This study examined the individual and school-level determinants of teachers' attitudes towards change in Greek primary schools using a multilevel analysis. The results showed that 22% of the variability in teachers' attitudes was explained by school-level factors. Teachers' perceptions of their principal's transformational leadership style, in particular, explained part of the between-school variability in attitudes. Schools influenced teachers' affective and cognitive dimensions of attitudes more than their behavioral dimension. Additional factors like teachers' age, principal's education level, and perceptions of past change diffusion also impacted teachers' attitudes. The findings provide insights into how leadership behaviors and other school characteristics can shape teachers' receptiveness to educational reforms.
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
Teachers' Attitudes Towards Change: A Multilevel Analysis of Individual and School Factors
1. Marina-Stefania Giannakaki
University of Thessaly, Greece & Open University of Cyprus
Eleftheria Belikaidi
Directorate of Primary Education of Kastoria, Greece
Individual and school-level determinants of
teachers’ attitudes towards change: A multilevel
analysis in the context of Greek primary
schools
2. The purpose of the study was to examine the
link between leadership style and teachers’
attitudes to change accounting for control
variables operating at two (nested) levels:
(a) the individual teacher and
(b) his/her school
3. Rationale of the study
1. Worldwide discussion on the causes of success or failure
of change efforts in education.
2. Teachers’ attitudes: key determinants of successful
change implementation and its sustainability in the long-
term.
3. Leadership: important covariate of teachers’ attitudes to
change.
4. In Greece, previous research on the link between
leadership and attitudes ignored group effects resulting
from the clustering of teachers within schools.
5. This study took account of the hierarchical structure of
the data.
4. Research Questions
1. What are teachers’ attitudes towards change in Greek
primary schools?
2. Is there between-school variability in teachers’ attitudes?
3. Can this variability be explained by differences in the
characteristics and leadership style of the principal?
4. Does the strength of the relationship between leadership
and teachers’ attitudes differ from school to school?
5. Can other school-level factors, such as school resources
and/or area characteristics explain between-school
variations in teachers’ attitudes?
6. What other characteristics of the teacher are associated
with their attitudes towards change?
5. Key explanatory variables of
interest
A. Leadership behaviours measured:
Transactional: Exchange of tangible rewards as a way to
motivate people to work for the achievement of
organisational goals. Useful for lower-order goals and non-
significant change.
Transformational: Inspires individuals to do more than
they thought possible and to work for the organisation’s
good. Articulates vision and stimulates people to view the
world from new angles. Useful for higher-order, long-term
goals and significant change.
Avoidant/laissez-faire: ‘No leadership’. Tends to react
only after problems have become serious. Avoids setting
goals and making decisions.
6. B. Earlier experiences of change implementation:
Teachers’ perceptions of the degree to which their
school had implemented innovative changes in the last
5 years.
Teachers’ perceptions of the extent to which these
innovative changes had been diffused within the school
and/or the local community (as a ‘proxy’ of successful
change implementation).
7. School-level
predictors
(LEVEL 2)
• School area: urban/rural/semi-rural
• Land prices in the area (‘proxy’ of SES)
• School size, student mix & infrastructure
• Principal’s demographics
• Principal’s education/training & experience
• Principal’s attitudes towards change
• Principal’s perceptions of earlier change implementation
• Principal’s perceptions of the diffusion of innovative changes
Teacher-level
predictors
(LEVEL 1)
• Teacher demographics
• Teacher education/training & experience
• Teachers’ perceptions of leadership
• Teachers’ perceptions of earlier change implementation
• Teachers’ perceptions of the diffusion of innovative changes
Overview of two-level model
(predictors of teachers’ attitudes to change)
8. Two stage cluster sampling
Stage 1
(Systematic) random sample of 50 schools from the
Directorate of Primary Education of Western Thessaloniki,
Northern Greece (based on available public school
catalogues).
Stage 2
Three teachers per school (on average) based on specific
criteria (i.e. the grade they taught, so that all grades were
equally represented).
Self-administered questionnaire to teachers and
principals.
139 teachers and 50 principals responded
10. ‘Attitudes towards change’ inventory by Dunham
et al. (1989) (sample items)
Response scale: (0) Strongly disagree, (1) Disagree, (2)
Neither agree nor disagree, (3) Agree, (4) Strongly agree
Cognitive dimension (8 items)
Cronbach’
s alpha
Change usually benefits the organization
0.91
I usually benefit from change
Affective dimension (5 items)
0.83
I do not like change
Change frustrates me
Behavioural dimension (5 items)
0.89
I am inclined to try new ideas
Other people think that I support change
11. Transformational (20 items)
Cronbach
’s alpha
Goes beyond self-interest for the good of the group
0.90
Instills pride in me for being associated with him/her
Transactional (8 items)
0.72
Keeps track of all mistakes
Provides me with assistance in exchange for my efforts
Avoidant/laissez-faire (8 items)
0.86
Fails to interfere until problems become serious
Is absent when needed
‘Multifactor leadership questionnaire’ (Avolio & Bass,
2004)
Response scale: (0) Not at all, (1) Once in a while, (2) Sometimes, (3)
Fairly often, (4) Frequently if not always
12. Other measures
A 13-item rating scale measuring perceptions of school
innovativeness in domains such as: school goals,
resources, organisation and management, pedagogy,
etc. (adapted from Johanessen et al., 2001)
Two 1-item rating scales measuring perceptions of the
diffusion of innovative changes within the school and in
the local community (r=0.52)
13. Descriptive results
Teachers:
37% men, 63% women
Age: 23 to 55 years (M=42.5, SD=6.1)
Teaching experience: 2 to 32 years (M=14.8, SD=7.3)
Principals:
76% men, 24% women.
Age: 37 to 61 years (M=49.3, SD=5.1).
Management experience: one to 30 years (M=8.9,
SD=6.7).
14. Teachers
Mean (S.D.)
Principals
Mean (S.D.)
Attitudes to change
Behavioural dimension
Affective dimension
Cognitive dimension
2.6 (0.7)
2.6 (0.8)
2.2 (0.7)
3.0 (0.5)
2.9 (0.8)
2.7 (0.7)
Perceptions of
leadership
Transformational
Transactional
Avoidant/laissez-faire
2.8 (0.6)
2.5 (0.7)
0.8 (0.7)
N/A
N/A
N/A
Scale range: 0 to 4.
15. Predictors
Overall scale
(Mean score)
Affective
dimension
Cognitive
dimension
Behavioural
dimension
FIXED PART
Intercept 0.01 (0.10) 0.01 (0.11) 0.004 (0.10) -0.01 (0.09)
RANDOM PART
School-level
Teacher-level
0.22 (0.11)
0.79 (0.12)
0.33 (0.12)
0.67 (0.1)
0.18 (0.10)
0.82 (0.12)
0.03 (0.09)
0.97 (0.14)
VPC
-2LL
0.22 (22%)
387.94
0.33 (33%)
380.61
0.18 (18%)
388.88
0.03 (3%)
394.39
Preliminary, exploratory analysis
Empty multilevel models for each dimension of attitudes to
change
• Variables have been standardized to have a mean of 0 and a S.D. of
1. This helps interpretation of findings.
• Statistically significant parameters are highlighted in red.
16. Empty multilevel models for each leadership style:
Model 1: transformational leadership:
Teachers’ perceptions of transformational leadership were
more similar within each school than between different schools
(i.e. statistically significant between-school variation).
VPC=31% (significant at 5% level).
Model 2: Avoidant leadership:
Similarly, perceptions of avoidant leadership were more similar
within each school than between schools. VPC=19%
(significant at 10% level).
Model 3: Transactional leadership:
No strong evidence that perceptions of transactional leadership
were more similar within each school than between schools.
VPC=15% (non-significant).
17. Main multilevel analysis
Outcome variable:
Teachers’ average score on the overall attitude
scale
A series of consecutive models were tested defining as
outcome variable the teachers’ mean score on the overall
attitude scale.
Explanatory variables were gradually entered in the analysis.
Every explanatory variable was tested for random coefficient
effects to test whether the strength and/or direction of its
association with the outcome variable differed between
schools.
18. Predictors M0 M1 M2 M3
FIXED PART
Intercept
Transformational (L1)
Transactional (L1)
Avoidant (L1)
0.01 (0.10) 0.01 (0.10)
0.09 (0.08)
0.001
(0.09)
0.42 (0.10)
-0.17 (0.10)
0.001 (0.09)
0.45 (0.13)
-0.18 (0.10)
0.05 (0.10)
RANDOM PART
School-level variance
Teacher-level variance
0.22 (0.11)
0.79 (0.12)
0.18 (0.09)
0.79 (0.12)
0.14 (0.09)
0.73 (0.11)
0.14 (0.09)
0.72 (0.11)
-2LL 387.94 386.79 371.82 349.56
• Significance level set at 10% due to small sample size.
• M1: teachers’ perceptions of transactional leadership explain 18 % of
the between-school variance in their attitudes to change.
• M2: teachers’ perceptions of transformational leadership explain a
further 22% of the between-school variance in their attitudes to
change.
(The above effects are not net of intervening control variables)
Main multilevel results
19. Predictors M4 M5
FIXED PART
Intercept
Transformational style (L1)
Transactional style (L1)
Avoidant style (L1)
Percept. of change diffusion in community (L1)
Land prices (mid-range) (L2)
Teacher’s age (L1)
Teacher’s ICT skills (L1)
Teacher trained in management(L1)
Principal’s foreign language skills (L2)
Principal with postgrad diploma (L2)
Principal’s years in current post (L2)
Teacher age x principal’s language skills (L1xL2)
-0.005 (0.09)
0.47 (0.12)
-0.30 (0.10)
-0.001 (0.10)
-0.09 (0.15)
0.36 (0.11)
-0.25 (0.09)
-0.12 (0.09)
0.21 (0.07)
-0.36 (0.14)
-0.30 (0.07)
0.35 (0.14)
0.65 (0.26)
-0.37 (0.17)
0.45 (0.23)
-0.18 (0.09)
0.30 (0.14)
RANDOM PART
Intercept variance
Teacher-level variance
Slope variance (avoidant)
Intercept-slope covariance
0.19 (0.09)
0.56 (0.09)
0.14 (0.07)
-0.17 (0.06)
0.11 (0.07)
0.46 (0.08)
0.10 (0.05)
-0.13 (0.04)
-2LL 357.25 297.16
20. Y = (predicted) attitudes towards change
X = perceptions of avoidant leadership style
21. Other predictors entered in the analysis, which did not
improve the model:
Teachers’ teaching experience and years in current school (L1)
Teachers’ knowledge of foreign languages, postgraduate
qualifications and training received in change management (L1)
Teachers’ perceptions of the degree of earlier change
implementation in their school (L1)
Number of students in school and student minority intake (L2)
School facilities, indoor and outdoor space, condition of the
building, area (urban/rural/semi-rural) (L2)
Principal’s demographics and years of (teaching and managerial)
experience (L2)
Principal’s training in school management & in change
management (L2)
Principal’s attitudes to change (L2)
Principal’s perceptions of the degree of earlier change
implementation in their school and of the degree of its diffusion
(L2)
22. Synopsis of findings
On average, Greek primary school teachers reported positive
attitudes to change.
The highest scores were observed in the affective and behavioural
dimensions of attitudes, while the lowest (but still high) scores were
observed in the cognitive dimension.
This means that teachers mostly tend to enjoy changes in their
schools and actively support them, but they recognize the benefits of
these changes for the school (or for them) to a smaller extent.
On average, principals reported more positive attitudes to change
than teachers.
Teachers’ perceived their school principals as mainly
transformational in their leadership behaviour, and to a lesser (but
still large) extent, as transactional.
Avoidant leadership did not appear to be a common behaviour of
principals in our sample.
Multilevel analysis showed that 22% of the variability in teachers’
attitudes to change could be explained by school-level factors.
23. Such school-level factors appear to have a stronger effect on
the affective dimension of attitudes, followed by the cognitive
dimension.
Schools do not appear to significantly affect the behavioural
dimension of teachers’ attitudes.
Teachers’ perceptions of their principal’s leadership behaviour
(esp. transformational leadership) explain part of the between-
school variability in their attitudes towards change.
Perceptions of transformational leadership are associated with
more positive attitudes to change and have the strongest effect
on attitudes compared to other leadership styles.
Teachers from the same school are more alike in their
perceptions of transformational leadership than teachers from
different schools.
This may be related to the fact that leaders use
transformational behaviours with all the teachers in their
school, as a group, rather than to selected individuals (having
homogenizing effect).
24. Transformational leaders tend to also adopt transactional
leadership behaviours.
Yet, such transactional behaviours are associated with negative
attitudes to change.
This may imply that transformational leaders often adopt
transactional behaviours to motivate certain (relatively resistant)
individuals to support/accept changes introduced in their schools.
Teachers from the same school do not seem to be more alike in
their perceptions of transactional leadership than teachers from
different schools (non-significant between-school variation).
This may imply that most principals use transactional leadership
behaviours with certain individuals in their schools.
Avoidant leadership behaviours appear to have a negative effect
on teachers’ attitudes only in schools with an above-average mean
attitude score. The opposite effect is observed in schools with a
below-average mean attitude score.
Teachers working in areas with either a low or a high SES, as well
as those whose principal has a postgraduate degree report more
positive attitudes towards change.
25. There is a negative association between teachers’ attitudes to
change and the years a principal has been in its current post. A
possible explanation could be fatigue and gradual loss of
interest/enthusiasm.
The younger the teacher, the more positive their attitudes to
change. The same applies to teachers with ICT knowledge and
who have received training in school organization and
management.
What mostly affects teachers’ attitudes to change is not the
extent of earlier change implementation per se, but the degree of
diffusion of such changes within the community (a proxy of
success).
The principal’s foreign language skills are negatively associated
with teachers’ attitudes to change, but this relationship is
reversed as the age of the teacher increases. This finding
warrants further investigation.
It would be interesting to carry out a multilevel analysis to
examine the factors that explain within-school and between-
school variation in teachers’ perceptions of leadership styles.
Editor's Notes
A single-level analysis (such as running a single level regression model) can often lead to an overestimation of effects - i.e. increased possibility for a type 1 error by underestimating standard errors of coefficients.
2: That is, are teachers’ attitudes more similar within schools than across schools? 4: That is, does leadership have a differential effect on teacher attitudes from school to school?
Transformational and transactional leadership behaviours often appear together, used by the same individual. Changes initiated through transactional leadership are often temporarily implemented, as teachers are not intrinsically motivated/committed to sustain them in the long-term. Oftentimes, it is difficult to find the rewards to motivate people (e.g. in centralized systems such as the Greek education system).Reinforcement through negative feedback can also be problematic – e.g. transactional leaders often feel uncomfortable to give negative feedback on performance. Feedback from the leader can be less important than feedback from colleagues. On the other hand, transformational leaders create a collective, collaborative culture within the school and in so doing, reinforce interaction and feedback from peers.
Teachers working in schools with a history of successful change implementation are more likely to develop positive attitudes towards change. Costly or unsuccessful change efforts affect individuals’ motivation to undertake future change initiatives (perceptions of earlier experiences).
Leadership style was measured at the teacher level, through the perceptions of individual teachers. The same applies to the extent of earlier change implementation within the school, and the degree of its diffusion within and outside the school. School level aggregates were also entered in the model (school averages), but individual level measures had a greater effect that outweighed the effect of aggregated measures. Measuring leadership at the teacher-level helped identify differences in perceptions within the same school (despite the existence of one and common principal/leader) and to start discerning whether the style differs according to the person the principal is addressed to and/or whether perceptions of leadership differ according to some personal attribute of the teacher (e.g. predispositions towards change and innovation).
Total number of schools 193. In the first 25 schools one teacher should be teaching Grade 1, another Grade 3, and another Grade 5, while in the remaining 25 one teacher should be teaching Grade 2, another one Grade 4, and another one Grade 6. In this way, we reduced the possibility of self-selection, i.e. the possibility of only innovative teachers participating in the study. Principals responded to question items measuring school-level variables and teachers responded to questions measuring individual-level variables.
Teachers’ attitudes towards change were measured using the ‘Attitudes towards Change Inventory’ developed by Dunham et al. The inventory comprises three subscales corresponding to the cognitive, affective and behavioral dimensions of attitudes. The cognitive subscale evaluates an individual’s recognition that change occurs and tends to benefit an organization and its members . The affective subscale evaluates an individual’s tendency to enjoy changes in organizations. The behavioral subscale measures the degree to which a person takes action to support or implement change. Different dimensions correlated significantly with r ranging from 0.53 to .55. An exploratory factor analysis (Maximum Likelihood with Varimax rotation) with the Greek sample of teachers gave the same dimensions as the initial scale tested by Dunham et al. (1989).
The principal’s leadership style was measured using the Multifactor Leadership Questionnaire (MLQ, 5X –Short) developed by Bass & Avolio (2004) which was completed only by teachers. Respondents are asked to state how frequently the respective principal behaviour is observed in their school, using a 5-point scale.
Teachers were asked to estimate, using a 5-point scale, the extent to which their school had introduced innovative changes, over the past 5 years, in 13 different domains, such as school goals, teaching methods, the physical environment, etc. The higher the score on this scale, the higher the degree of school innovativeness perceived by teachers. AS THE CONCEPT OF ‘INNOVATION’ IS RELATIVE AND WHAT WE PERCEIVE AS ‘NEW’ OR ‘INNOVATIVE’ DEPENDS ON THE CONTEXT OF EACH SPECIFIC SCHOOL/ORGANISATION, WE DID NOT GIVE A DEFINITION OF ‘INNOVATIVE CHANGE’ TO TEACHERS, BUT EXPLAINED TO THEM THAT THEY SHOULD CONSIDER AS SUCH WHATEVER THEY THEMSELVES THOUGHT AS ‘INNOVATIVE’ FOR THE SCHOOL CONTEXT IN WHICH THEY WORKED.
The scales ranged from 0 to 4, i.e. the mid-point was number 2. Dimensions of attitudes correlated to each other between 0.52 and 0.55. Leadership styles correlated as follows: transformational correlated positively with transactional (0.66) and negatively with avoidant (-0.58). Transactional correlated negatively with avoidant leadership (-0.20).
We first run an empty model for teachers’ score on the overall attitude scale and an empty model for their score on each sub-dimension of attitudes, to check for differential school effects. The results showed that school effects differ between sub-dimensions of attitudes. The affective dimension is the most influenced by school-level factors, followed by the cognitive dimension, whereas the behavioural dimension is the least influenced by school-level factors (non-significant result). In other words, there is evidence that two teachers from the same school will be more similar (on average) on their affective attitudes to change than any two teachers from two different schools. The same applies to the cognitive dimension. Schools, therefore, have a tendency to homogenize teachers’ attitudes to change, but there is evidence that this happens only with regard to the affective and cognitive dimensions of attitudes. The behavioural dimension of attitudes does not appear to be influenced by school-level factors, but mostly, by individual-level factors. These results justify the conduct of a multilevel analysis to examine the determinants of attitudes to change, rather than a single-level regression analysis (at least for the affective and cognitive dimensions).
This analysis showed that teachers’ perceptions of transformational leadership were more similar within schools than between schools by 31%. In other words, two teachers from the same school would differ (on average) by 31% LESS in their perceptions of transformational leadership than two teachers from two different schools. Similarly, two teachers from the same school would differ (on average) by 19% LESS in their perceptions of passive leadership than any two teachers from different schools. And finally, two teachers from the same school would differ (on average) by 15% LESS in their perceptions of transactional leadership than any two teachers from different schools (but, in this latter case, these differences were not found to be statistically significant, and therefore, we hypothesize that teachers’ perceptions of transactional leadership differ equally both within and between schools. An interpretation of this could be that leaders tend to use transactional leadership behaviours only with selected individuals in each school and not with the entire group of teachers working in their schools.
Transactional leadership style has a significant and negative effect only when transformational leadership style is held constant. Transformational leadership and transactional leadership have an r of 0.66, transformational and avoidant leadership have an r of -0.58, while transactional and avoidant leadership styles have an r of -0.20.
Land prices were used as a proxy of SES.
This is true for all dimensions. No significant random slopes for transformational or transactional leaderships styles.
Some of the above variables correlate with one or more of the statistically significant predictors presented earlier. For example, rural areas have more positive attitudes because they tend to be staffed with younger teachers than urban areas. Also, principal’s attitudes to change may be seen as just an aspect of his/her transformational leadership style, and therefore, its effect is outweighed by the variable measuring transformational leadership behaviors. Moreover, principal’s perceptions of the degree of earlier change implementation in his/her school and the degree of its diffusion within the community may affect teachers’ attitudes to change indirectly, through the principal’s leadership style (i.e. by promoting transformational leadership behaviours).