How sortition works

The citizens’ assembly for democracy was designed as a nationwide process. Anyone with German citizenship aged 16 and over was eligible to participate. When drawing from the sample of citizens, it had to be certain that all potential participants theoretically had the chance to be selected. 

It is possible to draw random samples of the population from the municipal population registers in Germany. This procedure is usually used for lot-based participation procedures, such as for this citizens’ assembly process. As opposed to drawing a random sample from the phone directories, drawing from the municipal registers offers the advantage that even those who do not have a landline connection may be selected.

In order to ensure that every person over 16 with German citizenship could be drawn by lot, a structured random selection was made:

• As a first step, municipalities in all federal states in Germany were chosen from official municipal registers. The registration offices of these municipalities were requested to draw a random sample of their citizens and hand the sample over to the citizens’ assembly for the invitation to participate.
• The selection of the municipalities by nexus Institute and the selection of the citizens by the registration offices both technically took place using algorithms.

The proportion of votes in the Bundesrat serves as a reference for the distribution of the population sample among the federal states:

Selection of municipalities according to size class

The selection was based on the official municipal directory of all political independent municipalities (with municipal association) in Germany (as of 31.12.2018), which is published by the Federal Statistical Office (www.destatis.de).

In the public discussion about political dissatisfaction, reference is often made to the contrast between urban and rural areas, especially when the feeling of being “left out” is analyzed. In order to better reflect the different living conditions among the group of participants, the size of the municipality was taken into account in the first stage of the selection process. 

For this purpose, we positioned ourselves according the municipalities’ size of inhabitants, as used, for example, by the German Association of Cities for the systematic recording of municipalities. To optimize the manageably of the random selection process, we have worked with five size classes. 

The city state level can be directly drawn from the federal state level. In all Flächenländern (federal states which are not city states), one or more municipalities (see below) were randomly selected from these five size classes. The next step was then to take a random sample from the population registers of these municipalities.

Selection of municipalities

The municipalities were selected from the official municipal directory using the “random number” function. In each size class, a selection range between 1 and 3000 was specified so that all municipalities of that size class were assigned a random number. The municipalities then were sorted in ascending order. The municipality that has received the lowest random number was selected. 

If the number of those to be invited to take part in the process was higher than 2 % of the population, the municipality with the next higher random number was included in the selection. The 2 % limit had been set so that families or circles of friends did not by chance represent a municipality for a federal state. 

Since some states have very small independent municipalities, up to 10 municipalities had been selected in the first size class according to this procedure (Schleswig-Holstein). On the contrary, in North Rhine-Westphalia, only 3 municipalities have less than 5000 inhabitants. In this case, the first and second size classes had been merged. 

Size of the sample

We wanted 160 participants to compose of the citizens’ assembly. With an estimated positive response rate of 3 %, 5333 addresses were planned to be extracted from the population registers. The response rate was set so low because, in the following selection process, not all who declare their willingness to participate could have been taken into account. At the end 4362 people from 76 municipalities were invited, because not all municipalities sent data.

Size of samples by federal state and municipal size class

The share of the municipalities of one size class in the total state sample corresponded to the share that the municipalities of this size class have in the total of all municipalities of the federal state.

Composition of participants according to criteria

The random selection of participants guarantees diversity. This cannot be achieved with other procedures based on public invitations or invitations from interested parties. Participation procedures based on a lottery are considered to be particularly inclusive, but even with this method, the older age groups and the highly educated tend to be overrepresented. In order to counteract this, the participants are compiled according to positive responses. 

The decisive factor here was that the distribution of socio-demographic characteristics in the citizens’ assembly should correspond as closely as possible the population as a whole. 

The following characteristics were considered: 

• Gender
• Age group
• Level of education
• Federal state
• Size of municipality
• Migrant background

Some of this data is already found in the data of population registers. In some cases, they were derived from the response forms that all invitees received.