NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Labour market expectations, relative performance and subject choice
Initial registration date
August 07, 2014
August 07, 2014 4:24 PM EDT
University of Birmingham, UK
Other Primary Investigator(s)
Additional Trial Information
This trial tested the effect on the subject choices of secondary school students in English schools of providing them with information about the variation in graduate salaries by undergraduate subject studied. We ran a cluster randomised controlled trial in 2011-12 in which 15-16 year old students were provided with information about graduate salaries through an activity-based lesson lasting one hour. The lesson was taught by the normal class teacher using resources provided for the intervention. The information on graduate salaries was taken from a study by O'Leary and Sloane (2011) which used data from the Labour Force Survey. The education system in England encourages relatively early specialization. Most students who remain in full-time education between the ages of 16 and 18 study a small (3-5) number of subjects which are designed to prepare for undergraduate study. Undergraduates typically study their major from their freshman year onwards. We measured the effect of the intervention on students' choice of subject using the records supplied by the school showing which subjects students were studying in 2012/13 when the students were aged between 16 and 17. Data were collected by questionnaires before and after the intervention lesson. These data were linked to data available on the National Pupil Database. We were therefore able to control for a range of student characteristics including prior achievement, parental education, ethnicity, cultural capital and expressed intentions towards subjects they might study.
A priori we conducted a power calculation assuming an intra-cluster correlation coefficient of 0.1 and an average school size of 83, which suggested that 48 schools would give 80% power to detect a 0.3 difference in outcomes using a two-tailed test at α=0.05. We generated a pool of potential participants through a list of all schools within a large and diverse geographical area within England which satisfied our criteria: serving the age range 13-18 and having at least 100 students in their ‘sixth forms’ (students in the academic year for students aged 16-17 and the academic year for students aged 17-18). We also stratified our sample to include 20 private schools and 30 state schools. These criteria meant that the average achievement of students in our sample was significantly higher than the average for all 15-16 year-old students in England. Schools satisfying these criteria were placed in a random order and approached in that order until we had secured firm agreement for participation. The total project sample included 50 schools and 5,593 students. Ten schools withdrew part way through the project. Six schools (571 students) allocated to the intervention arm did not take part in the intervention and four schools (487 students) allocated to the control did not take part in the questionnaire administered after the intervention. As per the recommendations for reporting randomised trials (Moher et al. 2001) we report results using an intention to treat analysis. We used the mi routines in Stata to impute missing data. Analysis of the trial data showed a positive effect on choosing to study math and a negative effect on choosing study biology or art.
In the intervention arm 15-16 year old students were given information on variation in graduate salaries in England. The students were told to assume that they were leaving school with at least the qualifications regarded as necessary for admission to university in England. Students were divided randomly into two groups. One group were told they would be making decisions as males (and given data for males) and one group were told they would be making decisions as females (and given information for females).The information was provided through an activity in which students were asked to choose between successive pairs of options.The information provided was based on the data on graduate salaries provided in O'Leary and Sloane (2011) scaled up to match evidence of average graduate salaries for a thirty year-old at the time when the intervention was conducted. The first choice for students was between not going to university or going to university to study Art. Students were then told the salaries for each. They were then given an option to switch from whatever they had chosen in this round to studying a History degree. In successive rounds they were given options to switch to other subjects including Mathematics, Science, Engineering and Languages. This activity was followed by two more which presented the same data in different formats and posed questions to the students which they answered in pairs or on their own. The lesson was taught by the normal class teacher using resources for the activities and teachers' guidance notes provided by the project.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
The primary outcome for the intervention was choice of subject to study when students were aged 16-17 in the academic year following the intervention. Our analysis concentrated on differences between the intervention and control arms in the likelihood of studying each of the following subjects: art, biology, business, chemistry, computing, economics, English, geography, history, languages, mathematics, physics.
Primary Outcomes (explanation)
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Baseline data were gathered through a questionnaire and matched with data from the National Pupil Database. The baseline data collected evidence of demographics and students' intentions towards subjects to study the following year. A shortened version of the questionnaire gathered evidence of intentions and beliefs about the graduate premium immediately after the intervention. Our analysis compared the likelihood that a student would be studying each of several advanced level subjects one year after the intervention, taking account of demographics and intentions expressed before the study.
Experimental Design Details
The randomization was carried out by an independent researcher at the Medical Trials Unit of the University of Birmingham
The unit of randomization was the school.
Was the treatment clustered?
Sample size: planned number of clusters
Sample size: planned number of observations
Sample size (or number of clusters) by treatment arms
25 schools to treatment, 25 schools to control
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
A priori we conducted a power calculation assuming an intra-cluster correlation coefficient of 0.1 and an average school size of 83, which suggested that 48 schools would give 80% power to detect a 0.3 difference in outcomes using a two-tailed test at α=0.05.
INSTITUTIONAL REVIEW BOARDS (IRBs)
University of Birmingham Ethics Committee
IRB Approval Date
IRB Approval Number
Post Trial Information
Is the intervention completed?
Is data collection complete?