Dutch Labour Force Survey (LFS)

Survey characteristics


To provide information on the relationship between individuals and the labour market. Characteristics of individual persons are connected to their current or future position on the labour market.

Target population

Persons aged 15 years and older in the Netherlands, excluding persons in homes and institutions.

Statistical unit

Persons and households.

First year of survey

The Dutch LFS has been carried out since 1987.


Every three months a limited set of variables concerning the labour market is published. This limited set and a large additional set of variables concerning the labour market is also published on a yearly basis. In addition, monthly figures on the labour force broken down by sex and age (15-25 years, 25-45 years, 45-65 years) are available approximately three weeks after the month under review.

Publication strategy

The monthly, quarterly and yearly figures are always definite.

How is the survey carried out?

Survey type

Sample survey.

Survey method

The Dutch LFS is a so-called rotating panel survey with five waves. Respondents are once interviewed thoroughly, by an interviewer from Statistics Netherlands, by means of a face-to-face interview (Computer Assisted Personal Interviewing, CAPI) or an interview by phone (Computer Assisted Telephone Interviewing, CATI). They are subsequently contacted by phone every three months four times. This survey method was introduced in the last quarter of 1999.


Per household a maximum of eight persons aged 15 years or older are interviewed.

Sample size

In 2011, 62 thousand addresses were drawn.

Checking and correction methods

The information given by the respondent is checked for internal inconsistencies. Overrepresentation and underrepresentation of certain groups in the response is corrected through weighting.


The weighting of the observations takes place in two steps. In the first step inclusion weights are assigned to the observations. These inclusion weights are calculated so that they can correct for biased inclusion probabilities caused by the sampling method. In the second step final weights are determined. This step reduces bias caused by non-response. by weighting for sex, age, country of origin, official place of residence and some other regional classifications. Moreover, information from registrations of country of origin, registration at employment office and income is also used. For the monthly figures all waves are weighted separately, which makes it possible to compute five separate series for the labour force. These series serve as input for the structural time series model.

How accurate are the results?


The outcomes of the Dutch LFS are subject to a margin of error. As the sampling design is complex, estimating the 95% confidence interval is not easy. Because of the considerable relative inaccuracy, figures under a certain threshold are not published. In the tables these figures are replaced by a dot (.). In principle, the minimum for quarterly figures is set at 30 thousand (rounded)in the total column. Subdivisions of groups of 30 thousand persons and over are published. For the yearly figures a minimum of 15 thousand for the total is set. Subdivisions of groups of 15 thousand persons or more are published.

Sequential comparability

Most tables on StatLine contain figures from 1996 and onwards. In some cases it is only possible to publish comparable figures from 2000 and onwards. For a limited number of data on the labour market, sequential comparable series are available from 1988.

For all tables based on the Dutch LFS there is a break in the time series from 2000 to 2001. From 2001 onwards the Dutch LFS figures are based on all five waves. In the years before, the Dutch LFS consisted of only one CAPI wave.

Monthly figures on the labour force are available from January 2003. These figures were published for the first time in July 2010. Before this, three-month averages were used. These figures were available from December 2000 - February 2001.

Quality strategy

As the response is typed into a computer during the interview, inconsistent answers are spotted immediately. For proxy response (i.e. another member of the household answers the questions for the sampled respondent) the interviewer checks whether this person has enough information to answer the questions.