This is the second step of a research project about the diffusion of statistical knowledge among young
generations, in which we focus our interest on statistical contents of primary school textbooks. From a
precedent survey, we have seen that teachers think that they do not contain sufficient information
about data analysis and forecasting of uncertain events, and so the teaching materials are mostly
taken from other sources. In the Italian primary school, statistics is placed within the mathematical
curriculum, with probability, measures and relationships, and the textbooks present about 6 pages
dedicated to statistics and probability at the end of the Math General book. Now we analyse a random
sample of 4th and 5th grade textbooks, edited in the last few years in Italy. By using a mixed-methods
approach we carry out the content analysis of these texts, starting from an ex ante conceptual grid.
This grid allows us to observe the contents of the pages dedicated to the subject. From an inspection
of the pedagogical and statistical literature we identify the principal characteristics (vocabulary,
structure, graphic system, technological links, exemplifications, etc.) and the statistical tools (mean,
mode, median, diagrams, tables, etc., and the mistakes in their presentation) which can not be missed
in a good schoolbook for the specific school level. Initially we use the CAQDAS package ATLAS.ti: this
scientific software makes our job easier. Indeed, the application of that package permits us to describe
accurately the relationships between structural and content aspects of the analysis units, i.e. the textbooks,
in which, according to a classical content analysis approach, a lot of characteristics are
identified. Subsequently, the codes obtained from the computerized content analysis are transformed
in the typical statistical data set: the matrix of data, in which the cases (each textbook) and the
variables (each category created in ATLAS.ti) are read and analysed by PASW Statistics package.
The aim is measuring the prevalence of the various textbooks characteristics and the level of
association between each other. The application of canonical non-linear correlation analysis
(OVERALS) to the set of these textbooks characteristics permits the spotlight of content typology. The
conjoint use of multi-method techniques - quantitative survey, content analysis, multivariate statistical
analysis - puts into act the “triangulation”. In a methodological perspective multi-methods, triangulation
can produce interesting results about the statistical contents of school textbooks and their evaluation.
