Statistical Analysis for Education and Psychology Researchers

Example from the Literature

In an experimental study on ways to improve the clarity of journal abstracts Hartley
(1994) designed a study to examine the effects of changes in type-size, layout and
language on the perceived clarity of published abstracts. In one of four studies to see
whether formal language is more appropriate for overseas students a 2×2 unrelated
factorial (unbalanced) design was used, one factor being the abstract which was
presented in original vs. revised condition (2 levels) and the other factor was student,
British students vs vs. Overseas students (2 levels). Data were analysed using a Two-way
ANOVA for unrelated measures.
Data represented in Table 8.13 is taken from Table 5 of the author’s original paper.

Table 8.13: Mean cloze (comprehension) scores

(out of 20) for British and overseas postgraduate

librarianship students on the original and revised

version of abstract A

(^) ABSTRACT
(^) Original version Revised version
British students M 7.6 10.5
SD 0.8 2.3
N 5 10
Overseas students M 3.6 6.8
SD 3.3 2.7
N 5 7
Hypotheses tested by the author include: i) main effects for Abstract, that is the means for
the two abstract groups will be equal, ignoring the effects of students; ii) main effects for
students; and iii) interaction effects, that is whether the effects of one variable depend on
the level of the other, for example, one hypothesis might be no different among students
for revised abstract.
The author reported two significant main effects. A main effect is the effect of one
factor ignoring the effect of the other. In this example the significant main effect for
revision of abstract indicated that participants did better with the revised abstracts than
with the original versions, F=7.58, df=1,23, p=0.01. Degrees of freedom for the factor
abstract is, number of conditions−1, (2−1=1), and the degrees of freedom for error are
given by dftotal−dfFactor 1 −dfFactor2−dfinteraction which is evaluated as (27−1)−(1−(1−(1)=23.
The degrees of freedom for the interaction term are: dfinteraction=dfF1×dfF2=(1×1)=1. The
main effect can be seen by examining the body of the table, the differences in means
between original and revised versions of the abstract are evident ignoring the main effect
of student. The main effect of student is also significant, British students had higher
scores than overseas students, ignoring the effect of abstract, F=13.83, df=1, 23, p =
0.001. The author reported no significant interaction effect.
Statistical analysis for education and psychology researchers 334