Two surveys published in November – on private school graduates and foreign domestic workers in Singapore – drew contrasting interpretations, raising questions about how statistics is used. And more specifically, on the statistical issues of sampling, as well as study or survey design.
The first concerns the employment prospects of graduates from the private universities. Across all the three employment indicators of overall employment rate, full-time permanent employment rate, and median gross monthly salary, a survey by the Committee for Private Education (CPE) found that these graduates underperformed their contemporaries from the autonomous universities. This survey found, for instance, that only 60.1 per cent of private school students found full-time work. This is 19.8 percentage points lower than the fresh graduates from autonomous schools. At $2,550, the gross monthly salary of the median private school student is also $775 lower. Taking issue with these findings, the Singapore Association for Private Education (SAPE) instead pointed to “good evidence” of private school graduates enjoying salary increments, and further argued that there was a “quantitative bias”, since students who enrol in private or autonomous schools do not have the same backgrounds.
The second survey relates to the well-being of foreign domestic workers (FDW) in Singapore. A report – titled “Bonded to the System” and published by independent consultancy Research Across Borders – documented the working conditions for these workers, suggesting “some 60 per cent of maids … are exploited by their employers”. Exploitative conditions include, the CNN added in its news report, excessive hours or days, poor living conditions, and low to no salary, and these are structurally premised upon the unequal power relations between the workers and their employers. In response, the Ministry of Manpower (MOM) said the survey paints a “misleading picture”, and pointed to its own regular surveys to “better understand the employment conditions and well-being of FDWs”. In interviews with ST, furthermore, some non-profit organisations added that the findings from the Research Across Borders study contradicted either their interactions with the FDWs or their own survey findings.
Paying attention to how statistics is used should, to a large extent, resolve the apparent contradictions of the reactions to both surveys. Sampling is the first statistical issue. Statistically, because it is rarely feasible or cost-effective to reach out to a large population, sampling of a subset – randomly and properly conducted to minimise bias – therefore allows for observations to be made about the population. The CPE survey had two sampling issues: First, the low response rate of 32 per cent, compared to the 70 per cent rate for the autonomous universities and the polytechnics, which means the actual employment rates and monthly salaries of the population may be even lower; and second, the disproportionate representation of respondents from the Singapore Institute of Management (SIM). The 2,109 respondents from SIM constitutes 60 per cent of the total, and despite the large school sample the survey does not compare across degrees or disciplines.
This comparison across degrees or disciplines is important, since the intent of the CPE survey is to help prospective students make more informed decisions. In addition, the CPE should be concerned that only four private education institutions had at least 49 respondents, and that response rates across the board were very low. The SAPE has a point too on the “quantitative bias” – “the assumption that people who enrol in the autonomous universities are similar to those who pursue courses in [the private schools]” – such that the findings from the private schools may not be comparable to the those of the autonomous ones. A distinction has to be made, moreover, between fresh graduates and those who return to the university after some years in the workforce, and in this vein it would be possible to ascertain whether in the latter group do earn higher salaries after upgrading their academic qualifications. Though on this note, if the SAPE alludes to contrarian evidence of private school graduates enjoying salary increments, the association needs to be more transparent with its methods.
Sampling could partly explain the differences between the surveys produced by Research Across Borders (which used convenience sampling, a non-probability method which increases the likelihood of sampling bias) and those by MOM or the other non-profit organisations (which has not clarified how their samples were assembled). Stratification further ensures that the sample is representative of the larger population, based on their race or ethnicity and their countries of origin. Though a more probable issue here is study or survey design.
Concerns over study or survey design includes questions on how respondents are prompted and how they may be influenced during the data collection process: Who the data collectors are, and how they introduce themselves (the responses are likely to vary, if the data collector is representing a government agency, a research agency, or a non-profit organisation); whether survey questions are positively or negatively framed (in the context of FDWs, Research Across Borders operationalised indicators of labour exploitation, whereas the different surveys by MOM and the non-profit organisations tend to instead focus on levels of satisfaction and whether they would recommend Singapore as a country to work in); as well as how the survey questions are constructed and collectively presented to the respondents (if they proffer consistent responses to questions anchored by similar themes).
Delving deeper into the survey questions and questionnaires should, in this regard, prove more instructive.
Tied to study or survey design is the purpose of the study, which in turn is based on the purpose of the organisation. Put otherwise, on the employment prospects of graduates from the private universities: What is the mission and vision of CPE – as a regulator of the private education sector – vis-à-vis the SAPE’s role as an advocate for the sector? On the well-being of FDW in Singapore, organisational incentives vary too. Paying heed to the aforementioned statistical issues of sampling as well as survey and study designs is an exercise of healthy scepticism, to not take surveys for granted, and to perhaps find a constructive middle ground between the contrasting interpretations.