This week we have been carrying out a qualitative analysis of a subset of the data resulting from the School Leavers Study, 1978.
The full dataset consists of 141 essays, too many to analyse in a single week, especially on a part-time basis. I chose to analyse the first twenty boy’s essays and the first 20 written by girls. I made an assumption that the data is in no particular order and that selecting the first twenty would be no more or less random than any other means of selection. Forty essays turned out to be far too ambitious as I have documented below.
There are no references at the bottom of this blog post because I’ve deliberately avoided looking at anyone else’s secondary analysis of the data so that I can determine for myself what it might be possible to infer from the data.
I used the online application Dedoose to code the data. Importing the essays and the table that provides demographic information about the participants was straightforward, as was linking the demographic information*, to the relevant essay (*described as descriptors in the software).
This has been a hugely informative exercise. As someone who prefers to learn by doing (and by learning from mistakes along the way) simply ‘having a go’ at coding made me realise how powerful it can be when applied appropriately.
The first insight for me was that creating new codes as I progressed through the data resulted in frequent returns to all of the previous data to find out if there was evidence of the code I had just applied. While this was frustrating at times it was interesting to observe my own thought processes emerging as I wrestled with how I might mine the most useful information from of the data. For this reason, I would probably work through this iterative process with any future coding exercise where I wasn’t familiar with the data, although giving some thought to codes that could be created apriori would save a lot of time.
The second insight into the coding process came when I realised I had made some of my codes too high level. For example, I started off coding mentions of having children, but later felt it might be useful to know when participants thought they would start a family and how many children each respondent was predicting they would have (or whether they mentioned children at all). That depth of empirical data would be useful in any comparative study, both with other groups and between the male and female participants in this study. This meant that some codes required a degree of mental arithmetic before they could be applied, such as working out what age the participant predicted they would be when they left home or got married.
I found that it’s easy to get caught up in the longer more descriptive essays and I was concerned that they might become more dominant in the output, so I tried to focus on the basic information that is present in both the more mundane essays and the sensational accounts.
It wasn’t until I had figured out that I needed to add a whole range of child codes into the high-level categories that I got a realistic feel for the amount of work required to do justice to the wealth of information in the data. Forty essays started to feel like I’d taken on far more than I could cope with in the time available and I could see why we’d been instructed to read 6 to 10!
Most of the essays I coded revealed relatively low-level aspirations, with many participants predicting that they would find it hard to get a job and would experience periods of unemployment and other adversity. However, in the subset of essays I analysed, there were also references to overcoming adversity through hard work, this was particularly evident amongst the boys. Most of the boys predicted starting jobs in engineering or low-level clerical roles and most of the girls expected to work in secretarial, clerical or retail. One of the boys essays was particularly telling; it was evident that he had aspirations of becoming a racing driver, but he expected his dreams to be thwarted by his economic background. Both the girls and boys essays referred to wives having to give up work to look after children and returning to work as soon as the children were of an age were that was possible.
All but two of the boys referred to having their own accommodation and many of these referred to owning their own house, despite their relatively low-level career aspirations.
Most of the boys also expected to get married (more than once in some cases) and of the essays I coded all of the boys who mentioned marriage also expected to father children. I started to code how many children participants predicted but I haven’t had time to finish that analysis, however, there appears to be a surprisingly frequent mention of twins amongst the girl’s essays and a few mentions by the boys. This looks like an example of memory being at odds with the actual data and the subset not reflecting the whole dataset. A quick search for the word ‘twins’ across all of the essays reveals only thirteen mentions in total, clearly I happened upon more than would be expected looking at all of the data.
Where the boys reflected on their lives many saw themselves as having failed and very few seemed to have a positive view (or any view) of what retirement might hold, with one going as far as expecting to end it all when they reach sixty. In general the girls seemed to be more content with how their lives had panned out, with many taking satisfaction in watching their children grow up and life events such as their own and siblings weddings.
What was noticeable by its absence was mention of possessions. Other than owning a house, a car, a small garden, or going on foreign holidays, few mentioned any of the luxuries youngsters might expect to have in their lives today. In the essays I analysed the most ambitious had hopes of owning a small yacht and one expected to be able to rent out a house he had bought as well as buying a flat.
What questions could be answered using this data?
I can see that there would be risks in inferring too much from a secondary analysis of this data. There’s a lot that we don’t know about the participants and the nature of the exercise means that there are gaps in many of the accounts. For example, it would be risky to infer that a participant who doesn’t mention marriage or children had no expectations of either.
What the data does seem to be useful for is analysing differences in expectations between the boys and girls and what appears to be important to each group. It would also be possible to draw some conclusions about how boys saw girls lives unfolding and vice versa and as such their attitudes toward the opposite sex.
At the time the data could have been used as evidence to support the need for interventions such as career advice, or the provision of training to enable school leavers to pursue careers in areas other than those that would have been in recession in the late 70s.
If I was conducting a project using this data, what would I want to do next?
There are a few follow-on projects that would be fascinating to pursue. A longitudinal study of students in the same area would enable researchers to show how attitudes and aspirations have changed over time.
Comparative studies with similar datasets (if they exist) might demonstrate differences and similarities with other areas that had different social and economic conditions.
And it would, of course, be interesting to revisit this particular group of school leavers and compare their actual life-course to their imagined version.