1 Introduction

Given the progressive technological and societal changes of our times, there is a need for individuals to continue their training after completing their initial education to acquire new knowledge and competences. This increases the demand for high-quality education and training. To meet this higher demand, professional teaching staff are needed, among other things, which is increasingly becoming the focus of scientific and educational policy. Of central interest here are employment and living conditions and their influence on the professionalism of adult educators (Autorengruppe wb-personalmonitor 2016). In particular, ongoing participation in continuing education and training is considered to be a driving factor for the professionalism of adult educators (Tippelt 2007). As has been empirically proven for the school context (Hattie 2009) it is also assumed for the field of adult and continuing education that educators are substantially responsible for the learning success of the learners (Hattie 2009; Collins and Pratt 2011; Schrader et al. 2010). Therefore, they must adapt their subject-specific, subject-didactic, and didactic competences to the changing requirements. In Germany, there is a mandatory obligation for schoolteachers to participate in Continuing Education and Training (CET), but there is a lack of such a regulation for teaching staff in adult and continuing education. Due to the increasing demand for high-quality continuing education, as well as the continuously changing requirements for adult educators, they need to invest in their professionalism themselves. In addition to the lack of a legal framework for mandatory participation in CET, such obligations are also not enforced by the educational institutions in which adult educators work due to the heterogeneous employment structure of these institutions in Germany. Consequently, the decision to participate in continuing education courses is entirely down to the individual and depends on individual decisions. The question arises as to what influences the individuals’ decisions to participate. Previous research on determinants of participation in continuing education shows a complex relationship between structural contextual conditions and individual characteristics (Kaufmann and Widany 2013). Participation in continuing education is thus seen as the result of interactions between actors at the micro, meso and macro levels (Boeren et al. 2010; Boeren 2017). In general, the determinants of participation in CET are well studied. At the micro level, characteristics of employment and socio-economic context have been shown to be important predictors. However, there are only a few studies that deal with living and working conditions and their effect on the participation in CET of adult educators (Alfänger et al. 2016; Autorengruppe wb-personalmonitor 2016; WSF Wirtschafts- and Sozialforschung 2005). In this context, it was shown that employment in the field of adult and continuing education is often characterised by heterogeneous and atypical forms of employment. Measured by the total number of employment relationships, employment relationships within the categories of fee-based work and service contracts dominate: About 70% of employment relationships in adult education are based on this type of contract (Autorengruppe wb-personalmonitor 2016). Furthermore, 38.4% of employees in adult education do not have statutory health insurance cover (Autorengruppe wb-personalmonitor 2016). Often embedded in a context of unfavourable socio-economic conditions and atypical employment, the situation of adult educators in Germany is described as precarious (Dörre 2017). In the discourse on the professionalisation of adult educators, the professional situation, characterised by uncertainties, is an almost obstructive factor. For this reason, the employment situation in the further education sector is often described as a “tension field between required professionalism and growing risk of precarious employment” (Dobischat et al. 2010, p. 164). Precarity has been well studied in large empirical studies (Bartelheimer et al. 2012; Bellmann et al. 2008; Brinkmann et al. 2006; Kalleberg 2009; Schmid and Protsch 2009). Specific risks associated with precarity—employment insecurity, low or unstable incomes, health risks or limited opportunities for participation as well as for further training—have also been empirically studied and discussed (Benach et al. 2014; Brehmer and Seifert 2008; Canivet et al. 2016; Lewchuk 2017). Despite the importance of teaching staff, very little research has been conducted on the influence of precarious situations on the participation in CET of adult educators.

In our study, we therefore examine the influence of precarious employment on adult educators’ participation in CET, focusing on the uncertainty and the associated inability to plan one’s life and career development in the medium or long term that is typical of precarious life situations (Bourdieu 1998; Bultemeier et al. 2008; Castel 2011). For this, we have used data from the wb-personalmonitor 2014 that addresses employees working in open-access continuing education in Germany. Methodologically, we have estimated linear and logistic regressions using the inverse-probability-weighted regression-adjustment (IPWRA).

2 Theoretical framework

2.1 Precarious employment and participation in further education

Due to the increased flexibility of work in developed countries, forms of employment are also changing. Within the service sector, the field of continuing education is an illustrative example of the dynamic development of modern forms of work (Alfänger et al. 2014). In this context, especially in Germany, this field is strongly characterised by atypical forms of employment (Elias 2017). The definition of atypical employment relationships is created through their difference from the so-called standard employment relationship (Mückenberger 1985; de Grip et al. 1997). The term “standard” in this context refers both to a regulatory model and to the fact that this type of employment relationship is expected to be the norm (Schoukens and Barrio 2017). More precisely, the standard employment relationship can be defined as a “stable, open-ended and direct arrangement between dependent, full-time employee and their unitary employer” (Walton 2016, p. 112). The following criteria of the standard employment relationship can also be found in literature: Full-time employment, integration into social security systems, permanent employment relationship, identity of work and employment relationship, bound by the employer’s instructions (Mückenberger 1985). All forms of employment that do not meet these criteria are considered to be atypical. Accordingly, atypical work can be more easily defined by what it is not than by what it is. Basically, however, the concept covers a variety of forms of work that deviate from the norm (Rodgers 1989). As atypical employment relationships increase steadily, aspects of precarious forms of employment are also being discussed more frequently in this context. Here, the boundaries between atypical and precarious employment are blurred. Also, the terms atypical and precarious are often used synonymously. Treu (1992), for example, prefers to speak of “precarious employment”, since such employment relationships have already become so widespread that the term “atypical” is no longer appropriate. However, atypical does not necessarily mean precarious. Nevertheless, previous studies show that atypical workers are exposed to significantly higher precariousness risks than workers in standard employment relationships, who are also not free of such risks, but experience them at a far lower level (Seifert and Keller 2011). So, standard employment relationships do not automatically protect against precariousness.

Precarious work is not a new phenomenon, rather it is as old as humanity itself (Castel 2000). But it is only at the beginning of the twenty-first century that precarity became a central theme in academic research and literature. It found its way into academic discourse partly as a reaction to political mobilisations, that especially in Europe took place against a background of unemployment and social exclusion (Kasmir 2018). In general, precarious work is defined as employment that is “uncertain, unpredictable, and risky” (Kalleberg 2009, p. 2) from the worker’s point of view; it can encompass a wide range of employment relationships. In the literature, other concrete characteristics of precarious work are not consistently defined. Depending on the perspective, the criteria may vary. A comparative analysis based on four dimensions of precariousness as distinct from the normal employment relationship is increasingly being proposed (Rodgers 1989; Seifert and Keller 2011).

In the following paragraphs, the four dimensions and their assessment regarding a possible risk of precariousness are briefly explained. The first dimension is income, or more precisely, a subsistence level of income. This can be evaluated based on the low-wage threshold. According to the internationally accepted definition, the threshold for low wages is less than two-thirds of the median wage of full-time employees (OECD 2006). In April 2021, the low-wage threshold in Germany was gross earnings of € 12.27 per hour (Statistisches Bundesamt 2021). Accordingly, lower hourly earnings are considered as low wages.

Second, there is employment stability. This refers to a period of continuous employment, i.e., an employment relationship that is as uninterrupted as possible. This is a condition for an independent income. Employment stability does not mean securing a single specific job (job security), but rather individual employability with possibly changing employment relationships (employment security), also with different employers, in different activities and occupations. The decisive factor here is to what extent the transitions between the employment relationships are seamless (Brehmer and Seifert 2008).

Third, there is employability. The term refers to the ability to enter into, maintain and/or replace an employment relationship with another employment relationship (Hirseland et al. 2008). Against the background of rapid structural change and the resulting changes in requirements and qualifications, employability is becoming increasingly important as a condition for employment stability. A central prerequisite for maintaining employability is lifelong learning and consequently the participation in CET (Brehmer and Seifert 2008). And fourth, there is social security. This refers to sufficient material security against the social risks of illness, unemployment and old age through the social security systems (Brehmer and Seifert 2008). Based on the mentioned criteria, different forms of precariousness, such as income precariousness or employment precariousness, can be differentiated. It should be noted, however, that the boundaries between precarious and non-precarious status are blurred (Brehmer and Seifert 2008). Moreover, the isolated occurrence of one of these characteristics is usually not sufficient to describe an employment relationship as precarious; rather, it is the combination of several characteristics that makes an employment situation precarious (Baron and Hill 2018). For example, this is the case when employees receive a lower income, are employed in an unstable occupation, and have limited access to training on the job. Due to low and also insecure incomes, it is difficult for them to compensate for their skills deficits on their own and thus to keep up with changing skills requirements, i.e., to secure their employability (Seifert and Keller 2011). Thus, socio-economic disadvantages accumulate, which subsequently leads to an increased perception of insecurity in general, and planning insecurity in particular (Bourdieu 1998; Castel 2011; Dörre 2017).

While participation in continuing education as a central indicator of professionalism, and its factors, is generally well studied, there are only a few studies that explicitly deal with the living conditions of teaching staff in adult and continuing education and the influence of these conditions on their individual participation in CET. The classic socio-demographic characteristics that influence participation in continuing education are education, age, and gender. The positive correlation of formal education attainments and participation in CET is one of the central results of research on adult education (Becker and Hecken 2009; Kruppe and Baumann 2019). The effect of increasing age is initially positive up to middle age, while a negative influence on training activity can be observed for older people who are nearing the end of their working careers (Hubert and Wolf 2007; Kruppe and Baumann 2019). Analysing participation in CET by gender, a positive effect can be observed for the participation of men. However, this effect is no longer found when activity-related characteristics are controlled for. Thus, at present, the influence of gender is still ambiguous (Hubert and Wolf 2007; Görlitz and Tamm 2011).

Previous studies dealing with characteristics of the fields of activity, the professional career, the qualifications as well as the individual living conditions of teachers in adult and continuing education have shown, among other things, that work in this field is often characterised by heterogeneous and atypical forms of employment (WSF Wirtschafts- and Sozialforschung 2005; Autorengruppe wb-personalmonitor 2016). Being embedded in a context of unfavourable socio-economic life circumstances and atypical forms of employment, the insecurity factors can accumulate, whereby the risk of precariousness increases. The influence of precarious living conditions on the continuing education behaviour of teaching staff in adult and continuing education has barely been investigated up to now. In this regard, the concept of uncertainty is of great importance.

2.2 Certainty, uncertainty, risk, ambiguity

The study investigates the influence of precarious employment on participation in continuing education and focuses on the uncertainty and the resultant inability to plan one’s life and professional development for the medium or long term—that is typical of precarious life situations (Bourdieu 1998; Castel 2011; Bultemeier et al. 2008). However, this ability is seen as a condition for rational behaviour and any economic calculus (Bourdieu 1998; Castel 2011). In contrast to previous studies, we differentiate the concept of uncertainty and in doing so we distinguish between risk with a known probability of occurrence and ambiguity, where the probability of occurrence of risks is completely unknown (Kron 2006).

However, terms like risk, uncertainty and insecurity lack precision and have different connotations in scientific decision theory than in everyday language. Moreover, various researchers use these terms interchangeably, which results in a lack of consensus on their exact definitions and a clear distinction between the concept of risk and the concepts of uncertainty and insecurity (Mildner and Boeckelmann 2011). Furthermore, many researchers limit their analysis to decision making under risk and do not consider decisions under uncertainty, or even assume that all forms of uncertainty can be summarised in terms of risk (Gigerenzer 2018). However, rational decision-making under uncertainty is not the same as rational decision-making under risk (Gigerenzer 2018). For this reason, the different terminology of the concepts will be discussed in the following and it will be made clear which linguistic differentiations are used in the paper.

Colloquially, terms such as uncertainty, risk and ambiguity are used to describe limited causal imputability and future-open decisions (Keiner 2005). But this does not correspond to the scientific differentiation proposed in decision theory. Still, this differentiation is important and above all necessary for the selection of the theoretical foundation and an appropriate operationalisation (Keiner 2005).

In principle, decisions can be made under certainty and uncertainty. Certainty refers to the fact that the occurrence probability of action consequences is known and amounts to 100%. Uncertainty occurs in all decision-making situations where the outcome and its probability of occurrence are not precisely known. Thus, decisions under risk and decisions under ambiguity are considered as subcategories of decisions under uncertainty (Keiner 2005). In order to distinguish the two concepts from each other, it is relevant to consider to what extent the probability of occurrence of the possible consequence of action is known. A decision under risk involves quantifiable probabilities. However, these probabilities do not amount to 100% and are therefore fraught with risks. Decisions that are classically referred to as rational are made in exactly this risk mode, since the different probabilities of occurrence are known and can therefore be weighed against each other. For decisions under ambiguity, such an estimation of probabilities is not possible (Müller 1993; Gigerenzer 2018) because sometimes even the alternatives for action are unknown.

Consequently, we can see that decisions under ambiguity are not the same as decisions under risk. In the following section, decision-making under ambiguity will be discussed and an approach will be taken to make it theoretically tangible. This is then applied to the decision to participate in CET.

2.3 Decision-Making under ambiguity

Theoretical explanations of individual participation in continuing education are often based on action theories. Economic approaches such as the human capital theory (Becker 1964) explain participation in continuing education as an individual educational decision and as a result of a rationally made investment (Kaufmann and Widany 2013). Classical rational choice theories (RCT) assume rationally acting actors who try to maximise their expected utility through their behaviour and decisions. In doing so, they evaluate alternative courses of action with regard to the expected utility of each one and select the course of action that generates the highest possible return with the least possible use of resources. Accordingly, individuals only invest in (continuing) education if the expected utility is higher than the expected costs (Esser 1999). Therefore, the individual decision to participate in CET can be seen as a cost-benefit analysis (Boeren et al. 2010). By specifying the value-expectancy theory into a subjective expected utility (SEU) theory, only the subjectively assessed costs, benefits and success expectations are considered (Esser 1999). In this context, the individual educational decision is still a result of rationally made investments, but it is no longer exclusively oriented towards economic benefit calculations. In addition, the perfectly informed actor is no longer assumed; the concept of bounded rationality is becoming more important. The concept of bounded rationality refers to a decision that, taking into account the cognitive limits of the decision-maker, can still be considered rational. In this context, cognitive limits mean the available knowledge as well as the ability to calculate costs and benefits. According to this, people are in principle uncertain, albeit to varying extents, about their expectations regarding certain consequences of action (Simon 1990). The concept of bounded rationality thus responds to the fact that decisions—according to some modern decision theories—are always to some extent vague (Winter and Kron 2015). This especially applies to people who are affected by uncertainty due to precarious working and living conditions. Some more recent decision theories consider the vagueness of decisions, as well as the so-called ambiguity problem, which also contradicts the predictions from classical rational choice theories. The Ellsberg (1961) addresses the ambiguity problem and illustrates that people have an aversion to ambiguity, preferring known risks over unknown risks—this challenges the traditional concept of rational decision-making under uncertainty. This paradox thus challenges the assumption that individuals always maximise expected utility and suggests that people prefer known probabilities even when the potential outcomes are the same. It also implies that individuals would accept higher risks if ambiguity were lower. This insight underscores the importance of subjective attitudes towards uncertainty and ambiguity in decision-making, as opposed to solely relying on objective probabilities. Overall, the Ellsberg paradox demonstrates the complexities of human decision-making and the role that subjective factors play in shaping our choices under uncertainty. It has also led to the development of alternative decision-making models that account for ambiguity aversion, such as Kron (2006) integrative theory of action, which we will use subsequently to explain decisions under ambiguity.

To avoid the problem of ambiguity and to take limited rationality into account, Kron (2006) proposes a modification of the classical RCT. Here, an additional variable is introduced into the existing decision equation that includes the vagueness of decisions, the so-called expectation vagueness (EV). This results in the following formula for the expected utility:

$$EU_{i}={\sum }_{j=1}^{n}p_{ij}\mathrm{*}\left(1-EV_{ij}\right)*U_{j}$$
(1)

This formalisation corresponds to the observation that individuals prefer an alternative course of action even if the probability of occurrence (p) of a benefit is lower, but is more certain to estimate. This further development is a significant one. Besides, this modification is suitable as a theoretical explanation for decisions under ambiguity since the variable of expectation vagueness represents exactly this ambiguity. Thus, if the actor is completely vague about their expectation, the variable EV takes the value 1. The expected value consequently becomes zero, no matter how great the utility of the consequence of the action is. In cases where the actor is completely certain about their expectation (EV(A) = 0) and therefore there is no expectation vagueness, then—and only then—the common modelling of p within the expected utility theory holds. All values between zero and one reduce the expected value of action alternative A accordingly (Kron 2006). In relation to our research question, participation in CET is considered as one possible action alternative among many other possible action alternatives. The focus in this study is on the expected probability of the action alternative “participation in CET”.

Individuals in precarious situations are more inclined to choose less favourable alternatives if they can calculate these and thereby have a higher degree of certainty. Applied to the participation in CET by adult educators, this means that participation occurs even if the professional benefit is low but known. The field of adult education offers an almost perfect framework to empirically test the theoretical assumptions, given the highly heterogeneous composition of the teaching staff and the diversity of employment conditions. As the personnel in continuing education mainly work in the external labour market, which is characterised by strong competitive conditions (Sengenberger 1987), it is plausible to assume that they take purposeful and utility-maximising considerations into account regarding the decision to participate in CET. Therefore, they can hardly afford not to make cost-benefit analyses. However, the aforementioned precariousness risks, arising from involvement in unusual and atypical forms of employment, are likely to lead to a heightened perception of uncertainty among the adult educators, thereby limiting their rationality. At the same time, it is observed that occupational motivational attitudes, which are central components of professional competencies, are strongly pronounced in this field (Autorengruppe wb-personalmonitor 2016). These motivational attitudes are also significantly influenced by working and employment conditions, and they also have an impact on participation in further education. Based on this, we assume that, under the control of the motivational aspect, the effect of perceived uncertainty on participation in further education should remain unchanged.

3 Hypothesis

Following the theoretical framework of the Ellsberg Paradox, which is considered in Kron’s integrative theory of action, it is assumed that decisions that are certain are preferred to decisions that are uncertain, even if the success probability is lower. If the level of uncertainty is high, one is more likely to consider choosing another “worse” action alternative, of which the probability is, however, known. Participation in continuing education is seen as a rational investment in one’s own human capital. However, an investment is only made if it is associated with returns. But those who are affected by uncertainty have limited ability to make cost-benefit calculations. Due to atypical and partly precarious living and employment conditions, adult educators are affected by uncertainty and the associated inability to plan their lives and professional development in the medium or long term. This follows the concept of ambiguity.

Based on the theory presented, the following hypothesis can be derived:

H1

Uncertainty regarding the assessment of one’s own career development opportunities leads to a reduction in participation in CET among adult educators.

4 Data and methods

4.1 Data description

To test the hypothesis, we used cross-sectional data from the wb-personalmonitor 2014 (Autorengruppe wb-personalmonitor 2016). The survey was conducted in cooperation between the Federal Institute for Vocational Education and Training (BIBB), the German Institute for Adult Education—Leibniz Centre for Lifelong Learning (DIE) and the Department of Business Education/Vocational Education and Training at the University of Duisburg-Essen. With the wb-personalmonitor, data on employees in adult and continuing education were collected for the first time, with the aim of achieving representativeness. To compensate for the sample bias, the survey data were weighted and extrapolated by provider type and form of employment (Autorengruppe wb-personalmonitor 2016). For Germany, this is still the current and representative data basis on personnel in adult and continuing education. The content of the survey focused on the qualifications of the employees and their employment situation, as well as the working conditions and their evaluation by the respondents. The wb-personalmonitor was designed as a full census; the population is the personnel working in open-access continuing education in Germany. The data collection was conducted in two stages, the second stage constitutes our analysis sample with N = 1243.

4.2 Variables

4.2.1 Uncertainty

The article focuses on the question of the effect of precarious living and employment conditions on the participation in CET of adult educators. The integrative theory of action theorises uncertainty, or rather ambiguity, as decisive for decision-making, especially regarding investments in one’s own professional career. As it is assumed that precarious living and employment conditions are associated with an increased potential of ambiguity regarding one’s own professional development, the item on perceived employment perspective (“How do you think your professional situation will change in the next few years? Will it improve, stay the same or get worse?”) is defined as the independent variable. Respondents were asked to rate their career perspectives for the next year, for the next 2–3 years and for the next 5 years on a 5-point scale (1 = improve, 2 = stay the same, 3 = get worse, 4 = will retire in this period, 99 = I can’t assess). Those respondents who are unable to assess their professional situation for the next 1–3 years are considered uncertain and thus ambiguous in our context. The variable was then binary coded (0 = no uncertainty, 1 = uncertainty) and forms the expectation vagueness within the decision model in the integrative theory of action.

For the variables used, item non-response cannot be considered random. To avoid bias, these values were imputed. Originally, however, there were missing values on this variable for about 1/3 of the sample. For this, we use the multiple imputation by chained equations (Rubin 1987; Azur et al. 2011). We use all control variables that are later included in the model as predictors. This also imputes the few missing values on the other variables. In order to adequately account for the estimation error, 100 imputations are made. Accordingly, the models are estimated 100 times and the results are then combined according to Rubin’s rules (Rubin 1987). A visualisation of the imputation by trace plots, and a table displaying the number of imputed observations, can be found on OSF (https://osf.io/fsvhu/?view_only=7d97628250d44d14979c49b2f09c4554).

4.2.2 Participation in CET

One of the key indicators of professionalism is participation in CET. This is completely voluntary for adult educators and therefore a result of rationally made investments and decisions. In order to examine how uncertainty affects the decision to participate in continuing education, the teachers’ own participation is used as an outcome variable. Participation in CET is measured with the item “Have you attended or are you currently attending at least one general or vocational continuing education course yourself in the last 12 months?”. The item is measured as a dichotomous variable (0 = no, 1 = yes). For those who reported participating in continuing education, the number of continuing education courses is also taken into account. The corresponding item in the questionnaire is: “How many continuing education courses have you participated in during the last 12 months?” It is an open question; the number given is treated as a continuous variable in the calculation.

4.2.3 Demographics and further variables

As it can be assumed that perceived uncertainty is attributed to a variety of individual characteristics, these characteristics need to be controlled. For this purpose, the wb-personalmonitor provides information on income, working hours, number of jobs in adult and continuing education, vocational qualification, employment status, age, and gender. Based on existing variables, we also generate variables that give us information about the net equivalent income and the proportion of income from working in adult education.

4.2.4 Estimation strategy

To test the theoretically derived hypothesis, linear and logistic regressions are estimated, depending on the operationalisation of the outcome. Because it is assumed that the perceived uncertainty is not random, but largely based on the control variables described earlier, a potential selection bias is methodologically compensated by using the inverse-probability-weighted regression-adjustment (IPWRA). IPWRA involves assigning weights to each observation based on the inverse of the estimated probability of being in the observed group. These weights are then incorporated into regression models to adjust for the potential selection bias. By giving higher weights to observations with lower probabilities of selection and lower weights to observations with higher probabilities, IPWRA aims to balance the sample and create a pseudo-population that represents the overall target population more accurately. We calculated the weight using a logistic regression of all control variables on the treatment.

$$p_{i}\left(X_{i}=1| z_{i}\right)=\frac{1}{1+e^{{-z_{i}}}}$$
(2)

Here, pi denotes the dependent variable describing the probability whether respondent i is affected by ambiguity as a function of control variables z. Then, e is the base of natural logarithm—Euler’s number.

Depending on whether the treatment is received (see Eq. 3) or not (see Eq. 4), the formula for the IPWRA weight is as follows:

$$w_{i\left(EV=1\right)}=\frac{1}{p_{i}}$$
(3)
$$w_{i\left(EV=0\right)}=\frac{1}{\left(1-p_{i}\right)}$$
(4)

The weighted observations are then used in a regression analysis to estimate the effect of treatment or group membership on the dependent variable. The weighted regression models using IPWRA provide unbiased estimates by accounting for the potential confounding effects introduced by the selection process. It helps to reduce the bias that could arise from the non-random assignment of individuals to different groups or conditions (Wooldridge 2007, 2010).

When using the continuous measurement (number of CET courses attended) as operationalisation of the independent variable and considering IPWRA, we specify the linear regression model as follows:

$$(Y_{i}={\ss}_{0}+{\ss}_{1}X_{i1}+\varepsilon )w_{i}$$
(5)

Here, Y denotes the dependent variable describing the number of CET courses attended for respondent i. Then, \({\ss}_{0}\) is the constant and \({\ss}_{1}\)is the regression coefficient. Then, Xi1 is the independent variable for the respondent i, while ε is the error term. By using the IPWRA, we already control for those variables that are theoretically assumed to influence the perception of uncertainty, whereby the influence of uncertainty on participation in CET could then be influenced. Thus, additional control variables can be omitted when estimating the effect of uncertainty on participation in CET.

All statistical analyses were conducted with Stata (version 17). The corresponding do-file as well as all tables can be found on OSF (https://osf.io/fsvhu/?view_only=7d97628250d44d14979c49b2f09c4554).

5 Results

5.1 Descriptive statistics

The descriptive statistics of the analytical sample are provided in Table 1.

Table 1 Descriptive statistics of the analytical sample
Full size table

5.2 Regression models

Regarding the dichotomous measure of the independent variable indicating participation in CET, we estimate a logistic regression model including the IPWRA weighting. The results displayed in Table 2 reveal, that perceived employment uncertainty does not have a significant effect on participation in CET. Neither the model nor the effect shows significance.

Table 2 Logistic Regression on participation in CET
Full size table

Regarding the continuous measure of the independent variable indicating the number of CET courses attended, we estimate a linear regression model including IPWRA weighting. The results presented in Table 3 show that perceived employment uncertainty leads to a significant reduction of 0.5 in the number of CET courses attended (p = 0.019). Based on the results, we can confirm Hypothesis 1.

Table 3 Linear Regression on CET courses attended
Full size table

6 Discussion

The much discussed and frequently demanded professionalism of adult educators is a multi-layered concept, both theoretically and methodologically. We have taken on this complexity and made an attempt to investigate the influence of living and working conditions on a central indicator of professionalism—one’s own participation in continuing education. From our analyses, the following conclusions can be drawn: The results differ considerably depending on the operationalisation used for the dependent variable. When a dichotomous measurement of the dependent variable is used, based on agreement or disagreement with the question about one’s own participation in CET activities, a negative effect of perceived uncertainty on participation in CET is observed—but it is not statistically significant.

When the participation is operationalised based on the number of CET courses attended, a significant negative effect can be observed. Thus, perceived employment uncertainty reduces participation in CET by approximately half a course per year. Following this operationalisation, we can confirm our hypothesis. When following the former operationalisation, the hypothesis would consequently have to be rejected. Teachers therefore still participate in continuing education, but to a reduced extent, i.e., they invest less. This confirms the theoretical assumption that uncertainty leads to a lower investment in one’s own human capital. Therefore, higher ambiguity makes this action alternative less attractive. Nevertheless, teachers follow the basic behavioural pattern—participation in CET; this can be seen as professionality. In this context, however, a possible selection effect should also be mentioned. Adult educators who generally do not participate in CET may not participate for reasons other than the employment insecurity we have studied. This can indeed be assumed, as we have methodologically well-defined uncertainty and the associated ambiguity. In doing so, we also control for other dimensions of precarity (income precarity and employment precarity). Therefore, the effect of ambiguity is well isolated, which eliminates confounding effects for observable characteristics. Thus, other individual non-activity-related unobservable characteristics could influence one’s own participation in CET.

With regard to precariousness research, it is of particular interest that here—to our knowledge—the effects of precarious insecurity have been studied for the first time, which goes beyond the conventional understanding of insecurity. Our results indicate that the issue of precarious employment not only concerns the occupational risks themselves, but above all, the inability to assess these risks. This reduces the actor’s ability to make rational decisions. The ambiguity of such life situations reduces the possibility of rationally weighing costs, benefits, and risks. On the contrary, in precarious situations there is a danger that alternative courses of action with little benefit will be chosen simply because the risks seem calculable. Precarious life situations thus seem to harbour the danger that actors make bad, irrational decisions.

Nevertheless, we have obtained an ambivalent result regarding the hypothesis. Thus, our study provides an interesting insight into how strongly results and the consequent acceptance or rejection of hypotheses depend on the respective operationalisation.

7 Limitations and conclusion

Although our study is one of the first to investigate the influence of insecure living and employment conditions on the participation of adult educators, it has certain limitations. The data used is almost 10 years old, but it is still relevant for our research question, as the assumed relationship between uncertainty and participation in continuing education is of a generic nature and does not necessarily require time-specific examination. While longitudinal studies would be needed to determine whether precarious living conditions have a long-term or temporary impact (Kraemer 2008; Fuller 2009), our focus was on examining the generic effect. Recent descriptive surveys (Martin and Schrader 2022) confirm that precarity is still relevant in the field of adult and continuing education. For this reason, but also due to the lack of more recent representative data on the personnel, we used the available data to address our research question.

Nevertheless, due to the cross-sectional design, controlling for unobserved heterogeneity is not readily possible and causal statements are limited.

Furthermore, the subjective assessment of perceived uncertainty may not be sufficient to operationalise precarity and ambiguity. More current data and a more accurate way of operationalising uncertainty caused by precarity could allow further research on this topic. In this context, we would like to mention a current project at the German Institute for Adult Education (DIE) which deals with the professionalisation and competence development of teachers in adult and continuing education (TAEPS). This will provide a data basis to address further research questions concerning teachers in adult and continuing education.

Our study provides additional evidence for the assumption that uncertainty in the assessment of risks hampers rational decision-making. This is particularly significant for adult educators, who are in a highly competitive market; they have to act as rational actors in the external secondary labour market, which is challenging due to uncertainty in their living conditions. Thus, our research builds upon descriptive findings that have shown the existence of precarity and goes a step further by studying its effects on the professional behaviour of adult educators. In this way, we are bringing a topic onto the research agenda that will become increasing relevant due to societal and labour market developments, such as the progression of digitisation and access to online training at lower costs, which may also influence participation in continuing education.

Looking ahead, it would be possible to expand the research question to other employment sectors and professions. The distinction between risk and ambiguity could again be taken into account and researchers could examine how occupation-related uncertainty affects participation in CET. For this purpose, more up-to-date, representative data could be used, that may allow a more appropriate operationalisation of ambiguity.

In conclusion, in our study it became clear that adult educators already act professionally with regards to the examined professionality indicator. However, whether this professional behaviour may partially be based on self-selection into the field of activity and thus on individual, non-task-specific factors, cannot be determined at this point. Nevertheless, it would be desirable if the organisation of employment situations in adult and continuing education were not only fraught with risks. Rather, transparency and dissemination of information could be used to enable employees to practice some form of risk management. Even if the chances of continued employment may not be certain, at least knowing and calculating the risk could be beneficial. This already works in other self-employed activities in traditional professions. Concerning the discourse on the development of professionalism of adult educators, the influence of employment conditions on key professionalism indicators should still be addressed by future research.