. A three-factor model (see Table V) was identified for secondary schools (with
an internal consistency reliability of 0.889). This points to the extent of changes in the
behavioural climate in school conditions and positive change (reductions) in specific
types of poor pupil behaviour over the last three years.
It was hypothesised that changes in school conditions and intermediate pupil
outcomes such as attendance, behaviour and motivation would precede change
(improvement) in schools’ performance measured in terms of pupils’ academic
outcomes, and that both would be affected by leadership, but mainly indirectly via
effects on staff and school processes and conditions. After deletion of missing data the

Dimensions Questionnaire items

Setting directions1d. Demonstrating 1e. Demonstrating 1f. Demonstrating 1g. Working

high expectations high expectations high expectations collaboratively
for staff’s work for pupil for pupil with the governing
with pupils behaviour achievement body
Developing people 2b. Encouraging 2e. Promoting 2f. Promoting a 2g. Encouraging
staff to consider leadership range of CPD staff to think of
new ideas for their development experiences learning beyond
teaching among teachers among all staff the academic
curriculum
Redesigning 3a. Encouraging 3e. Improving 3h. Allocating 3j. Structuring the
organisation collborative work internal review resources organisation to
(internal among staff procedures strategically based facilitate work
strategies) on pupil needs
Use of data 4g. Encouraging 4h. Encouraging
staff to use data in all staff to use data
their work in planning for
individual pupil
needs
Use of observation 4b. Regularly 4c. After 4d. Using coaching
Table IV. observing observing and mentoring to
Questionnaire items that classroom classroom improve quality of
underpin the five-factor activities activities, working teaching
CFA model of leadership with teachers to
practice (secondary improve their
heads) teaching
SEM analysis was conducted with data for 309 secondary schools and 363 primary Impact of school
schools. Results for the primary and secondary samples showed strong similarities and leadership
this paper focuses on the SEM results based on survey data from secondary head
teachers (see Appendix 2, Table AII).

5.1 A structural model of leadership practice in effective/improved secondary schools

The full SEM model is shown in Appendix 3, Figures A1-A2. It identifies relationships 95
between dimensions related to school leadership and processes and predicts changes
(improvement) in schools’ academic results over three years (2003-2005) using survey
data from secondary head teachers. The strength of the loadings indicates the nature
and strength of the relationships between 19 dimensions (latent constructs) that
emerged from the data set. Four levels of relationships were identified.
Level 1 comprises three key dimensions of leadership – i.e. “setting directions”,
“redesigning the organisation” and “head trust” – plus three other dimensions – i.e.
“developing people”, “use of data” and “use of observation” – linked with the first two.
There is a strong positive correlation between the first two constructs (r ¼ 0:70), i.e.
“setting directions” and “redesigning the organisation”, both relating to change in the
practice of leadership over the three-year period. However, no significant links were
found between either of these two constructs and “head trust”, suggesting that the two
aspects of leadership practice by the head and their “relational trust” in the staff may
have differing roles in improving school performance and pupil outcomes.
In addition to these three key dimensions of leadership, there are another three
dimensions at this level:
(1) “use of data”;
(2) “developing people”; and
(3) “use of observation”.

These three dimensions, together with “setting directions” and “redesigning the
organisation”, form a five factor structural model of change in leadership practice over
three years. We also identified links between these measures of leadership practice and
the dimensions of distributed leadership (described below).
Level 2 comprises four dimensions in relation to leadership distribution in the
school:

Dimensions Questionnaire items

Improvement of school 12f. School experienced: 12g. Promoted an 12h. Improved pupil
conditions (ImpSchoC) enhanced commitment orderly and secure behaviour and
and enthusiasm of staff working environment discipline as a result of
a whole-school Table V.
approach Questionnaire items
underpinning the
Improvement in pupil 13f. Changes in level of 13i. Changes in level of 13j. Changes in level of
three-factor CFA model of
behaviour (PupMisBe) physical conflict among physical abuse of verbal abuse of teachers
improvement in school
pupils teachers
conditions and
Improvement in pupil 13a. Changes in: pupils’ 13b. Changes in pupils’ 13d. Changes in pupils’ intermediate pupil
attendance (PupAtten) lateness to lessons lateness to school missing class outcomes (secondary)
IJEM (1) “distributed leadership”;
25,1 (2) “leadership by staff”;
(3) “senior leadership team (SLT) collaboration”; and
(4) “the SLT’s impact on learning and teaching”.

The dimension “heads trust” has a moderate direct effect on the dimension “SLT
96 collaboration” and direct but relatively weaker effects on the other dimensions
“leadership by staff”, “SLT’s impact on leaning and teaching”, and “distributed
leadership”. The two dimensions “redesigning the organisation” and “setting directions”
have an indirect impact on “distributed leadership” operating through the dimension
“developing people”. The effect of change in the emphasis given to “developing people”
over three years on “distributed leadership” is by contrast weak but negative. This may
be because where more attention is required for “developing people”, the conditions may
not yet be seen as appropriate for greater “distributed leadership”.
As the SEM results for secondary schools suggest, the leadership practices of the
head and of the SMT (Levels 1 and 2 dimensions) appear to influence, directly or
indirectly, the improvement of different aspects of school culture, processes and
conditions (Level 3 variables) which then indirectly impact on the change in pupil
academic outcomes through improvements in several important intermediate
outcomes (Level 4 variables).
Level 3 comprises four dimensions relate to improved school and classroom
processes which seem to function as mediating factors in this structural model:
(1) “teacher collaborative culture”;
(2) “assessment for learning”;
(3) “improvement in school conditions”; and
(4) “external collaborations and learning opportunities”.

Level 4 comprises four dimensions:

(1) “high academic standards”;
(2) “pupil motivation and learning culture”;
(3) “change in pupil behaviour”; and
(4) “change in pupil attendance”.

These constructs appear to be important intermediate outcomes that are important in

their own right but that also have direct or indirect effects on changes in pupil
academic outcomes over three years.
It is also evident that some dimensions (latent constructs) in our model have direct
effects on other dimensions at more than one level. For example, in addition to its
impact on Level 2 dimensions, “head trust” also has a direct moderate impact on
“teacher collaborative culture” (Level 3). “Redesigning the organisation” also has a
direct moderate effect on “Improvement in school conditions” (Level 3), which
indirectly promotes positive change in pupil academic attainment outcomes through
“change in pupil” behaviour.
The dimension “teacher collaborative culture” likewise has a direct moderate impact
on improvement in “pupil motivation and learning culture”, which in turn has an indirect
effect on change in the “pupil academic outcomes” measure through the impact on the two Impact of school
dimensions “improvement in pupil attendance” and “improvement in pupil behaviour”. leadership
6. Summary and conclusions
Full details of the leadership and pupil outcomes project findings are presented in
detailed reports (Day et al., 2007, 2009, 2011) that draw together both the qualitative
case study and quantitative results. 97
The current findings on academically effective and improved schools in England
provide new evidence that demonstrates the links between various leadership
dimensions and school organization and processes. This paper has reported only part
of the rich data and analyses conducted for the MM study. Taken together, the survey
results for heads and key staff indicate that rapidly improving schools in England
show improvement across a range of areas to do with practice, climate and learning
conditions that are likely to have a mutually reinforcing impact and help schools to
break out of an existing low attainment state into an upward trajectory. Such change
was particularly evident in successful secondary schools that had a past history of low
performance, the “low start” group. The SEM analyses illustrate the existence of direct
effects of leadership on a range of important dimensions of school and classroom
processes and point to modest but statistically significant indirect links with changes
in school conditions that in turn lead to improvements in students’ academic outcomes
at the school level.
The effects of leadership on improvements in pupils’ academic outcomes, however,
are weaker and seem to operate indirectly as might be expected given the role of
leaders, especially the head teacher in schools in England. School and leadership effects
are shown to influence changes in school academic outcomes via their effects on
teachers, and teaching quality and on promoting a favourable school climate and
culture that emphasises high expectations and academic outcomes. The present study
highlights, in particular, the importance of a model of leadership practice that promotes
an orderly and favourable behavioural climate, positive learner motivation and a
learning culture that predicts positive changes in pupil behaviour and attendance as
intermediate outcomes that themselves promote improvement in attainment.
The current project is the first of its kind to focus explicitly on studying leadership
and pupil outcomes in a large sample of schools identified as more academically
effective and improved. We built on earlier research by Silins and Mulford (2004) and
Mulford (2008) Our study differs, however, because we chose an explicit focus in
predicting change in a measure of school performance based on improvement in pupil
attainment. Thus our SEM can be seen to fit a dynamic rather than a static theory of
educational effectiveness (Creemers and Kyriakides, 2008).
Our results differ somewhat to those reported by Kruger et al. (2007) that
investigated the impact of school leadership on school-level factors. Their study also
used a SEM path analysis to test and validate a causal model. They found no direct or
indirect effects of educational leadership on student commitment in a secondary
analysis of data on secondary schools and leadership in The Netherlands. A major
difference is that they did not study attainment or improvement in attainment and thus
direct comparisons are not possible. Nonetheless, their models do indicate significant
relationships between leadership and quality of the school organisation that accord
with our findings. More comparative research on leadership effects in a range of
IJEM different countries and investigating a greater variety of student outcomes are needed
25,1 to test the generalisability of the present findings in other contexts (Pashiardis and
Ribbins, 2003; Brauckmann and Pashiardis, 2009) Recent multilevel longitudinal
research by Kythreotis et al. (2010) in Cyprus examined changes in primary students’
attainments and examined the relationships among school leadership, school culture
and student achievement and provided some evidence of direct effects of school
98 leadership.

References
Brauckmann, S. and Pashiardis, P. (2009), “From PISA to LISA. New educational governance and
school leadership: exploring the foundations of a new relationship in an international
context”, paper presented at the AERA Annual Meeting, San Diego, CA, April.
Creemers, B.P.M. and Kyriakides, L. (2008), The Dynamics of Educational Effectiveness,
Routledge, London.
Day, C.W., Sammons, P. and Gu, Q. (2008), “Combining qualitative and quantitative
methodologies in research on teachers’ lives, work and effectiveness: from integration to
synergy”, Educational Researcher, Vol. 37 No. 6, pp. 330-42.
Day, C., Sammons, P., Leithwood, K., Hopkins, D., Gu, Q., Brown, E. and Ahtaridou, E. (2011),
School Leadership and Student Outcomes: Building and Sustaining Success, Open
University Press, Buckingham (forthcoming).
Day, C., Sammons, P., Hopkins, D., Harris, A., Leithwood, K., Gu, Q., Penlington, C., Mehta, P. and
Kington, A. (2007), The Impact of Leadership on Pupil Outcomes: Interim Report, DSCF
Research Report RR018, Department of Children, Families & Schools/National College of
School Leadership, Nottingham.
Day, C., Sammons, P., Hopkins, D., Harris, A., Leithwood, K., Gu, Q., Penlington, C., Mehta, P. and
Brown, E. (2009), The Impact of Leadership on Pupil Outcomes: Final Report to DSCF,
Department of Children, Families & Schools/National College of School Leadership,
Nottingham.
Gu, Q., Sammons, P. and Mehta, P. (2008), “Leadership characteristics and practices in schools
with different effectiveness and improvement profiles”, School Leadership & Management,
Vol. 28 No. 1, pp. 43-63.
Hallinger, P. (2005), “Instructional leadership and the school principal: a passing fancy that
refuses to fade away”, Leadership and Policy in Schools, Vol. 4 No. 3, pp. 1-20.
Harris, A., Chapman, C., Muijs, D., Russ, J. and Stoll, L. (2006), “Improving schools in challenging
contexts: exploring the possible”, School Effectiveness and School Improvement, Vol. 17
No. 4, pp. 409-24.
Haydn, T. (2001), “From a very peculiar department to a very successful school: transference
issues arising out of a study of an improving school”, School Leadership and Management,
Vol. 21 No. 4, pp. 415-39.
Kruger, M., Witziers, B. and Sleegers, P. (2007), “The impact of school leadership on school level
factors: validation of a causal model”, School Effectiveness and School Improvement, Vol. 18
No. 1, pp. 1-20.
Kythreotis, A., Pashiardis, P. and Kyriakides, L. (2010), “The influence of school leadership styles
and culture on students’ achievement in Cyprus primary schools”, Journal of Educational
Administration, Vol. 48 No. 2, pp. 218-40.
Leithwood, K., Day, C., Sammons, P., Harris, A. and Hopkins, D. (2006), Seven Strong Claims
about Successful School Leadership, DfES, London/NCSL, Nottingham.
Muijs, D., Harris, A., Chapman, C., Stoll, L. and Russ, J. (2004), “Improving schools in Impact of school
socio-economically disadvantaged areas: a review of research”, School Effectiveness and
School Improvement, Vol. 15 No. 2, pp. 149-75. leadership
Mulford, B. (2008), “The leadership challenge: improving learning in schools”, Australian
Education Review, No. 53, Australian Council for Educational Research, Camberwell.
Pashiardis, P. and Ribbins, P. (2003), “On Cyprus: the making of secondary school principals”,
International Studies in Educational Administration, Vol. 31 No. 2, pp. 13-34. 99
Silins, H. and Mulford, B. (2004), “Schools as learning organisations – effects on teacher
leadership and student outcomes”, School Effectiveness and School Improvement, Vol. 15
No. 4, pp. 443-66.
Sammons, P. (2008), “Zero tolerance of failure and new labour approaches to school improvement
in England”, Oxford Review of Education, Vol. 34 No. 6, pp. 651-64.
Sammons, P. (2010), “The contribution of mixed methods to recent research on educational
effectiveness”, in Tashakkori, A. and Teddlie, C. (Eds), Handbook of Mixed Methods in
Social and Behavioral Research, 2nd ed., Sage Publications, Thousand Oaks, CA, Ch. 27,
pp. 697-723.
Tashakkori, A. and Teddlie, C. (2003), Handbook of Mixed Methods in Social and Behavioral
Research, Sage Publications, Thousand Oaks, CA.
Teddlie, C. and Reynolds, D.S. (2000), The International Handbook of School Effectiveness
Research, Pergamon, Oxford.

Further reading
Leithwood, K., Louis, K., Anderson, S. and Wahlstrom, K. (2004), The Learning from Leadership
Project (2004-2009), funded by the Wallace Foundation, New York, NY.
Sammons, P., Thomas, S. and Mortimore, P. (1997), Forging Links: Effective Schools and Effective
Departments, Paul Chapman, London.

Corresponding author
Pamela Sammons can be contacted at: pamela.sammons@education.ox.ac.uk

Appendix 1

Table AI.
Model x2 df GFI AGFI RMSEA NFI CFI PNFI PGFI Standardised RMR
Fit indices for the
Four-factor 176.41 71 0.93 0.90 0.064 0.98 0.99 0.76 0.63 0.049 four-factor CFA model
(secondary head teacher
Note: n ¼ 362 sample)

Table AII.
Appendix 2 Fit indices for the CFA
measurement model
including all 34 latent
Model x2 df GFI AGFI RMSEA NFI CFI PNFI PGFI Standardised RMR variables (to be used for
the structural model)
34-factor 5,439.04 3,448 0.72 0.66 0.043 0.88 0.95 0.74 0.59 0.054 (secondary)
IJEM Appendix 3
25,1

100

Figure A1.
Sample of Wave 1
headteacher survey
questions
Impact of school
leadership

101

Figure A2.
An SEM based on
secondary head teachers’
survey responses linked
with changes in students’
academic outcomes