Page 216 - Invited Paper Session (IPS) - Volume 2

IPS215 Elizabeth S. et al.
2015 PISA Dataset: The 2015 study was the sixth cycle of PISA since it was
first conducted in 2000. The present study presents results for Australia from
a total of 14 530 student participants with 7163 females and 7367 males.
Australia took a larger sample than required in order to oversample smaller
jurisdictions and Indigenous students to ensure reliable estimates. The main
mode of assessment changed from a paper-based delivery to a computer-
based delivery with students completing a series of questions with stimulus
material comprising a short written passage or text accompanying a table,
chart, graph, photograph or diagram. A range of item-response formats
including multiple-choice and open ended responses were used to cover the
full range of cognitive abilities and knowledge identified in the Assessment
Framework.
In addition to the standardised student questionnaire that students were
assigned, an additional student questionnaire offered was the educational
career questionnaire which sought information from students about their
learning habits and preparation for their future career with 6863 students
completing the additional career questionnaire, representing 47% of the
original sample. One question asked about additional instruction in school
subjects attended by students in the school year. This instruction could take
place at school or somewhere else, but is not part of students’ mandatory
school schedule and students were asked to consider all regularly attended,
institutionalised, organised additional learning activities in which they received
some kind of instruction, guidance, or support and to approximate how many
hours per week they attended the additional instruction in mathematics.

Statistical Analyses
Variables were used on two different levels in the proposed analyses, with
student mathematics anxiety measures, gender and preferred learning
methods measured on the individual student level, while school was used at
class level. The effect of gender on anxiety and preferred learning methods
will be assessed descriptively using percentages and chi-squared tests and will
be validated using multilevel binary logistic regression models, also known as
generalised linear mixed models, to take into account the hierarchical
structure of data with students nested within school groups, and
simultaneously model data from the different levels. Multilevel regression
models will be used to assess differences in time spent in additional instruction
in mathematics between genders.

203 | I S I W S C 2 0 1 9