referred in Cohen and Manion (2000: 144), apply to all mailed questionnaires. The
first limitation refers to whether respondents have given accurate answers to the
questions they were provided with. As the questionnaires were anonymous, it was not
possible for the researcher to verify how accurately each person had responded.
However, this limitation was partly overcome through the classroom observations and
interviews which were planned in order to allow the researcher compare results
coming from different sources. The second limitation refers to whether the persons
who did not respond to the questionnaire would have provided the same distribution
of questions with the ones who responded provided that, as Moser and Kalton (1971:
267-268) stress, it has been repeatedly confirmed in practice that ‘…people who do
not return questionnaires differ from those who do’. Finally, as far as the observations
are concerned, despite the fact that the observer made every effort to pass as
unnoticed as possible, there is always the possibility that either the teacher or the
students observed were affected by the his presence, even subconsciously. Labov
(1970) identified that the mere act of observing people’s language behaviour is
inclined to change that behaviour, coining the term ‘the observer’s paradox’.
A final limitation to the study refers to the analysis of data and the way these
are presented. Thus, it should be noted that the small number of questionnaires which
were collected makes difficult the valid classification of results into categories as,
sometimes, a category was established as a result of a single response. Moreover, as it
is stressed in the literature (Research Methods in Education Handbook: 2001: 83), the
fact that small numbers are converted to percentages, can have misleading effects as
small differences appear as dramatically large percentages. For example, a difference
of 3% between two categories represents the difference in response of just one person.
To overcome this problem, the tables which accompany the charts in appendix IV (see
pp. 110-131) provide the reader with the actual number of respondents for each
question.
2.5 Ethical issues considered
Before the research was carried out, both the supervisor and the school advisor
were asked for their consent as part of the research involved visits in schools and
classroom observations.
During questionnaire construction, care was taken so that the questions were
not biased by the researcher’s own beliefs and that they were representative of all the
40