Course Profiles and Participation in German Adult Education Centers During Times of Migration: A Longitudinal Study

Abstract

The increasing demand for integration courses in Germany (especially due to the crisis-induced migration of refugees) has raised the question in research and practice of how this change affects the supply and participation of continuing education providers. To address this issue, the present study explores the impact of increasing integration course counts and participation on regular course counts and participation in German state-funded public adult education centers (Volkshochschulen; VHS), which are the main providers of integration courses. For this, we apply longitudinal fixed effects analyses to the VHS statistics. Controlling for VHS-specific trends, we find that on average, an increase in integration course counts slightly decreases regular course counts. Regarding participation, we observe the opposite: Increasing participation in integration courses leads to a small increase in participation in regular courses.

Keywords

adult education migration integration courses segregation

Since 2005, every individual who enters Germany as a nonnative and non-EU citizen, and who intends to stay in Germany without fundamental knowledge of the German language, is obliged to participate in an integration course. The immigration office determines whether their participation in an integration course should be mandatory (Federal Office for Migration and Refugees, 2018). Integration courses usually comprise 700 hours, with 600 hours of language instruction and 100 hours of cultural instruction. Graduates must achieve a B1 language level. Although most European countries have requirements regarding the language skills of individuals applying for a permanent residency permit, the requirements in Germany are comparatively high, with only Denmark and United Kingdom with comparably high requirements (Rocca et al., 2020). Consequently, the demand for integration courses in Germany tracks continuing migration movements and skyrocketed with increased crisis-induced refugee migration in 2015 (Federal Office for Migration and Refugees, 2019a). While several scholars evaluated whether integration courses are a successful means of facilitating migrants’ integration into society (Hochman & Davidov, 2014; Lochner, 2015), there is still little knowledge on developments that take place within integration course providers, that is, within the German adult education centers (“Volkshochschulen”; VHS). Alongside the general educational mandate of VHS, which is the provision of education for all citizens, covering a broad range of topics and subjects, VHS conduct the largest share of integration courses.

So far, only Käpplinger (2018), Käpplinger and Reuter (2019), and Käpplinger (2020) have investigated how the program structure and subjects in German adult education centers have changed in the context of migration. We build on this research and extend it by two important aspects: First, we complement the question regarding the influence of migration on the program structure with an analysis of the influence of migration on participation. Second, we investigate both questions with causal methods. Our objective is to measure the causal influence of migration on both the supply side and the demand side of continuing education. For this, participation counts and course counts are appropriate performance indicators (Widany et al., 2020).

We hypothesize that open course counts are unaffected by increases in integration course counts. Our argument is based on the idea that course counts depend on available resources and the demand for courses. When the demand for open courses remains stable and when there is no resource stress due to the supply of additional integration courses, the open course counts should also remain stable. We do not expect resource stress to arise, since VHS often operate in flexible, cooperative networks and are financially supported by municipalities and federal states.

Regarding participation counts, we analyze whether mechanisms of ethnic and socioeconomic segregation comparable to segregation in schools or voluntary organizations (Makles & Schneider, 2015; Owens et al., 2016; Schneider et al., 2012) come into effect. We assume that increasing integration course participation decreases open course participation. Moreover, we argue that the extent to which this segregation process sets in depends on the proportion of migrants among potential open course participants and on the attitudes of potential open course participants toward migrants.

To test our assumptions, we use the VHS statistics with N = 864 VHS and N = 8,640 observations over 10 years combined with public statistics of the German municipalities. For the analysis, we estimate fixed effects (FE) and fixed effects individual slopes (FEIS) models.

Dynamics in Course Counts and Participation

Following the integrated model of participation in adult education (Boeren et al., 2010), course counts and participation in adult education can be seen as the result of an interplay of supply and demand. The match between supply and demand follows from mechanisms that take place both within and between the two levels: Individuals decide for or against participation in adult education depending on their socioeconomic situation, the expected costs and benefits and psychological attitudes toward learning opportunities (Boeren et al., 2010). The adult education center similarly decides to provide educational offers depending on its resources and educational mandate (Meisel, 2019). Both the participant choices and the organizational choices are embedded within broader social contexts (Boeren et al., 2010). This approach guides the structure of our argument; we distinguish mechanisms on the supply side (the VHS) and mechanisms on the demand side (open course participants and integration course participants), embedded in the broader social context of municipalities.

Open Course Supply as a Function of Integration Course Supply

Adult education centers that offer increasing numbers of integration courses inevitably face an increased need for resources, such as rooms, teachers, and administrative staff. In the event that generating new resources fails, but integration courses still have to take place, a lack of resources arises. One option to compensate for the lack of resources would be to cut resources that were intended for open course offerings. This would result in decreasing open course counts. Acknowledging that the VHS receives funding for integration courses and open courses separately from distinct funders, the shortage of funds may in general be unlikely. Open courses take place on the base of participation fees and funds from the federal government, the federal states and the municipalities (Reichart et al., 2018). The financial means for the supply of integration courses are put together also partly from participation fees but mostly from funding by the Federal Office for Migration and Refugees. Additionally, cooperation with other organizations may compensate for shortages of rooms and staff (Martin & Muders 2018). Based on this, we assume that there is no structural displacement of open courses by integration courses.

Hypothesis 1: The increase in integration course counts does not lead to a decrease in open course counts.

Demand for Open Courses as a Function of Participation in Integration Courses

Prior research on the determinants of participation in adult education provides evidence on different sociodemographic characteristics that are associated with adult education participation, for example, higher levels of (formal) education, no migration background, being middle-aged, female, and being employed (Bélanger, 2011; Bilger et al., 2017; Reichart et al., 2018).

However, with increasing numbers of integration courses, the participant structures in a VHS change. The regions of origin of migrants and asylum seekers in Germany are very heterogeneous and have changed considerably over the last decade. Before 2015, the majority of migrants came from former Yugoslavia, Iraq, Afghanistan, and Turkey. Since the so called “2015 Refugee Crisis,” the majority of migrants has come from Syria, Iraq, and Afghanistan (Federal Office for Migration and Refugees, 2019b). Despite these diverse origins of migrants, as participants in integration courses in Germany, they all have in common that they have no German citizenship and no or only poor German language skills. The grouping of these heterogeneous individuals in exclusive integration courses, which already implies that the participants are not (yet) part of the receiving society, may strengthen the perception of migrants as belonging to an out-group.

Due to these inevitable changes in the participant structures, one may observe a scenario of segregation similar to the scenarios of segregation in schools and voluntary organizations (Makles & Schneider, 2015; Owens et al., 2016; Schneider et al., 2012; Wiertz, 2016). A basic mechanism for segregation is in-group favoritism. In sociology, this mechanism is discussed as the basis for ethnocentrism and is applied to social behavior such as voting, consumer choice, and discriminatory behavior toward migrants (Hammond & Axelrod, 2006). In particular, the social identity theory (Tajfel, 1982) can be applied when theoretically approaching segregation processes in a VHS: Individuals build up a self-concept based on their social identity, which itself is based on individuals’ perceived membership in social groups. A general striving for self-esteem calls for the construction of a positive social identity. Self-esteem highly correlates with in-group favoritism, which means that those individuals with high levels of self-esteem particularly prefer their in-group (Aberson et al., 2000). Following this logic, individuals who join an educational organization may choose an organization that is also favored by their in-group members. Consequently, they avoid organizations that are perceived as dominated by out-group members or appear to be increasingly populated by out-group members. The same mechanism may apply to VHS participants. However, whether such a mechanism can be effective also depends on the proportion of persons with a migration background among the potential participants in open courses. The population in Germany as a country with a long history of migration is also very diverse. Hence, in view of many individuals with a migration background in the German receiving society, there are high regional proportions of migrants. When this factor is controlled, the following hypothesis emerges:

Hypothesis 2: The more integration course participation rises, the lower open course participation becomes.

The Influence of Social Context on Potential Segregation Processes

The mechanisms behind segregation heavily build on the assumption that processes of social categorization take place. However, this highly depends on the perception of the integration course participants as belonging to an out-group as well as the attitudes and prejudices toward integration course participants. Both may substantially vary depending on the social context in the municipalities. A fundamental factor in the perception of a group is its size (Laurence & Bentley, 2018). Two contrasting approaches make claims as to how the size of an out-group influences the perceptions toward this out-group, and thereby how the out-group size moderates mechanisms of segregation in VHS. Hence, the following considerations are conditional for Hypothesis 2 to hold true.

According to the group threat theory (Blalock, 1967), prejudices and negative attitudes toward minorities or subordinated groups increase when these groups are perceived as a danger or as a threat to one's own group. The larger the group is, the more it is perceived as a threat. Thus, negative attitudes and prejudices toward the out-group should increase with its increasing size (Campbell, 2006; Hjerm, 2007, 2009; Kanas et al., 2015; Quillian, 1995; Schaller & Abeysinghe, 2006).

However, a high proportion of migrants in a municipality can also reduce segregation in a VHS. The higher the proportion of migrants in a municipality is, the higher is assumedly the proportion of migrants among potential participants in open courses. This reduces the proportion of potential participants for whom migrants participating in the integration courses are perceived as belonging to an out-group. This plays a decisive role, especially in large municipalities. While in small German municipalities with up to 2,000 inhabitants, the average proportion of people with a migration background is 9.7%, it is 35.8% in municipalities with more than 500,000 inhabitants (Destatis, 2020). According to intergroup contact theory (Allport, 1954), prejudices and negative attitudes toward migrants may also decrease because of a higher proportion of migrants in the municipality. The key assumption here states that with increasing size of the out-group, intergroup contact becomes more likely. As a consequence of this intergroup contact, prejudices, and perceived threats may be reduced when similarities between the groups become visible (Allport, 1954; Al Ramiah & Hewstone, 2013; Paluck et al., 2019). A meta-study by Pettigrew and Tropp (2006) shows that even when accounting for the self-selection problem, a clear causal effect in the direction predicted by the intergroup contact theory can be observed. Based on these arguments, we test the following hypothesis

Hypothesis 3: The higher the proportion of migrants in the population of the municipality in which the VHS is located, the lower the segregation effect in that VHS.

Data Description

To test our hypothesis, we use longitudinal data from the German VHS statistics (Reichart et al., 2018). Initiated by the German Institute for Adult Education—Leibniz Centre for Lifelong Learning in 1962, the annual online survey collects organizational data on VHS, capturing information on staff, finances and expenditures, course offerings, participation rates, cooperation with other organizations, and other activities such as exhibitions or symposia. Besides this rich set of variables, only eight centers did not participate in the survey. This makes the data an almost complete sample of every German VHS (N = 864) and therewith, we work with an exceptional database.

In our study, we use data from the survey years 2007 to 2017 for two reasons: First, to test Hypothesis 3 on the moderating effect of the social context, we ought to match regional context data to the VHS statistics. This is only possible from 2007 onward. Second, after 1991 and until the early years of the new millennium, there were several structural changes in the VHS landscape in Germany: Many VHS closed, merged with other VHS or split into more than one division. These changes may inevitably cause changes in open course counts and participation, which may distort the results of our analysis.

In 2007, 973 VHS participated in the survey. In 2017, 883 VHS participated. Throughout the whole observation period, 864 VHS took part in the survey, which then shapes the base of our analytical sample. While the observation period covers 11 years, the inclusion of lagged variables in our models shrinks our observation period to 10 years. This results in 8,640 observations.

In addition to the VHS statistics data, we use data from the German federal statistical office, which supplies us with regional context data (details in the next section). This data is available at the municipal level for the period 2007 to 2017 and stems from the official registers of the municipalities. Since the data from a single municipality (not named here) was unfortunately not consistently available, we have disregarded data from the corresponding VHS from our course participation dynamics models. Thus, the analytical sample for the participation models sums up to 8,630 observations.

Variables

Our hypotheses target the potential effects of course counts and participation in integration courses on course counts and participation in open courses at VHS. We separated model course counts and participation. The VHS statistics include course counts and participation rates per course within each calendar year, recorded for 86 subject areas. The subject area titled “German as a foreign language” includes integration courses. Hence, course counts and participation in this subject area form our main independent variables. All courses of the remaining 85 subject areas are considered to be open courses, since in contrast to integration courses, the majority of these courses are openly accessible. The course offerings and participation rates in these open courses form our dependent variables.

Our first hypothesis builds on the assumption that VHS are able to generate additional resources to conduct integration courses. With the available data, we are able to distinguish between human, spatial, and financial resources. We cover human resources by the number of employees, which include both permanent employees and freelancers. To approximate the extent of spatial resources, we include the number of courses that took place in cooperation with other institutions. Financial resources are operationalized with the amount of money gathered from participation fees and funds from the state, the federal states, the European Union, municipalities and job centers.

To estimate potential influences of municipality-specific characteristics on changes in VHS and VHS participants, we generate variables covering the population size, the proportion of natives as well as the proportion of migrants (citizens without German citizenship) living in the municipality and the average net household income within municipalities. We include these variables in the estimation (1) to control for heterogeneity regarding the migration background among potential participants in open courses in Hypothesis 2, (2) to test Hypothesis 3, and (3) to control for the most important confounding changes on the municipal level.

Table 1 shows the descriptive statistics for all variables and for our two analytical samples (course models and participation models). The two samples do not substantially differ. In the following description, we use the full sample (course count models with 8,640 observations from 864 VHS) as a reference. On average, a VHS conducts 581 open courses and 50 integration courses per year. These courses serve an average of 6,266 open course participants and 630 integration course participants. However, the high standard deviations and the large ranges indicate that there are substantial differences between individual VHS, but also within any given VHS over time (as the within standard deviation suggests). The average development over time in course and participation counts may also be drawn from Figure 1. While open course counts increased until 2014 and only then sharply decreased, open course participation rates decreased throughout the whole observation period. The opposite holds true for integration courses and participation rates. Here, as expected, we observe a steady rise, which becomes steeper toward the end of the observation period.

Table 1.

Descriptive Statistics of Dependent, Independent, and Control Variables.

	Sample for course count models, N = 8,640			Sample for participation count models, N = 8,630			r
	M	SD	w-SD	M	SD	w-SD	r
Regular courses	581	684	82	581	684	76	1 to 10,796
Participation in regular courses	6,266	8,095	1,510	6,264	7,997	1,326	13 to 147,161
Integration courses	50	94	34	50	97	33	0 to 1,541
Participation in integration courses	630	1,420	483	630	1,421	482	0 to 26,726
Federal funds (1,000€)	64	199	110				0 to 3,741
Job center (1,000€)	61	337	141				0 to 6,311
EU funds (1,000€)	24	107	56				0 to 2,389
Other revenues (1,000€)	85	297	131				0 to 5,939
Communities (1,000€)	250	826	119				0 to 20,446
Municipalities (1,000€)	57	167	81				0 to 6,925
Federal state (1,000€)	156	328	58				0 to 6,412
Participation fees	461	768	157				0 to 15,002
Human resources	221	233	41				1 to 3,398
Cooperation	29	64	30				0 to 935
Germans in municipality				247,921	353,521	3,682	31,257 to 2,988,274
Net household income in municipality (1,000€)				5,607,254	7,638,127	630,400	612,872 to 7 × 10⁷
Population in municipality				274,130	411,796	9,349	33,944 to 3,574,830
Proportion of migrants				7.64	3.96	0	1.10 to 29.18
Proportion of migrants 33th percentile				0.33	0.47	0	0/1
Proportion of migrants 33th to 66th percentile				0.33	0.47	0	0/1
Proportion of migrants 33th to 100th percentile				0.33	0.47	0	0/1

Note. w-SD = within standard deviation; r = maximum range.

Source. Own calculations using VHS (Volkshochschulen) statistics, 2007 to 2017.

Figure 1.

VHS (Volkshochschulen) integration course and regular course counts and participation in 2007 to 2017.

Analytical Strategy

Analyzing the effects of integration course counts and participation on open course counts and participation calls for methods that account for the high heterogeneity of VHS. This becomes a challenge when unobserved baseline heterogeneity and unobserved growth heterogeneity are related to the number of integration course offerings or related to the participation rates in integration courses. To control for time-invariant unobserved heterogeneity, we employ FE models. However, FE models cannot account for growth heterogeneity. In order to control for this unobserved heterogeneity, we extend the FE approach to a FEIS model (Brüderl & Ludwig, 2014). We use both model specifications, FE and FEIS models, to test our Hypotheses 1 and 2. The comparison between estimates from both approaches shows the influence of unobserved time-specific confounders.

For the number of course counts (Hypothesis 1), we specify the FE model as follows:

y_{i t} = a_{i} + β_{0} + β_{1} x_{1 i t} + \sum_{j = 2}^{j = 11} β_{j} x_{j i t - 1} + ε_{i t}

(1)

Here, $y_{i t}$ is the number of open courses that took place at a VHS $i$ at time $t$ . $a_{i}$ is a specific constant for each individual VHS. $x_{1 i t}$ denotes the number of integration courses, $x_{2 i t - 1}$ to $x_{11 i t - 1}$ are the values of the resource variables (human, spatial, financial resources), which we set using the VHS resources from the previous year, since the strategic course planning is presumably influenced by the resources of the previous year. The integration course counts and open course counts are included without a time lag, since open courses are planned at the same time as integration courses. Hence, planners must weigh out both types of planned courses for the next year and make a decision based on the available resources from the previous year. Since the course counts only include realized courses, it is possible that courses were offered initially but then they did not take place due to a lack of demand. To control for this, we also include the number of participants in the model. When the number of courses decreases because of low demand, the number of participants would accordingly also decrease. $ε_{i t}$ is the error term.

The FEIS model, which determines the specific trends of VHS, is specified as follows:

y_{i t} - {\hat{y}}_{i t} = β_{1} (x_{1 i t} - {\hat{x}}_{1 i t}) + \sum_{j = 2}^{j = 11} β_{j} (x_{j i t - 1} - {\hat{x}}_{j i t - 1}) + ε_{i t}

(2)

Here, ${\hat{y}}_{i t}$ describes the predicted value from estimate (3) for each VHS and time point, $a_{1 i}$ is the individual intercept and $a_{2 i} t$ the individual slope.

{\hat{y}}_{i t} = a_{1 i} + a_{2 i} t + ε_{i t}

(3)

The same calculation is used for ${\hat{x}}_{1 i t}$ and ${\hat{x}}_{2 i t - 1}$ to ${\hat{x}}_{11 i t - 1}$ .

First, we estimate both models without including the resources ( $x_{2 i t - 1}$ to $x_{11 i t - 1}$ ) and then, in a second calculation, add the resources to the models.

We apply comparable FE and FEIS models to represent the effects of integration course participation on open course participation, while controlling for the number of courses and characteristics on the municipality level (Hypothesis 2):

y_{i t} = a_{i} + β_{0} + β_{1} x_{1 i t - 1} + β_{2} x_{2 i t} + β_{3} x_{2 i t}^{2} + \sum_{j = 4}^{j = 6} β_{j} x_{j i t - 1} + ε_{i t}

(4)

Here $y_{i t}$ describes the participation counts in open courses within a VHS $i$ at time $t$ .The value $x_{1 i t - 1}$ is the number of participants in integration courses of the previous year. The implementation of the time gap of 1 year is based on the assumption that potential open course participants must first notice a change before changing their action. $x_{2 i t}$ is the open course count in the current year. $x_{4 i t - 1}$ to $x_{6 i t - 1}$ are the values from the municipality level control variables.

Accordingly, we specify the FEIS model as follows:

y_{i t} - {\hat{y}}_{i t} = β_{1} (x_{1 i t} - {\hat{x}}_{1 i t}) + \sum_{j = 2}^{j = 5} β_{j} (x_{j i t - 1} - {\hat{x}}_{j i t - 1}) + ε_{i t}

(5)

Again, $y_{i t}$ is the number of participants in open courses, $x_{1 i t}$ is the current open course count. $x_{2 i t - 1}$ to $x_{5 i t - 1}$ are, respectively, the participation in integration courses in the previous year (the variable of interest) and the municipal data from the previous year (for control).

To test Hypothesis 3, we specify FEIS models. The assumption here is that segregation in VHS is moderated by the experiences of the local population in the respective community with migrants. The contact theory describes long-term mechanisms that depend on the proportion of migrants, but not on short-term and minor changes. Since the ICC of the variable “percentage of individuals without German citizenship in the municipality” is 0.93, we use the between-variance to test the moderator effect. For this, we choose two strategies: (1) including an interaction term in the FEIS model and (2) running the FEIS models for distinct samples. We build the interaction term in the FEIS model with dummy variables representing the percentiles of the variable “percentage of individuals without German citizenship” (divided into 33th; 33th-66th and 66th-100th percentile). In more compact notation compared with equation (5):

{\tilde{y}}_{i t} = β {\tilde{I P}}_{i t - 1} + δ ({\tilde{I P}}_{i t - 1} x_{i}) + β {\tilde{x}}_{i t - 1} + β {\tilde{x}}_{i t} + β ({\tilde{x}}_{i t - 1} x_{i}) + β ({\tilde{x}}_{i t} x_{i}) + ε_{i t}

(6)

Here, ${\tilde{y}}_{i t} = y_{i t} - {\hat{y}}_{i t}$ is the participation count in open courses. Equivalently, ${\tilde{x}}_{i t} = x_{j i t} - {\hat{x}}_{j i t}$ is the vector of the mean and trend-centered variables. In detail, ${\tilde{x}}_{i t - 1}$ is the vector of the time-lagged variables including population, number of Germans in the population and net household income. ${\tilde{I P}}_{i t - 1}$ is the participation count in integration courses (also mean and trend-centered). ${\tilde{x}}_{i t}$ is the current number of open course offerings. $x_{i}$ is the vector of the three dummies (33th; 33th-66th and 66th-100th percentile) of the proportion of migrants (mean 2007-2017) in the municipalities. This variable is time-invariant. The interaction effects between the proportion of migrants in the municipality and the other covariables are not reported due to limited space.

In FE models, interactions with time-constant variables are modeled by mean-centering the values of the time-constant variables within units (Giesselmann & Schmidt-Catran, 2018). Since the interaction variables are mean- and trend-centered in our FEIS model (Ludwig, 2015), an unproblematic interpretation of the effects is given. However, for greater certainty, we then estimate the FEIS model as described in equation (5) again in three separate models, which include three distinct samples of VHS. The samples are formed according to the proportion of migrants in the municipalities (33th; 33th-66th and 66th-100th percentile). We then compare the effects of participation in integration courses on the participation in open courses using z value statistics.

Results

In the following, we focus on the course count models investigating the structural displacement hypothesis on the supply side, and the participation count models for the segregation hypothesis on the demand side. Finally, we report the results of the interaction models regarding the context of the VHS and their participants.

First, we tested whether an increase in integration course counts affects open course counts (Hypothesis 1). The Hausman test showed that the FE model was preferable to the random effects model (χ² = 2,153; degrees of freedom [df] = 12). The results from the FE model that does not control for resources showed a substantial positive and significant effect of integration course counts on open course counts (Table 2, Model 1). On average, for each four (4.3) additional integration courses a provider conducts, we observed one additional open course in the same year. When controlling for available resources (Table 2, Model 2), the effect size decreased substantially to a nonmeaningful number. Hence, in the event that there would be no change in available resources, integration course counts would neither positively nor negatively affect open course counts.

Table 2.

Regular Courses Depending on Integration Courses in VHS—FE and FEIS Models.

	FE				FEIS
	Model 1		Model 2		Model 3		Model 4
	b	SE	b	SE	b	SE	b	SE
Integration courses	0.23***	0.02	0.06**	0.02	−0.11***	0.03	−0.10***	0.03
Participation in regular courses	0.04***	0.00	0.04***	0.00	0.04***	0.00	0.04***	0.00
Federal funds (t − 1)			−0.06***	0.01			−0.04***	0.01
EU funds (t − 1)			−0.00	0.00			−0.01	0.01
Job center (t − 1)			0.04***	0.01			0.01	0.01
Communities (t − 1)			0.00	0.00			−0.02***	0.00
Municipalities (t − 1)			0.01	0.01			−0.06***	0.01
Federal state (t − 1)			0.03***	0.01			0.02*	0.01
Fees (t − 1)			0.06***	0.01			−0.02	0.01
Other revenues (t − 1)			0.06***	0.00			0.02***	0.00
Human resources (t − 1)			0.20***	0.02			0.04*	0.01
Cooperation (t − 1)			0.19***	0.02			0.10***	0.02
Constant	311.32***	3.28	243.38***	4.33
N	864		864		864		864
Observations	8,640		8,640		8,640		8,640
R² (within)	.50		.55		.49		.51

Note. VHS = Volkshochschulen; FE = fixed effects; FEIS = fixed effects individual slopes; SE = standard error; EU = European Union.

p < .05. ^**p < .01. ^***p < .001.

The more restrictive FEIS models, which considered organizational specific profile trends, revealed opposing results. Irrespective of resources, additional integration courses here showed a negative association with open course counts (Table 2, Models 3 and 4). Here, around 10 additional integration courses would relate with a decrease of one course in open course counts.

Hypothesis 1 assumed no significant effect of integration course counts on open course counts. The results stand against this hypothesis.

We analyzed the effect of integration course participation on open course participation (Hypothesis 2). The Hausman showed that the FE model was preferable to the random effects model (χ² = 205; df = 6). The FE model (Table 3, Model 1) suggested a negative association between an increase in integration course participation and open course participation: With around two additional integration course participants, there was one open course participant fewer at the VHS in the following year. The effect also remained stable after including the regional context control variables (Table 3, Model 2).

Table 3.

Participation in Regular Courses Depending on Participation in Integration Courses in VHS—FE and FEIS Models.

	FE				FEIS
	Model 1		Model 2		Model 3		Model 4
	b	SE	b	SE	b	SE	b	SE
Participation in integration courses (t − 1)	−0.58***	0.03	−0.55***	0.02	0.10**	0.03	0.08**	0.03
Regular courses	10.96***	0.18	11.03***	0.18	11.54***	0.19	11.61***	0.19
Regular courses²/1,000	0.25***	0.03	0.27***	0.03	0.14***	0.03	0.11***	0.03
Migrants in municipality (t − 1)			0.04***	0.00			0.04***	0.01
Net household income (t − 1)/10⁶			−0.19***	0.02			−0.75***	0.08
Population in municipality (t − 1)			0.02***	0.00			0.02***	0.00
Constant	63.00	92.93	658***	681.73
N	863		863		863		863
Observations	8,630		8,630		8,630		8,630
R ²	0.50		0.27		0.49		0.50

Note. VHS = Volkshochschulen; FE = fixed effects; FEIS = fixed effects individual slopes; SE = standard error.

p < .05. ^**p < .01. ^***p < .001.

Again, accounting for VHS specific trends, the results from the FEIS models opposed the results from the FE models. Under control of unobserved growth heterogeneity, there was a significant positive effect: When the number of participants in integration courses increased by 10, the number of participants in open courses increased by 1 in the following year (Table 3, Model 3). When controlling for municipality specific variables, the effect size dropped to 0.08 (Table 3, Model 4). Consequently, the segregation hypothesis (Hypothesis 2) is not confirmed.

Hypothesis 3 was conditional to Hypothesis 2. Since Hypothesis 2 is not confirmed, Hypothesis 3 may not be confirmed either. Nevertheless, we analyzed whether the number of migrants in the spatial context (municipality) had a moderating effect on the association between integration course participation and open course participation. The results of our models suggested a negative interaction effect (Table 4, Model 1). In districts with a proportion of migrants up to 5.2% (a rather low proportion), there was a significant positive effect of integration course participation on open course participation (0.24). In districts with an average proportion of migrants (between 5.2% and 8.9%) the effect reversed. There was a negative significant effect of 0.25. In municipalities with a high proportion of migrants (>8.9%), the negative effect decreased slightly to 0.20. The distinct sample models (Table 4, Models 2, 3, and 4) also confirmed these results. When comparing the z scores we found a significant difference in the coefficients between the Models 2 and 3 in Table 4 as well as between 2 and 4 (Model 2 vs. Model 3: z = 4.3274, p = .00; Model 2 vs. Model 4: z = 2.48, p = .01; Model 3 vs. Model 4: z = −0.84, p = .40).

Table 4.

Participation in Regular Courses Depending on Participation in Integration Courses in VHS, Moderated by the Proportion of Migrants in the Municipality—FEIS Models.

	FEIS
	Model 1^a		Model 2 (33th)		Model 3 (33th-66th)		Model 4 (66th-100th)
	B	SE	b	SE	b	SE	b	SE
Participation in integration courses (t − 1)	0.24**	0.08	0.24***	0.05	−0.01	0.04	0.05	0.064
Participation in integration courses (t − 1) 33th to 66th (Ref. 33th)	−0.25*	0.09
Participation in integration courses (t − 1) 66th to 100th (Ref. 33th)	−0.20*	0.08
Regular courses	13.40***	0.15	13.40***	0.19	10.10	0.17	12.53***	0.31
Germans in municipality (t − 1)	0.00	0.02	0.00	0.01	0.00	0.01	−0.08***	0.01
Net household income (t − 1)/10⁶	0.44	0.38	0.44*	0.22	0.17	0.10	−1.36***	0.16
Population in municipality (t − 1)	0.00	0.01	0.00	0.01	0.01	0.00	0.03***	0.00
N	863		288		288		287
Observations	8,630		2,880		2,880		2,870
R ²	0.51		0.69		0.61		0.45

Note. VHS = Volkshochschulen; FEIS = fixed effects individual slopes; SE = standard error.

Interaction effects between the proportion of migrants in the municipality and covariables are not listed.

p < .05. ^**p < .01. ^***p < .001.

Discussion

This study examined the effects of continuing migration movements on the profile of public adult education centers (VHS) in Germany. To the best of our knowledge, our work is among the first studies to empirically address this issue on the organizational level. We focused on the associations between integration course counts and participation and open course counts and participation. Our analyses provide three main conclusions.

First, we found a negative association between the number of integration courses and the number of open courses that a VHS conducts. However, this association only held in the FEIS models, which controlled for VHS-specific trends in course counts, for example, a general increase or decrease of open courses and integration courses. We conclude that on average, an increase in conducting integration courses decreases the number of open courses, which a VHS would otherwise conduct in the case of no general trends. However, the effect size is rather small, since it would take 10 additional integration courses before there was a single open course fewer. This initially refutes our assumption that the increase in integration courses has no effect on the open course offer. Furthermore, the comparison of Models 3 and 4 in Table 2 showed that the resources in this issue do not have the importance we assumed.

Second, the results of the participation models revealed exactly the opposite pattern compared with the course count models. The FEIS models showed a weak, but positive association between integration course participation and open course participation. We conclude that, on average, an increase in integration course participation led to an increase in open course participation within a VHS in the following year. Our assumption that segregation processes occur in VHS as in other organizations under similar circumstances cannot be confirmed. This suggests that the situation in adult education centers is not comparable to schools or voluntary organizations. It may be assumed that the perception of in-group and out-group, as presupposed by social identity theory, does not take place in this context.

Third, depending on the proportion of migrants in the municipality of a VHS, the effect of integration course participation on open course participation varied substantially. The interaction effect between the proportion of migrants in the municipality and participation in integration courses on the participation in open courses was negative. We only observed an average positive effect in municipalities with a relatively low proportion of migrants. In municipalities with higher proportion of migrants, the effect disappeared. In any case, the negative interaction effect refutes the assumptions of contact theory. If this assumption were correct, we would have to observe the opposite. The observed effect rather agrees with the assumptions of the group threat theory.

Our findings therefore set limits on the idea that a VHS is able to conduct any number of additional integration courses on top of its open courses without an effect on open course counts. The negative effect of integration course counts on open course counts suggests a structural displacement of open courses to set in when the increase in integration courses is high (10 or more additional courses). However, the fact that the effect only shows to be negative when controlling for trends suggests unobserved growth heterogeneity, which means that the effect differs according to general trends in course counts during the observation period. Most VHS show a decline in open course counts. In this respect, the positive effect in the FE models and the negative effect in the FEIS models can be interpreted as follows—the VHS that are on a less declining or a positive open course count trend are more likely to have increased their integration course counts.

It is also worth noting that the negative effect in the FEIS model is insensitive to controlling for resources. When the change in the supply structure is not caused by a lack of resources in general, other factors, such as program planning, may come into focus. VHS act on behalf of the public interest and their actions follow a normative rationale rather than an economic rationale (Schrader, 2011). Although the interest of municipalities often lies in saving money, it can be assumed that program planners in VHS are also willing to set priorities that result from promoting societal interests such as the integration of migrants. In particular, the political and social debate around the “2015 Refugee Crisis” suggests that program planners have shifted the prioritization of the course offers. Hence, it appears as if they rather based their decisions on the perceived normative urgencies than on economic conditions.

Regarding participation, we did not find evidence for our general assumption that segregation sets in at a VHS with increasing integration course participation. This means that the number of participants in open courses was not found to decrease. Indeed, it actually increased with increasing integration course participations, when allowing for individual growth. The differences in the results between the FE models and the FEIS models showed that VHS with declining open course participation rates also had increasing integration course participation. One can assume that VHS with declining participation rates in open courses try to compensate for these declines by gaining additional funding from offering integration courses, which ultimately leads to higher integration course participation. However, when we controlled for trends in open course participation, we found that, on average, a positive effect of the participation rates in integration courses led to an increase in the participation rates in open courses in the following year. The best explanation for this positive effect is the assumption that participants from the integration courses also attend open courses in the following year (Brose, 2013; Lücker & Mania, 2014). This explanation is less plausible in municipalities with high proportion of migrants, where the effect again reverses. It can be assumed that the higher the proportion of migrants, the higher the possibility that integration course graduates established supportive social groups. This, in turn, may lower the motivation to establish contacts with members of the receiving society in open courses of a VHS. Respectively, when the proportion of migrants is low, integration course graduates may be more motivated to get involved with the local receiving population.

Limitations

Despite the exceptional data and models employed in our study, it has clear limitations. First, our analyses were based on organizational data. However, our theoretical considerations also refer to the individual level. Thus, we cannot directly test any of the assumptions that we made on the individual level. Consequentially, we also cannot test the mechanisms. The ideal data set to address this issue would include longitudinal microdata of participants nested in VHS. However, such data is currently not collected. Second, while we controlled the heterogenity of the receiving society based on the regional variations in proportions of migrants, we could not account for the full cultural and ethnic diversity of the German society. Third, we were not able to rule out reversed causality and unobserved variable bias.

Conclusion

Although course planners reduce the open course offerings for the benefit of integration course offerings, these dynamics were only observed on the supply side, since the number of participations in open courses has not decreased over time. Thus, no segregation effect was observed. Here, it is important to note that these dynamics on the supply side were not applicable to a lack of resources. Hence, it is of particular interest whether all parts of the open course offers are affected by the decline, or whether normative motives are more important than economic motives when deciding which course offers are dispensable. This question on the impact of reduced course offerings on different program areas should be addressed in future research. Furthermore, it would be of particular interest to compare how other providers of continuing education react to these social challenges in other reproduction contexts (Schrader, 2011).

The results also have implications for integration and migration policies, since they show that meeting the challenges of migration needs more than only sufficient resources: Actors of program planning in further education organizations and their evaluations also play a major role. Furthermore, the findings reveal that the effectiveness of integration courses is not independent from not-intended mechanisms and the regional context. However, further research and more evidence that focuses on the regional context as well as the key role of program planners is needed.

Footnotes

Previous Oral Presentations

25/02/2019—27/02/2019. The 7th annual conference of the Gesellschaft für Empirische Bildungsforschung (GEBF—Association of Empirical Educational Research) 2019, Cologne, Germany

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Andreas Martin

Ina Elisabeth Rüber

Author Biographies

Andreas Martin, PhD, is head of junior research group “Effects and revenues of education-political interventions in the field of continuing education” at the German Institute for Adult Education (DIE), Bonn, Germany. His research interests include supply and participation structures of continuing education, regional disparities in continuing education, and migration.

Julia Sophia Granderath is a PhD student at the German Institute for Adult Education (DIE), Bonn, Germany. Her research focuses on the effects of continuing education on integration and acculturation of migrants.

Ina Elisabeth Rüber is a researcher at the German Institute for Adult Education (DIE), Bonn, Germany. Ina Rüber researches the non-monetary effects and returns of continuing education.

References

Aberson

C. L.

Healy

Romero

(2000). Ingroup bias and self-esteem: A meta-analysis. Personality and Social Psychology Review, 4(2), 157-173. https://doi.org/10.1207/S15327957PSPR0402_04

Al Ramiah

Hewstone

(2013). Intergroup contact as a tool for reducing, resolving, and preventing intergroup conflict: Evidence, limitations, and potential. American Psychologist, 68(7), 527-542. https://doi.org/10.1037/a0032603

Allport

G. W.

(1954). The nature of prejudice. Addison-Wesley.

Bélanger

(2011). Theories in adult learning and education. Verlag Barbara Budrich.

Bilger

Behringer

Kuper

Schrader

(2017). Weiterbildungsverhalten in Deutschland 2016. Ergebnisse des Adult Education Survey (AES) [Continuing education behavior in Germany 2016—Results of the adult education survey (AES)]. W. Bertelsmann Verlag.

Blalock

H. M.

(1967). Toward a theory of minority-group relations. Capricorn Books.

Boeren

Nicaise

Baert

(2010). Theoretical models of participation in adult education: The need for an integrated model. International Journal of Lifelong Education, 29(1), 45-61. https://doi.org/10.1080/02601370903471270

Brose

(2013). Soziale Inklusion durch Erwachsenenbildung [Social inclusion through adult education]. Zeitschrift Für Bildungsforschung, 3(1), 47-62. https://doi.org/10.1007/s35834-013-0050-z

Brüderl

Ludwig

(2014). Fixed-effects panel regression. In Best

Wolf

(Eds.), The SAGE handbook of regression analysis and causal inference (pp. 327-358). Sage. https://doi.org/10.4135/9781446288146.n15

10.

Campbell

D. E.

(2006). Religious “threat” in contemporary presidential elections. Journal of Politics, 68(1), 104-115. https://doi.org/10.1111/j.1468-2508.2006.00373.x

11.

Destatis. (2020). Bevölkerung und Erwerbstätigkeit: Bevölkerung mit Migrationshintergrund - Ergebnisse des Mikrozensus 2019 [Population and employment: Population with migration background - results of the 2019 Mikrozensus]. https://www.destatis.de/DE/Themen/Gesellschaft-Umwelt/Bevoelkerung/Migration-Integration/Publikationen/_publikationen-innen-migrationshintergrund.html

12.

Federal Office for Migration and Refugees. (2018). Ausländische Staatsangehörige [Foreign citizens]. https://www.bamf.de/DE/Themen/Integration/TraegerLehrFachkraefte/TraegerIntegrationskurse/Organisatorisches/TeilnahmeKosten/Auslaender/auslaender.html?nn=282656

13.

Federal Office for Migration and Refugees. (2019a). Bericht zur Integrationskursgeschäftsstatistik für das Jahr 2018 [Report on integration course statistics for the year 2018]. https://www.bamf.de/SharedDocs/Anlagen/DE/Statistik/Integrationskurszahlen/Bundesweit/2018-integrationskursgeschaeftsstatistik-gesamt_bund.pdf?__blob=publicationFile&v=3

14.

Federal Office for Migration and Refugees. (2019b). Das Bundesamt in Zahlen 2018—Asyl, Migration und Integration [The federal office in figures 2017—Asylum, migration and integration]. https://www.bamf.de/SharedDocs/Anlagen/DE/Statistik/BundesamtinZahlen/bundesamt-in-zahlen-2018.pdf?__blob=publicationFile&v=14

15.

Giesselmann

Schmidt-Catran

(2018). Interactions in fixed effects regression models (DIW Berlin Discussion Paper No. 1748). https://ssrn.com/abstract=3227779

16.

Hammond

R. A.

Axelrod

(2006). The evolution of ethnocentrism. Journal of Conflict Resolution, 50(6), 926-936. https://doi.org/10.1177/0022002706293470

17.

Hjerm

(2007). Do numbers really count? Group threat theory revisited. Journal of Ethnic and Migration Studies, 33(8), 1253-1275. https://doi.org/10.1080/13691830701614056

18.

Hjerm

(2009). Anti-immigrant attitudes and cross-municipal variation in the proportion of immigrants. Acta Sociologica, 52(1), 47-62. https://doi.org/10.1177/0001699308100633

19.

Hochman

Davidov

(2014). Relations between second-language proficiency and national identification: The case of immigrants in Germany. European Sociological Review, 30(3), 344-359. https://doi.org/10.1093/esr/jcu043

20.

Kanas

Scheepers

Sterkens

(2015). Interreligious contact, perceived group threat, and perceived discrimination. Social Psychology Quarterly, 78(2), 102-126. https://doi.org/10.1177/0190272514564790

21.

Käpplinger

(2018). Addressing refugees and non-refugees in adult education programs: A longitudinal analysis on shifting public concerns. European Journal for Research on the Education and Learning of Adults, 9(2), 161-177. https://doi.org/10.3384/rela.2000-7426.rela242

22.

Käpplinger

(2020). Modernisation of organisations due to migration? Mixed blessings in adult education centres in Germany. European Journal for Research on the Education and Learning of Adults, 11(3), 293-308. https://doi.org/10.3384/rela.2000-7426.ojs1699

23.

Käpplinger

Reuter

(2019). Verschiebungen bei Personal- und Programmstrukturen an Volkshochschulen: Explorationen der Volkshochschulstatistik und Forschungsdesiderate [Shifts in staff and program structures at adult education centers: Explorations of adult education statistics and research desiderata]. Hessische Blätter Für Volksbildung, 2019(2), 142-150. https://doi.org/10.3278/HBV1902W142

24.

Laurence

Bentley

(2018). Countervailing contact: Community ethnic diversity, anti-immigrant attitudes and mediating pathways of positive and negative inter-ethnic contact in European societies. Social Science Research, 69(January), 83-110. https://doi.org/10.1016/j.ssresearch.2017.09.007

25.

Lochner

(2015). Integrationskurse als Motor für gesellschaftlichen Zusammenhalt? [Integration courses as a motor for social cohesion?]. Budrich UniPress.

26.

Lücker

Mania

(2014). Übergänge innerhalb der Weiterbildungseinrichtungen am Beispiel von Angeboten für Menschen mit Migrationshintergrund [Transitions within continuing education institutions using the example of offerings for people with a migration background]. Report. Zeitschrift Für Weiterbildungsforschung, 36(2), 71-83. https://doi.org/10.3278/REP1402W071

27.

Ludwig

(2015). XTFEIS: Stata module to estimate linear fixed-effects model with individual-specific slopes (FEIS). https://econpapers.repec.org/software/bocbocode/s458045.htm

28.

Makles

Schneider

(2015). Much ado about nothing?: The role of primary school catchment areas for ethnic school segregation: Evidence from a policy reform. German Economic Review, 16(2), 203-225. https://doi.org/10.1111/geer.12048

29.

Martin

Muders

(2018). The benefit of cooperation for adult education centres. Zeitschrift für Erziehungswissenschaft, 21, 383-405. https://doi.org/10.1007/s11618-017-0779-6

30.

Meisel

(2019). Herausforderungen an die öffentliche Erwachsenenbildung und strukturelle Ressourcenprobleme. Hessische Blätter Für Volksbildung, 4, Article 324. https://doi.org/10.3278/HBV1904W324

31.

Owens

Reardon

S. F.

Jencks

(2016). Income segregation between schools and school districts. American Educational Research Journal, 53(4), 1159-1197. https://doi.org/10.3102/0002831216652722

32.

Paluck

E. L.

Green

S. A.

Green

D. P.

(2019). The contact hypothesis re-evaluated. Behavioural Public Policy, 3(2), 129-158. https://doi.org/10.1017/bpp.2018.25

33.

Pettigrew

T. F.

Tropp

L. R.

(2006). A meta-analytic test of intergroup contact theory. Journal of Personality and Social Psychology, 90(5), 751-783. https://doi.org/10.1037/0022-3514.90.5.751

34.

Quillian

(1995). Prejudice as a response to perceived group threat: Population composition and anti-immigrant and racial prejudice in Europe. American Sociological Review, 60(4), 586-611. https://doi.org/10.2307/2096296

35.

Reichart

Lux

Huntemann

(2018). Volkshochschul-Statistik: 56. Folge, Arbeitsjahr 2017. wbv Media.

36.

Rocca

Carlsen

C. H.

Deygers

(2020). Linguistic integration of adult migrants: Requirements and learning opportunities (Report on the 2018 Council of Europe and ALTE Survey on Language and Knowledge of Society Policies for Migrants). Council of Europe. https://rm.coe.int/linguistic-integration-of-adult-migrants-requirementsand-learning-opp/16809b93cb

37.

Schaller

Abeysinghe

A. M. N. D.

(2006). Geographical frame of reference and dangerous intergroup attitudes: A double-minority study in Sri Lanka. Political Psychology, 27(4), 615-631. https://doi.org/10.1111/j.1467-9221.2006.00521.x

38.

Schneider

Schuchart

Weishaupt

Riedel

(2012). The effect of free primary school choice on ethnic groups: Evidence from a policy reform. European Journal of Political Economy, 28(4), 430-444. https://doi.org/10.1016/j.ejpoleco.2012.05.002

39.

Schrader

(2011). Struktur und Wandel der Weiterbildung [Structure and change in continuing education]. Bertelsmann W. Verlag.

40.

Tajfel

(1982). Social identity and intergroup relations. Cambridge University Press.

41.

Widany

Reichart

Ambos

Huff

(2020). Datennutzung der VHS- und Verbundstatistik. Potenziale für Bildungsforschung, -politik und -praxis [Data use of VHS statistics. Potential for educational research, policy and practice]. Zeitschrift Für Weiterbildungsforschung, 43(1), 75-95. https://doi.org/10.1007/s40955-019-0136-x

42.

Wiertz

(2016). Segregation in civic life. American Sociological Review, 81(4), 800-827. https://doi.org/10.1177/0003122416651312