gms | German Medical Science

GMS Zeitschrift für Hebammenwissenschaft

Deutsche Gesellschaft für Hebammenwissenschaft e.V. (DGHWi)

ISSN 2366-5076

Achieving a random sample: The potential of field access via residents’ registration offices

Research Article

  • corresponding author Manuela Raddatz - University of Applied Sciences Osnabrueck, Germany; University Witten/Herdecke, Germany
  • Uwe Bettig - Alice Salomon University of Applied Sciences Berlin, Germany
  • Claudia Hellmers - University of Applied Sciences Osnabrueck, Germany; University Witten/Herdecke, Germany
  • Friederike zu Sayn-Wittgenstein-Hohenstein - University of Applied Sciences Osnabrueck, Germany; University Witten/Herdecke, Germany

GMS Z Hebammenwiss 2023;10:Doc04

doi: 10.3205/zhwi000028, urn:nbn:de:0183-zhwi0000280

This is the English version of the article.
The German version can be found at:

Received: April 1, 2022
Accepted: October 5, 2022
Published: December 20, 2023

© 2023 Raddatz et al.
This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 License. See license information at


Background: In midwifery research, mothers are an important study group which is often hard to gain field access to. In quantitative research, questions relating to field access are linked to considerations concerning how to achieve representativeness. One – albeit insufficient – prerequisite for achieving representativeness is drawing a random sample. To date, studies that have taken random samples from population registration data do not describe this procedure in detail.

Objective: To share methodological experiences from a cross-sectional survey of mothers that used field access via residents’ registration offices (RRO). Furthermore, the article seeks to promote critical reflection on methodological decisions in midwifery science.

Methods/material: Research progression data from the “ROSE” research project have been used. The “ROSE” project was funded by the Lower Saxony Ministry of Science and Culture and implemented at Osnabrück University of Applied Sciences. The overarching goal of the maternal healthcare sub-project was to determine to what degree maternal health-care services in the Osnabrück-Emsland region are needs-based.

Results: The success of sample collection using RRO depends on knowledge of the legal situation as well as on flexible study conditions. In the ROSE project, despite the achievement of a random sample, there is no coherence between target population and inference population.

Conclusion: Obtaining address data via RRO is an option that must be included in research decisions in order to achieve a random sample in quantitative surveys. When it comes to achieving representativeness, however, further methodological approaches are required.

Keywords: access to the field, random sample, representativeness, residents’ registration data


In midwifery research, women of reproductive age are an important study group. There is a wide range of different research interests related to this group: What is this phase of women’s lives like for them – what healthcare situations do they encounter and what specific experiences do they have? What are women’s healthcare needs when it comes to pregnancy and childbirth, what healthcare services do they use and in what form?

Irrespective of which methods are used to address research questions – whether qualitive, quantitative or a combination of the two – the key question in research planning is always how the required data can be collected and analysed. To this end, firstly the research field has to be defined, e.g. target group, suitable sample or sampling procedure, as well as suitable field access [5], [14], [18]. Given that not all eligible individuals can participate in a survey, it is essential that a suitable sample is identified, comprising a realistic representation of the parent population. After defining the research field and identifying the sample, the next issue that arises is access to the field. Atteslander categorises decisions on field access as part of determining the research subject, as they mirror decisions about sample composition ([2], p. 32ff). Approaches that are geared more towards practical research tend to deal with field access under feasibility aspects. In day-to-day research practice, a balance has to be achieved between these two poles, in which methodological considerations regarding the application of quality criteria to the research approaches being used should be given priority [10]. The external validity quality criterion, with its requirement of generalisability of results, is seen as central in quantitative research [11] and as such plays a crucial role in the planning of field access. This article describes field access to mothers for a postal cross-sectional study based on the example of the research project “ROSE – the learning healthcare system in the Osnabrück Emsland region” (“ROSE” stands for Region Osnabrück-Emsland). It also discusses the methodology for such field access with a particular focus on generating samples. The main aim of the ROSE project, based at Osnabrück University of Applied Sciences and funded by the Lower Saxony Ministry of Science and Culture, was to link regional healthcare research and practice. This aim was pursued in seven regional sub-projects from the field of healthcare research. The overarching goal of the maternal healthcare sub-project was to determine to what degree maternal healthcare services in the Osnabrück-Emsland region are needs-based.

Methodological background

The aim of quantitative research is to extrapolate the general from the specific, whereby the focus – in contrast to qualitative research – is on statistical generalisation and not theoretical. When using the term generalisability, it is important to take the reference point into account. At this juncture, an explanation of the term parent population is therefore warranted. The parent population refers to all objects of study on which a statement is to be made. Defining the parent population precisely is of crucial importance because it prevents false conclusions from being drawn [19], [24]. In healthcare studies, the definition of the parent population might include variables such as age, gender, specifics regarding health and healthcare, as well as precise temporal and geographical limitations. The results can only be generalised to the predefined parent population and not to a broader population, for example, to another age group or other healthcare contexts. If this were to happen, it would result in over-generalisation and incorrect conclusions would thus be drawn [19], [24].

In the healthcare sciences, as in the social sciences overall, it is rarely useful or even possible to study the population of interest in its entirety In such a case, every single unit of the parent population is included in the study. This poses two challenges related to the size of the parent population. If a large group of people are studied, this group is, first, not readily or easily accessible in its entirety and second, incorporating all individuals incurs disproportionately high costs. Drawing a sample can help resolve this problem. Based on the sample, conclusions can be drawn about the parent population using statistical methods – the results are generalised to a defined parent population. To make it possible to draw such conclusions, the sample must fulfil two criteria: first, it must be big enough and second, it must be representative [11].

Mathematical or statistical methods can be used to decide on an adequate sample size. The crucial consideration here is with how much confidence you want to be able to make your statements, in other words, what your designated confidence level is. The required sample size can be calculated using the relevant formulas, irrespective of the size of the parent population (see Figure 1 [Fig. 1]) ([2], p. 264ff). Statistical programs such as G*Power allow for exact a priori sample size calculations depending on the planned statistical tests and desired effect sizes [13].

Achieving representativeness is more complex. A sample is representative if its units and the units of the parent population are similar when it comes to key variables, that is if the sample represents a miniature image of the parent population ([24], p. 278). Achieving this depends first and foremost on the sampling method. Essentially, we can distinguish between two types of sampling: random and purposive selection. With purposive selection, certain criteria are defined according to which the selection is made. Using this sampling method, a distribution of characteristics which corresponds to the parent population is not ensured per se. Only random selection guarantees that every unit of the parent population has the same non-zero probability of being part of the sample ([19], p. 58; [24], p. 247). This is a prerequisite for the parent population and the sample to be similar when it comes to key characteristics. Moreover, random selection makes it possible to limit what is known as volunteer bias, which is a variant of selection bias. Volunteer bias describes the error that can occur when people are more likely to participate in surveys due to a particular interest in healthcare or because they had particularly good or particularly bad experiences with the healthcare system ([12], p. 425; [21], p. 237). It is plausible to assume that a general call for study participation, in contrast to directly approaching people in a random sample, is more likely to result in volunteer bias.

The sampling generates a population every unit of which – theoretically – has the opportunity to take part in a survey. It is not yet equivalent to the actual sample, meaning the population of all actual participants. The latter is created by the willingness to participate and the practical possibility of participating, which can, for example, also be limited by language barriers. The type of selection bias which can occur at this point is non-response bias, which, however, cannot be influenced by the sampling method per se. It depends far more on methodological decisions about the language and scope of the survey instrument as well as the mode of questioning.

In order to be able to make a random selection, all elements of the parent population must be known and nameable ([19], p. 59). This requirement is not fulfilled for all research questions in health science studies. In the following, a study from midwifery sciences is described, which serves as an example of a study where a random sample was used.

Cross-sectional survey within the ROSE research project

In the last decade, discussions among the (expert) public on regional shortages of midwifery services have been on the rise. At the same time, the premise enshrined in the German healthcare system is to provide care – including midwifery care – that is needs based, offers equitable access and is close to the citizens’ place of residence [25]. This situation led to the implementation of research projects in which mothers were surveyed on their needs and the midwifery care they had received [3], [4], [17], [22]. The results of these studies revealed, at the level of the respective German federal states, regional, service-specific and user group-specific shortages of midwifery care. So far, such a survey of mothers has not yet been conducted in the federal state of Lower Saxony. In its health reports on midwifery care from 2019 and 2021, the State Health Authority in Lower Saxony highlights a lack of consistent and systematically collected data, which prevents a reliable assessment of the situation regarding midwifery care provision [7], [8]. From research on medical service planning we know that the provision of healthcare has to be analysed and planned on a small scale in order to effectively assess and impact people’s situation in their specific local environment [23]. This is where the survey of mothers conducted as part of the ROSE project comes in. The aim of the survey was to reveal differences in access to healthcare throughout pregnancy and childbirth based on place of residence within the Osnabrück-Emsland region. The aim was also to identify gaps in meeting service users’ needs. The following questions were addressed: Were there, within the ROSE region, any geographical differences in access to maternal healthcare, in utilisation behaviour or in the wishes and preferences of the users of midwifery services? Were any needs not met and if so, what were they?

A postal cross-sectional survey was conducted targeting women who had had a live birth between 01/07/2018 and 30/06/2019 and were resident in one of the predefined sub-areas of the Osnabrück-Emsland region. The three study areas were defined based on the findings of a previously conducted study depicting the factors influencing maternal healthcare across the whole Osnabrück-Emsland region. Another inclusion criteria was that the mother was of legal age when she participated in the survey. The survey was open from December 2019 to April 2020.

Recently, surveys of mothers on the provision of midwifery care using similar questions have been conducted in various federal states [3], [4], [17], [22]. The ROSE survey differs from the surveys conducted so far in that 1) it considers healthcare throughout the entire process of pregnancy and childbirth and not exclusively in the sub-area of midwifery care and 2) it focuses on geographical differences in access to care in a small region.


The aim of this paper is to provide researchers from midwifery science with useful information with regard to decisions about possible field access for cross-sectional surveys as well as to discuss the importance of random samples when it comes to the generalisability of study results. Overall, the article aims to stimulate a methodological discussion in healthcare professions that are currently undergoing academisation in Germany.

To achieve this, the following questions are addressed: How can field access for a survey be achieved via municipal residents’ registration offices and what potential does this approach have in terms of achieving representativeness?

Methods and material

This article is based on research progression data for the “ROSE – the learning healthcare system in the Osnabrück Emsland region” project. The methodological approach to field access will be described and discussed in the context of the generalisability of quantitative study findings.

Field access in the ROSE cross-sectional survey: methodological decisions and practical approach

The decision-making process when it came to suitable field access for conducting the survey of mothers was initiated by defining the interested parent population. Since there was an interest in the variation between the different areas in the region when it comes to provision of care, three parent populations were defined which corresponded to the respective study areas. From a research theory perspective, the optimal sample type generated by this is a random sample stratified by area [24]. The choice of stratification was made based on the fact that, by defining the study areas, the total sample had already been divided into sub-groups in a theoretically meaningful way. In order to be able to draw a random sample, it has to be possible to depict each unit of the parent population, in other words, each unit has to be known. The names and contact details of all women in the target population therefore had to be available. This objective can only be achieved through access to the residents’ registration data.

The legal requirements for accessing a large quantity of address information from municipal population registers have been set out in the Federal Registration Act (Bundesmeldegesetz) since 2015. Section 46 of the law describes the conditions for the release of what is referred to as “group information”, as well as the required data input and possible data output. In principle, it is only possible to release group information “if doing so is in the public interest” [9]. This requirement is fulfilled in the case of a scientific study with no economic interests. For “information from the population register on a large number of persons not referred to by name” [9] the inputs that can be used are date of birth, current address as well as date moved in and date moved out. In our study, this corresponds to a date of birth during the period from 01/07/2018 to 30/06/2019 and a place of residence within the relevant municipality at time of birth. The output for these data pertaining to the newborn is “legal representative, including surname, given names and address” [9].

The literature review conducted to glean practical tips on approaches to retrieving address information via residents’ registration offices excluded foreign publications, as specific features of the German registration system are relevant here. The literature review focused, in particular, on finding tips on how to overcome obstacles to retrieving address data, as – according to the German Data Forum – despite the well-defined legal situation, not every residents’ registration office is prepared to provide group information for research purposes, which can make retrieving complete address data difficult and jeopardise sampling concepts [20]. Before reaching a decision about methodology, the potential risk to the sampling concept should be made calculable. However, neither in the publications on midwifery care, nor in other publications on health sciences based on random samples using residents’ registration data provided any useful practical research tips in this regard [6], [22]. The only detailed field report was published by Albers in 1997 [1]. Besides the legal requirements, which can no longer offer any guidance due to a change in the residents’ registration laws in 2015, Albers also described the financial and time-related requirements for this form of field access. The latter were mainly determined by the leeway of the individual municipal residents’ registration offices. For instance, the fees charged for providing the address data varied, in some cases quite significantly, and could be unexpectedly high. In terms of the amount of time it took to acquire the information, three to four months should be anticipated, although this does not all involve just waiting to receive the address data but is more of an intensive period of fostering and maintaining contact with each individual residents’ registration office [1]. A personal consultation with a researcher who used residents’ registration data in 2013 confirmed that the advice provided by Albers continued to apply.

Recommended measures to increase the willingness on the side of the residents’ registration offices to release the address data were implemented at the same time as the preparatory phase that now followed. Anticipating and dispelling any potential data protection concerns and convincing the offices that the planned study was in the public interest played an important role here. To this end, the initial cover letter to the residents’ registration offices should include an ethics’ committee approval for the research project, a letter of endorsement from a political figure and sufficient information about the study. Addressing the person responsible for the request by name is also likely to increase the willingness to supply the required information (see Infobox).

Once preparations had been completed, the 32 relevant residents’ registration offices in the region were contacted in writing: 14 in the district of Emsland, 17 in the district of Osnabrück as well as the registration office in the city of Osnabrück. Responses from the offices came back in bursts over a period of ten weeks. Here three different response patterns could be observed:

Prompt positive response providing information; the first response was received 16 days after sending the request for information.
Delayed provision of information with an essentially positive or neutral attitude. The delays resulted from, for example, changes in staff responsibilities, staff absences and excessive workload as well as changes in data processing. In these cases, the registration offices were contacted by telephone at regular predefined intervals, and as needed. The content of the conversations was documented.
Initial refusal to provide information; two residents’ registration offices initially refused to provide group information. In one of the cases, a written notification was received explaining that it was not possible to grant the request for data privacy reasons. In the second case, it took several weeks to reach the member of staff responsible by phone and during the phone conversation they refused the request. The member of staff had consulted the most senior authority – the Lower Saxony Ministry of Internal Affairs – to ask for a decision on the issue. At that point, a response had not yet been received. Both offices could ultimately be persuaded through letters and telephone calls, and in the latter case, through permission from the Ministry of Internal Affairs, to provide the information requested.

Ten weeks after the request for information had been submitted, address data had been received from all 32 residents’ registration offices approached – in different digital formats and in some cases in paper form. During this period, a total of 63 telephone conversations were conducted with the clerks responsible in the residents’ registration offices. The fees charged for the group information varied considerably from one local authority to the next, but the overall total cost was not as high as expected.

After cleaning and synchronising the address data, there were three electronic lists with no duplicates, sorted by area, from which to draw the stratified random sample. The sample was drawn separately for each study area – each stratum – in accordance with the following system [15]: First, each entry on the list was assigned a random number. In the next step, the entries were sorted in ascending order, based on these numbers. In the last step, the first entries – n is the number of units to be drawn according to the sampling plan – were selected. For Area 1 (A1), the stratified random sample comprised 1,170 women, for A2 and A3, they comprised 957 women each. When the size of the random sample to be drawn was determined, it was based on the assumption of a response rate of 30 percent, which was similar to the above-mentioned comparable surveys on midwifery care (see Figure 2 [Fig. 2]).

Sample achieved in the ROSE cross-sectional survey

All women in the stratified random sample received a questionnaire with 86 question items in German, an information letter asking them to participate in the study, a data privacy statement in duplicate with an envelope along with a prepaid return envelope. A conscious decision was made regarding the mode and the language for the survey, taking all potential consequences into account. In order to enable women with limited or non-existent German-language skills to participate in the survey nevertheless, midwives as well as all day-care facilities in the region were contacted personally in advance of the survey and informed that a random sample of women were being written to with the request to participate. The people contacted were also provided with information material. They were informed that should anyone wish to participate in the survey, they could contact the study team in order to find a solution for the individual concerned, including to any language difficulties.


These measures implemented in the ROSE cross-sectional survey were not, however, as successful as expected. Of the total of 629 valid questionnaires, which were submitted by the end of the five-month return time, only 8.5 percent were women who were not born in Germany. This does not correspond to the share in the parent population. Apart from the underrepresentation of women with a migrant background, the participation of women with a low level of education and low household income was also limited (see Table 1 [Tab. 1]). Thus the inference population in the ROSE survey of mothers – in other words, the population based on which conclusions could be drawn – does not correspond to those of the parent population defined for the study with regards to all variables. Despite this under-coverage, the survey can still provide valuable insights, provided that reference is made to the inference population: German-speaking women with a medium to high social status. The response rate was 20 percent for Area 1, 18 percent for Area 2 and 22 percent for Area 3.


When planning the survey of mothers for the ROSE project, the motive for using a random sample was the desire to maintain a representative sample. By drawing the random sample stratified by area every mother in the respective study area had the same chance of being included in the sample. Creating the sample in this way meant the criterion of representativeness was met, which is a condition for an area-based generalisation of the results.

Apart from the sampling concept there are also other factors in the planning of a study which influence the achievement of representativeness. In the case of ROSE, these factors were the choice of scope of the survey instrument as well as the mode, which was a postal survey, and the related decision to opt for a monolingual questionnaire. The latter decision in particular was the basis of a systematic nonresponse bias [16], as non-German speaking women were excluded from participating from the very start or their participation was made more difficult.

The inference population in the ROSE survey of mothers does not correspond to the parent population with regards to all variables. This situation is however not unusual for health science studies, and this applies in particular to general calls to participate in a study [3], [4]. The fact that this finding is also evident in a study where field access is via personal written invitation suggests a general methodological challenge. The samples achieved in a federal state-specific study on midwifery care, the field access for which was also via residents’ registration offices, showed a similar socio-demographic constellation as in ROSE. In the midwifery care study conducted in Bavaria, depending on the settlement structure, between 49 percent and 79 percent of respondents stated that Abitur (school qualification required for university entrance) or Fachabitur (Abitur obtained at a vocational training school) was their highest school-leaving qualification. The share of women with a migrant background (8.6–25.5 percent) was rated as non-representative of the parent population in the Bavarian study [22]. Participants in the Bavarian study were not asked about their net household income.

Although, by drawing a random sample, the best possible methodological conditions had been created for the achievement of a high level of representativeness, other methodological decisions in the ROSE study meant that it did not prove possible to prevent a nonresponse bias, which thus reduced the representativeness of the sample achieved. The achievement of a random sample by using address data from municipal residents’ registration offices can thus be considered a necessary but not sufficient measure for the achievement of representativeness.


A random sample of mothers for surveys in midwifery science studies can be achieved through data acquisition from municipal residents’ registration offices. Since the costs and time required to acquire address data is impossible to accurately predict, a certain degree of flexibility within the research project at hand is beneficial. Further, it should be decided in advance what cost-benefit ratio would be deemed acceptable. During the process of reaching consensus on the study design, the importance of achieving a random sample in relation to the aspired degree of representativeness should be taken into consideration: random sampling is indispensable for this, but not yet sufficient. Additional measures for the achievement of maximum possible representativeness should also be given due consideration. The insights acquired must also be placed in a methodologically legitimate context.

What appears questionable with regards to the insights the studies on the provision of care seek to obtain – that being an assessment of the care received by all women, in other words by all user groups – is the extent to which a cross-sectional survey is able to deliver the answers sought. If no findings are produced on specific user groups, the degree of fair distribution of and equal access to midwifery care cannot be determined. Looking at the methods used and, more importantly, at the subject of the study, an innovative new methodology and approach seems necessary to find solutions to the pressing problems seen in the practice of providing midwifery care. One potential next step here might be to focus on service users who have so far been difficult to access, for instance through the use of participatory and combined methods.


Recommended preparation for submitting a request for group information according to Section 46 of the Federal Registration Act (Bundesmeldegesetz, BMG) [9]:

  • Formulate a persuasive and personal letter to the clerk responsible in the residents' registration offices being approached
  • Lists of members of staff responsible in the relevant residents' registration offices with names and full contact details
  • Create information material on the planned study tailored to the target audience
  • Request a letter of support from public (political) figure, preferably from a higher level than local government
  • Submit an application to an ethics committee (prerequisite: completion of the survey instrument, together with accompanying documents)


Competing interests

The authors declare that they have no competing interests.


Albers I. Einwohnermelderegister-Stichproben in der Praxis. Ein Erfahrungsbericht. In: Gabler S, Hoffmeyer-Zlotnik J, editors. Stichproben in der Umfragepraxis. Opladen: Westdeutscher Verlag; 1997. p.117-26.
Atteslander P. Methoden der empirischen Sozialforschung. 13th ed. Berlin: Erich Schmidt Verlag; 2010.
Bauer NH, Blum K, Loeffert S, Luksch K. Gutachten zur Situation der Hebammenhilfe in Hessen. 2019 [Access 01 Oct 2022]. Available from: External link
Bauer NH, Villmar A, Peters M, Schaefers R. HebAB.NRW - Forschungsprojekt „Geburtshilfliche Versorgung durch Hebammen in Nordrhein-Westfalen“. Abschlussbericht der Teilprojekte Muetterbefragung und Hebammenbefragung. Hochschule fuer Gesundheit Bochum; 2020 [Access 01 Oct 2022]. Available from: External link
Baur N, Blasius J. Handbuch Methoden der empirischen Sozialforschung. Wiesbaden: Springer; 2014.
Berens E, Riedel J, Reder M, Razum O, Kolip P, Spallek J. Postalische Befragung von Frauen mit tuerkischem Migrationshintergrund – Identifizierung, Stichprobenbereinigung und Response im Rahmen der InEMa-Studie. Gesundheitswesen. 2017;79:1000-3. DOI: 10.1055/s-0035-1564076 External link
Bruns-Philipps E, Heidrich S, Reissner K, Schicktanz C, Zuehlke C. Gesundheitsbericht Hebammenversorgung in Niedersachsen, Hannover. Niedersaechsisches Landesgesundheitsamt; 2019 [Access 01 Oct 2022]. Available from: External link
Bruns-Philipps E, Heidrich S, Reissner K, Zuehlke C. Gesundheitsbericht Hebammen in Niedersachsen, Hannover. Niedersaechsisches Landesgesundheitsamt; 2021 [Access 01 Oct 2022]. Available from: External link
Bundesmeldegesetz (BMG) § 46 Gruppenauskunft. Zuletzt geaendert durch Art. 7 G v. 15.1.2021. Bundesgesetzblatt. 2013;I(22):1084-103.
Doering N. Qualitaetskriterien fuer quantitative empirische Studien. Enzyklopaedie Erziehungswissenschaft Online. 2015. DOI: 10.3262/EEO07150345 External link
Doering N, Bortz J. Forschungsmethoden und Evaluation in den Sozial- und Humanwissenschaften. 5th ed. Berlin, Heidelberg: Springer; 2015.
Dreier M, Kramer S, Stark, K. Epidemiologische Methoden zur Gewinnung verlaesslicher Daten. In: Schwartz FW, Walter U, Siegrist J, Kolip P, Leidl R, Dierks ML, Busse R, Schneider N, editors. Public Health. Gesundheit und Gesundheitswesen. Muenchen: Elsevier; 2012. p.409-49. DOI: 10.1016/B978-3-437-22261-0.00017-4 External link
Faul F, Erdfelder E, Lang AG, Buchner A. G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods. 2007;39(2):175-91. DOI: 10.3758/bf03193146 External link
Flick U. Sozialforschung, Methoden und Anwendungen - Ein Ueberblick fuer die BA-Studiengaenge. Hamburg: Rowohlt Taschenbuch Verlag; 2009.
Haeder S. Stichproben in der Praxis. In: GESIS – Leibniz-Institut fuer Sozialwissenschaften, editor. GESIS Survey Guidelines. Mannheim: GESIS; 2015. DOI: 10.15465/sdm-sg_014 External link
Koch A, Blohm M. Nonresponse Bias. In: GESIS – Leibniz-Institut fuer Sozialwissenschaften, editor. GESIS Survey Guidelines. Mannheim: GESIS; 2015. DOI: 10.15465/gesis-sg_004 External link
Loos S. Hebammenversorgung in Thueringen. Gutachten zur Versorgungs- und Bedarfssituation mit Hebammenleistungen sowie ueber die Einkommens- und Arbeitssituation von Hebammen in Thueringen. Berlin: IGES Institut; 2015 [Access 01 Oct 2022]. Available from: External link
Micheel HG. Quantitative empirische Sozialforschung. Muenchen, Basel: Ernst Reinhard Verlag; 2010.
Poetschke M. Datengewinnung und Datenaufbereitung. In: Wolf C, Best H, editors. Handbuch der sozialwissenschaftlichen Datenanalyse. Wiesbaden: VS Verlag fuer Sozialwissenschaften; 2010. p.41-64.
Rat fuer Sozial- und Wirtschaftsdaten (RatSWD). Die sozial-, verhaltens- und wirtschaftswissenschaftliche Survey-Landschaft in Deutschland: Empfehlungen des RatSWD. German Data Forum; 2017. DOI: 10.17620/02671.5 External link
Razum O, Breckenkamp J, Brzoska P. Epidemiologie fuer Dummies. 2nd ed. Weinheim: Wiley-VCH; 2011.
Sander M, Albrecht M, Loos S, Stengel V. Studie zur Hebammenversorgung im Freistaat Bayern. Berlin: IGES Institut; 2018 [Access 01 Oct 2022]. Available from: External link
Schang L, Weinhold I, Wende D, Sundmacher L. Monitoring und Bewertung des regionalen Zugangs zur ambulanten aerztlichen Versorgung in Deutschland. BARMER Gesundheitswesen aktuell. 2019:230-71. DOI: 10.30433/GWA2019-230 External link
Schnell R, Hill B, Esser E. Methoden der empirischen Sozialforschung. 11th ed. Berlin, Boston: De Gruyter Oldenburg; 2018.
Sozialgesetzbuch (SGB) Fuenftes Buch (V) – Gesetzliche Krankenversicherung. § 70 Qualitaet, Humanitaet und Wirtschaftlichkeit. Zuletzt geaendert durch Art. 1 G vom 07.11.2022.