Measures to improve the quality of the LFS survey

On this page, Statbel collects all important methodological comments about the publication of the Labour Force Survey. Quality problems were identified in early 2025 in the results of the province of Flemish Brabant, for which Statbel has taken additional measures. Furthermore, actions were also taken to improve the quality of the occupations, and a new questionnaire was introduced by Eurostat for the publication on lifelong learning.

Quality problems in the province of Flemish Brabant

On 13 March 2025, Statbel published the quarterly figures for the fourth quarter of 2024, and the first annual figures for 2024. A warning was placed for the figures down to the provincial level that the results for the province of Flemish Brabant are insufficiently reliable for the period 2021-2024. Additional details can be found in the detailed explanation.

The employment rate in the province of Flemish Brabant has been on a downward trend for some time now, diverging from the other provinces and the Flemish region.

In cooperation with VDAB Vlaams-Brabant and Steunpunt Werk, possible explanations were sought in certain socio-economic shifts or effects of the methodology without being able to identify one clear cause. An extensive internal investigation revealed that the problem was due to the work of one interviewer in the province of Flemish Brabant.

As a result of these quality problems, the employment rate for the reference years 2021 to 2024 for Flemish Brabant is underestimated. The impact on the figures for the Flemish Region and Belgium remains limited. It does not change the conclusions that can be drawn.

Since it is not the calculation method that is at issue, but the collected data themselves, it is not possible to repeat the fieldwork carried out between 2021 and 2024. This is why Statbel has indicated in its various publications (graphs, tables, etc.) that the published figures for Flemish Brabant are insufficiently reliable.

Statbel takes this problem seriously and has since taken various initiatives to guarantee that the results from the first quarter of 2025 (publication on 10 June 2025) will provide a more accurate picture of the reality of the labour market in Flemish Brabant.

Break in time series of occupations

The respondents’ occupation is an important variable in the LFS (Labour Force Survey). Based on this variable, results are compiled on the evolution of certain occupational groups or on the background characteristics of those occupations.

Specialised coders categorise employed people according to the ISCO classification, with Statbel going down to the most detailed level of 4 digits. The descriptions from the interview between interviewer and respondent serve as the basis for that classification.

In recent years, we have observed an upward trend in the occupational groups of managing directors and chief executives (ISCO 1120), office supervisors (ISCO 3341) and general office clerks (ISCO 4110). Because this trend did not seem to completely match reality, additional internal quality checks were introduced and the coders were given more intensive training to enable them to detect a more specific category.

The extra attention for the specific versus the general codes in the training of the specialised coders, in combination with more intensive monitoring of these and other codings, had a clear impact. We observe a clear decrease in the number of respondents who are categorised in one of the three more general occupational groups. This improvement in quality is also clearly visible in the data, which leads to a break in 2024. Additional information can be found in the note on improving the quality of the coding of the occupations.

Break in time series of lifelong learning

Eurostat urged the various Member States to improve and extend the questions on lifelong learning in the Labour Force Survey (LFS). Statbel also made these changes to the questionnaire, leading to a break in the time series in the results on lifelong learning in 2024.

Lifelong learning refers to the participation of adults (aged 25-64) in education and training (also referred to as formal and non-formal learning). This indicator is usually measured over the last year or the last 4 weeks. Formal learning is education that is recognised by the ministries of education and usually leads to a diploma or certificate such as a master's degree or a training course in adult education. Non-formal learning refers to trainings outside the context of formal learning but still organised in a structured way, e.g. a course on health organised by the health insurance fund or a webinar on time management organised by a self-employed, etc.

Formal learning is usually quite clear: you are either enrolled in a higher education institution or you are not Measuring non-formal learning in a survey is much more difficult: the respondent must know what is meant by non-formal learning and, during the survey, one must remember whether one participated in such a training during the reference period. Different questions in the various Member States resulted in different results, which led to the Eurostat warning.

In an extensive note, Statbel goes into more detail about the Belgian improvements in the questionnaire.