Microsoft Word juntunen väitös docx

6. VALIDITY AND RELIABILITY 
The characteristics of good design research (Pernaa, 2013, 20) have guided this design 
process. The conducted design research was holistic in how it included both theories and 
models for guiding and describing the research. The models and theories were based on cyclic 
design and the testing and evaluation of the research. This design research project produced 
transferable and authentically-tested educational solutions through well-documented cyclic 
phases in Section 4. (Pernaa, 2013, 20) 
The particular epistemic values guiding scientific research are: reliability, testability, 
accuracy, precision, generality, simplicity of concepts and heuristic power. These values 
define the character of science as a distinct way of knowing (Allchin, 1999). This chapter first 
discusses the validity aspects and then the reliability aspects of this dissertation, both of which 
were considered throughout the problem analysis and other phases of this design research 
project (Edelson, 2002).
Validity refers to how well an instrument measures what it is purported to measure. In this 
thesis, accuracy was considered by inspection of the research frame and how the chosen 
methodology corresponded to the nature of the studied phenomena and the research questions 
that were set. The accuracy is explained in terms of internal, external, structural and content 
validity. 
Internal validity dictates how an experimental design is structured and encompasses all of the 
steps of the scientific research method (Anttila, 2006). In this thesis, internal validity was 
increased by using methodological triangulation on the qualitative and quantitative research 
tools. Besides methodological triangulation used in 
Articles
I–IV
, the multiple levels of 
analysis were drawn from the mixed methods used in 
Article
II
. These levels touched upon 
the individual, the community and the global society. Such a multi-level approach increases 
both the validity and the reliability of the conclusions.
Content validity (relevance) is the extent to which the measurement method covers the entire 
range of relevant factors that define the construct being measured. The structural validity 
(consistency) of the analysis concerns the question of whether the variables provide 
information on the relevant aspects mentioned in the theoretical framework. (Anttila, 2006) 
Content and structural validity were both increased by extensively analysing cross-curricular 
educational research material. In this dissertation, the theoretical problem-analysis represents 
mainly well-known and significant educational theories, leaving out single articles and 
weaker theories that may well be of interest in their own right. The theoretical basis seems to 
be well-saturated. It is limited to the core concepts of this thesis. Because the measurement 
tools used in this thesis are selected and formulated on the basis of the theoretical problem-
analysis, it is assumed that the used tests assess the main domains of the selected criteria.
To improve internal and content validity, the cyclic design process was visualised (see Figure 
6.). The visualisation clarifies the decision-making and progression of the thesis.

---

41
External validity refers to generalisation of the findings (Anttila, 2006). The group of 
participants in this study was rather small, which makes quantitative generalisation of the 
results impossible. However, the collaborative design of the teaching concepts together with 
the chemistry teachers and the testing of the concepts in authentic school settings partly 
answer the challenge of testability and generalisation. Representative conclusions can be 
drawn at least in the context of the research problem of the dissertation.
The results also seem to corroborate previously published educational research and, therefore, 
have a high face-validity. In other words, the results logically support the usefulness of 
product life-cycle analysis in particular as a cross-curricular approach to ESD in chemistry.
In terms of reliability, the aim was to select reliable research tools from the reviewed 
literature. Reliability is the degree to which the assessment tools produce stable and consistent 
results. The reliability of a method is assured if it produces reproducible results when applied 
to a sample. Reliability stems from the individual instruments used, and the high quality of 
the analyses and arguments presented in a study. (Anttila, 2006) The individual empirical 
methods that were selected for this study are briefly presented in Section 4. and analysed 
more deeply in