The German Doctoral Candidates and Doctorate Holders Study ProFile

2.1 Sampling Design

Registration procedures of doctoral candidates differ from university to university, since registration of doctoral candidates is not mandatory in Germany. ^[2] The German Federal Statistical office published two studies in 2012 and 2016 that estimate the number of doctoral students (Wolters/Schmiedel 2012; Hähnel/Schmiedel 2016). However, these estimations were not intended to function as a selection frame for sampling and the underlying total population of doctoral candidates remains unknown. Thus, drawing a simple random sample is not possible (Fräßdorf/Fräßdorf 2016). Therefore, several universities, graduate schools and scholarship providers were contacted before the ProFile survey started in 2009 in order to get access to registers of doctoral candidates. The selection of participating institutions was primarily driven by the intention to reflect heterogeneity in contextual factors, respectively settings, in which a doctorate might be pursued and obtained and to yield sufficient sizes of subsamples for comparative analysis. The sample of universities was therefore intended to vary in terms of size, composition of disciplines and geographical location (Hauss et al. 2012: 49). In addition, the selection was also influenced by the fact that the respective institution needed to be willing to participate in the project and capable of assessing and transmitting the relevant data of their doctoral candidates (Hauss et al. 2012: 51). The participating institutions annually submit contact data of their doctoral candidates (encompassing email address, name, year of birth, gender and discipline). Even though the participating institutions were not sampled randomly, the selection of participating institutions in ProFile basically follows the logic of a cluster sampling where ProFile attempts full coverage at the level of the participating institutions – the primary sampling units – listed in Table 1.

2.2 Survey design

The study follows a panel design in the sense that it ideally surveys a doctoral candidate at the beginning of his or her doctoral candidacy, at least one time during the doctoral candidacy, at the conferral of the doctorate and once after the conferral of the PhD as illustrated in Figure 1. Every year, a new cohort of doctoral candidates is added to the survey.

Figure 1:

ProFile longitudinal design.

The study foresees three surveys which are conducted by the DZHW: the Initial Survey, the Evaluation Survey and the Follow-Up Survey. Participants are invited through a personalized link to the online questionnaire via a bilingual email with English and German being the available survey languages.

The Initial Survey targets doctoral candidates who participate in ProFile for the first time. Even though the Initial Survey targets doctoral candidates at the beginning of their doctoral candidacy, due to the lacking necessity of prompt registration at universities, not all persons delivered are doctoral candidates. Instead, the data also contain persons who recently completed their doctorate or persons who stopped working on their doctorate (drop outs). About 5 % of the respondents in the Initial Survey are doctorate holders and less than 1 % report that they quit their PhD. The yearly Evaluation Survey focuses on persons who already participated in ProFile and stated that they are still doctoral candidates in the previous Evaluation or Initial Survey or those who completed their doctoral training in the meantime. The Follow-Up Survey targets doctorate holders who participated in ProFile at least once and completed their doctoral candidacy. ^[3]

2.3 Questionnaire Design

Since the status of the respondent is unknown prior to the invitation to the Initial and Evaluation Surveys, event triggered questionnaires have been implemented. Depending on the actual status of their doctoral project, which is assessed in the beginning of the Initial and Evaluation Survey, respondents are directed to varying questionnaires as depicted in Table 2. Thus, the status of the doctoral project serves as key variable for routing the participant to the correct questionnaire.

The shaded cells in Table 2 display the questionnaires a ProFile participant ideally answers: The Initial Questionnaire for Doctoral Candidates (PI) focuses on information on the educational background, motives for a candidature, the supervision situation, the quality of courses offered, resources and questions about how doctoral candidates finance themselves during their doctoral candidacy. If the respondent’s status of the doctoral project does not change during the next year he or she receives the Evaluation Questionnaire for Doctoral Candidates (PE) which aims at generating more information about the (possibly changing) conditions of their doctoral training. Conceptualized as a short questionnaire between the beginning and the conferral of the doctorate, it primarily considers changes in individual funding and supervision. If respondents indicate to have completed their degree, they receive the Evaluation Questionnaire for Doctorate Holders (PtE) which determines the successful conferral of the doctorate as well as job-related aims and perspectives. After the doctorate is finished, the Follow-Up Questionnaire (FQ) focuses on job placement in and outside of academia as well as career trajectories after the conferral of the doctorate. It also aims at gaining more insight into job satisfaction, possible job strains and adequate occupation within or outside of academia. Table 3 gives a brief overview of the single topics covered in the different questionnaires.

3.1 Number of Respondents

Altogether 15,649 doctoral candidates participated at least once in an Initial Survey from 2009 to 2015. Nearly half of them could be surveyed in an Evaluation Survey afterwards, from which 3,271 were questioned about their conferral of the doctorate. Up to now, 1,554 individuals were surveyed at all stages of their doctoral project (see Table 4). Since ProFile is still ongoing, this number of respondents will increase. As mentioned above, a few respondents are already doctorate holders in the Initial Survey and receive a specific questionnaire which surveys the beginning of the doctorate in retrospect and are questioned in the Follow-Up Survey, too.

4.2 Sample composition

The fact that besides universities also graduate schools and scholarship providers serve as primary sampling units doctoral candidates from these types of institutions are oversampled in the ProFile data. This was anticipated in order to allow in-depth analyses for these subgroups. Therefore, three samples need to be distinguished. Specific analyses on stipends can be run using the stipend sample which consists of doctoral candidates from graduate schools and scholarship providers exclusively. For analyses that are not solely focused on stipends the uni sample can be used. It consists of doctoral candidates at participating universities including scholarship holders from the participating graduate schools and scholarship providers, in case those doctoral candidates intend to graduate from one of the participating universities. The ones not intending to graduate from the participating universities are excluded from this sample. Both samples overlap in the parts of scholarship holders intending to graduate from participating universities. Taking into account the oversampling of stipends in the analyses, it is also possible to use the full sample. However, for descriptive statistics it is recommended to use the uni sample to avoid bias. Table 5 gives an overview of the number of respondents in the three samples for the Initial Surveys. The variation in the sample sizes results from changes in the composition of the participating institutions.

4.3 Data Quality

Data quality is ensured by means of data monitoring and error correction at different stages. At the stage of data delivery the whole contact data delivered by the participating institutions is checked and corrected for duplicates. ^[4] During field phase three reminders are sent and bounced email addresses are researched manually and invited again to increase response rates. ^[5] In addition, respondents were provided with results from the survey in order to show its value and increase panel commitment. The response rates vary between 20 % and 40 % for Initial Surveys, 30 % and 50 % for Evaluation Surveys and over 50 % for Follow-Up Surveys. ^[6] After data collection data editing includes anonymization, plausibility checking, coding of missing values and editing of central variables.

Furthermore, external validation was conducted on the level of the participating institutions and on the level of the whole sample. On the level of participating institutions the distributions of the data provided (gender, year of birth and discipline) and the data collected in the Initial Surveys are compared. On the level of the whole sample the uni sample of the Initial Survey 2015 can be compared with up-to-date official results of the Federal Statistical Office (Hähnel/Schmiedel 2016) which estimated the number of doctoral candidates at German universities by means of a two stage random sample where professors (as supervisors) and then doctoral candidates were asked to participate. The study concluded that there were about 196,200 doctoral students in Germany in winter semester 2014/2015. The comparison of the ProFile uni sample with the current results of the Federal Statistical Office is depicted in Table 6. ^[7]