The Scottish Government

« Previous | Contents | Next »

Listen

APPENDIX D: TECHNICAL DETAILS OF THE MAIN SURVEY ANALYSIS

This Technical Appendix refers to the analytical framework for the data from the main survey.

Sampling Strategy

The sampling strategy and details of the final sample size and breakdown by level of study is contained in Chapter 1.

Weighting Strategy

The data was initially broken down by level of study into the groups which we were primarily interested in i.e. Full-Time HE College, Full-Time HEHEI, Part-Time HE College, Part-Time HEHEI, Part-Time HE Open University, Full-Time FE, Part-Time FE. It became clear early on that we had insufficient responses from PTFE students to draw any valid conclusions regarding their behaviour and we therefore decided to look only at FTFE students. We then decided to weight the other groups in relation to the population totals for FTHE, PTHE and FTFE. The FT totals were further separated for weighting by whether they were degree or sub-degree students. The four populations were weighted separately, giving in effect four different surveys. This separation was done because institution and student participation varied so much between these groups: while full-time degree only students across Scotland participated in reasonable numbers, participation by sub-degree HE students was less representative across Scotland (most coming from a small number of participating FE colleges). In order to protect the results for full-time HE degree only students, the other populations were dealt with separately. In terms of the analysis, in order to be able to discuss figures for the FTHE sector as a whole, a weighted mean was taken of the degree and sub-degree means where necessary. The weights here were again based on the population totals found in the HESA and FES data

The final totals for each group are as follows:

FTHE:

Degree only 3959

Sub-Degree 372 (of which 116 attended HEIs and 256 attended Colleges)

Total 4331

PTHE 521

FTFE 114

The population totals which were used to weight the variables were the relevant HESA and FES totals for the year 2006-07 which is the most recent available data for both sectors.

In order to check whether the so weighted survey was biased by social class, a check was made on the proportion of students in receipt of the Young Students Bursary in the survey and it was found that the proportion in the survey was less than that of the student population, as indicated by data from SAAS. In this regard the results from the survey may not accurately reflect the finances of students from lower income background.

Weighting Variables

A number of variables were considered in this respect including, gender, age, level of study and year of study. Given the size of the sub-samples it was not considered sensible to use all these variables in each case. The actual weighting variables which were used for each of the sub-samples can be seen from the weighting tables which follow. Where the year of study was not given, the cases were distributed across the other years in proportion to the known sample totals in each year. Full-time sub-degree students reporting 3+ years were considered to be in year 2.

Trimming

The initial weights were examined to see if there were any very large or small weights which might bias the analysis. On examination it was found that only the weights for PTHE males studying at college had very high weights and they were trimmed to 4.5 in each case ( i.e. for the two age bands). There were no weights below 0.25 and therefore there was no trimming at the bottom end.

Final adjustment

The final sets of weights were then adjusted to bring the un-weighted and weighted totals back into line for all the sub-samples.

Weighting Tables

The following tables contain the weights for all of the sub-samples:

Table D.1: FTHEHEI

Table D.2: FTHE College

Table D.3: PTHEHEI

Table D.4: PTHE College

Table D.5: PTHEOU

Table D.6: FTFE

Construction of Variables

The grouped variables were made up as detailed in the tables below. Each variable described here corresponds to a question(s) in the survey recalculated over the academic session.

Time Periods

The survey respondents were asked to give figures for all the income, expenditure and debts questions in weeks, months or years in order to make it easier for them to reply and to maximise the completed responses. These responses then had to be annualised ( i.e. defining a year as an academic session). We therefore needed to define the length of an academic year. In discussion with the Scottish Government it was agreed that the most common length of term was 36 weeks and the data was therefore annualised on a 36 week or nine month basis.

Capital expenditure

A very small number of students reported significant items of capital expenditure i.e. car purchase. This figure was discounted using readily available information on the depreciation rate of cars in order to annualise these figures.

« Previous | Contents | Next »