Teaching outdoor and adventure activities: an investigation of a primary school physical education professional development p

Data Analysis 
Analysis of quantitative data.
All quantitative data were analysed using SPSS 
for Windows, version 14.0. Data were manually and statistically searched for 
unexpected values and original data were consulted in order to clarify any unusual set.
Data were presented descriptively as means, standard deviations and percentages and 
where appropriate gender- and age-specific means and standard deviations were 
calculated. The Pearson, chi-square statistics with standard residuals was used to 
investigate any categorical relationships in the data. Paired sample t-tests, or Mann-
Whitney U tests were conducted to compare differences and binary logistic regression 
analysis was used to evaluate children’s perceptual data around physical activity and 
physical education. Relevant effect sizes were calculated and reported as r-value. An r-
value of 0.10, 0.30 and 0.50 represented small, medium and large effect sizes 
respectively (Field, 2005).
Analysis of qualitative data.
Data were coded and categorised using constant 
comparative technique, this facilitated the identification of similarities and differences, 
the grouping of data into categories and the development of propositional statements. A 
journal was kept throughout the process which recorded the analytical process and 
methodological decisions taken. The literature was then interwoven with the data and 
used to confirm or refute findings.
Coding strategy.
The qualitative data gathered at each phase of the project were 
analysed initially, and then tracked individually over time through each phase. All 
qualitative data were transcribed and the transcripts of interviews, observations and field 
notes were entered into NVivo (QSR NVivo Version 8). Nvivo was chosen as it can act 
both as a depository for all data and many simple and more complex searches can be 
automated. Due to the nature of the study and the large amount of data involved, 
NVivo proved an excellent piece of software to store, code, cross code, perform many 

---

104 
analytical tasks as well as providing a central place to hold all notes, comments and 
memos (Figure 3.4). It also became a way of ensuring reliability and trustworthiness in 
the analysis process. Coding took the format of broad to narrow analysis and then 
expanding out again to gain an overall view of the themes developed. At each stage of 
coding any ideas, thoughts, literature relationships etc. were logged as 
memos/annotations and assigned/linked to the relevant data. Each code/category/theme 
etc was then carefully defined and recorded.
Figure 3.4 Screen shot of data layout (as tree nodes) in NVivo 
The following stages were followed through the analysis. 
Stage 1 Broad thematic coding.
This automated coding was carried out using 
the questions from the interview schedules as broad themes for analysis. Similar coding 
was applied to the field notes taken, observations and evaluations as appropriate. These 
were stand-alone themes or categories which were achieved by coding ‘down’ from the 
research question/interview schedule. There was also a list of categories and codes 
based on a visual analysis of the data which was also used when assigning 
categories/codes. 
Parent Node 
Child Node 
Data organised 
into each phase 
Attached 
memo 
Attached 
memo 
Child Node 

---

105 
Stage 2 Cross coding.
The automated coding was examined across each of the 
participants; teachers, children and facilitator notes (and any other data at each Phase) to 
establish common links and/or differences. As the data were analysed further and 
interpreted and read, categories that were discovered and were assigned. These 
categories came from the data rather than the other way around as in Phase 1. Nvivo 
allowed both processes to be combined and coding continued. This constant 
comparative method of analysing data combining inductive category coding with a 
simultaneous comparison of full units of meaning obtained (Charmaz, 2006; Lincoln & 
Guba, 1985; Maykut & Morehouse, 1994). As each new unit of meaning was selected 
for analysis it was compared to all other units of meaning and subsequently grouped 
with similar units of meaning. If there were no similar units of meaning a new category 
was formed (Appendix Ki). 
Stage 3 Grouping and re-ordering data.
At this stage all the themes were 
categorised using the research question and related questions associated with the study.
Hierarchies of categories were also established. Groupings were established where 
codes were matched to themes and the research question. Sub-themes/categories were 
established as they arose and any relationships to main themes or other sub-themes 
identified and linked (Appendix Kii). 
Stage 4 Coding on.
When the data were coded and assigned categories, it was 
important to read the categories’ content and consider whether there were other places 
to code it to. The content was selected and coded at the new or existing category. Ideas 
were beginning to be developed beyond the original coding (Appendix Kiii).
Throughout the process of coding, memo-writing was undertaken. Memo writing is a 
technique advocated by Charmaz (2006), whereby writing memos can ‘catch your 
thoughts, capture the comparisons and connections you make and crystallise questions 
and directions for you to pursue’ (p. 73). Memo writing allows the researcher to 
explore ideas about categories; they allow propositional statements to be developed. An 
example of this is outlined below: 
Text 
He taught the lesson very well. Because I was there, every so often he would look over 
and say ‘is that ok is that alright’ and you just check to see if he is doing the right thing 
but he had it all written down on a card in note form and he had gone through it the 
previous evening with the other third class teachers so he was very familiar with it.