Student selection in higher education – the organisational performance dilemma

ABSTRACT

Research on student selection mostly focuses on accepted applicants and the effects of selection procedures. In this sense, most samples seem to be biased, which is well-reflected in the literature. The present study investigates student selection regarding students who had been initially de-selected but finally succeeded in the admission process. Our results show that students from the de-selected group reveal lower performance, i.e. higher dropout, and thus lead to decreases in relative organisational performance. We argue that higher education institutions may be confronted with a performance dilemma if external stakeholders prefer institutional growth, and internal actors prefer maintenance of educational standards. In this situation, reduced graduation rates have the potential to undermine the output legitimacy of higher education institutions. Therefore, we recommend focusing more on the balance between academic standards, stakeholders’ demands, and processes of student selection. From a theoretical point of view, we suggest broadening the perspective and combining selection theory with organisation theory.

KEYWORDS:

• Higher education
• student selection
• organisation theory
• graduation
• dropout
• performance

Introduction

The selection of students is one of the most influential management instruments higher education institutions can use to respond to stakeholder demands (Ahola & Kokko, 2001). University management must decide who is allowed to enter the academic institution. However, the management must also strike a balance between the societal demand for the massification of higher education (Akalu, 2014; Alexander, 2000) and the need for measures against social inequality concerning the acquisition of education (Lynch & O’riordan, 1998) and for increasing diversity (Jungblut et al., 2015), using performance- and excellence-oriented elitist pre-selection criteria (Ashcroft et al., 2021). Hence, selection and massification do not necessarily go hand in hand. Selection refers to the setting of specific standards that lead to a homogenous student body, upholding a specified standard of excellence, and granting access to those who are ready for competition among the best and most creative minds (Silva et al., 2020). In contrast, massification refers to adjustments in the selection standards, e.g., by reducing entry scores (Marnewick, 2012) or adjusting the standards towards minimum thresholds, giving access to those who meet the most basic requirements to study further (Bore et al., 2009; Rawlinson & Burnard, 1978).

Regardless of the type of selection procedure used, it reflects a higher education institution’s preference for an exclusive or inclusive approach to student admission (Ahola & Kokko, 2001; Harman, 1994). The selection procedures may also relate to various stakeholders’ preferences. If higher education institutions and stakeholders’ preferences are aligned, tensions are unlikely to occur because educational programs and external expectations are congruent. The situation changes with diverging preferences.

Consequently, considering the conflicting demands of higher education institutions and their stakeholders, the type of selection procedure (massification or selection) implicitly influences stakeholders’ perceptions of a higher education institution’s performance and the assumed quality of teaching and learning, as uncovered by measurement instruments (Ball & Wilkinson, 1994; Pounder, 1999). Hence, admission procedures are key elements in the measurement of the performance of students and organisations, and they have the potential to create institutional pressure.

This paper aims to broaden the perspective on student selection. Consequently, we investigated the factors that influence student selection with an organisational perspective of stakeholder’s preferences for student performance. Based on the theoretical foundations from the literature review, this paper aims to improve the understanding of how stakeholders’ preferences for student selection can impact student and organisational performance.

We explore this by examining evidence of student performance at a cooperative state university that educates police inspector candidates (police students) in a federal state in Germany. The selection of police students is a special case of at-the-gate selection (Arnold, 2015): it is not the university itself that sets the admission criteria but a state authority responsible for police education and training under the administration of the federal state Ministry of the Interior. To investigate the effects of stakeholders’ selection preferences, we used a unique dataset from 2015. In that year, the Ministry of the Interior demanded a supplementary selection round to recruit additional police students who failed the initial selection process.

Our paper argues, in line with the results of research in other fields providing evidence, that selection procedures and organisational performance are related (Becker & Gerhart, 1996; Guest, 1997; Hiltrop, 1996). Furthermore, organisational factors, internal processes, and stakeholders’ views are becoming increasingly relevant for recent research because they complement former views on selection procedures (Ansmann & Seyfried, 2022; Neumann et al., 2021; Niessen et al., 2018). We argue that stakeholders’ preferences for increased student output may place the higher education institution into a dilemmatic situation. For example, lowered selection standards may lead to high numbers of graduates (high absolute output) in the long run. However, they may also lead – ceteris paribus – to higher dropout rates (lower relative output) and may be misinterpreted as decrease in organisational performance, when contextual factors are not taken into account.

Stakeholders’ preferences for student selection in the context of a cooperative university of applied sciences in Germany

Student selection and admission processes are internationally accepted methods to decide which student gets access to higher education. Nevertheless, there is no uniform implementation of selection instruments, and in various contexts, higher education institutions take into account the views and needs of relevant stakeholders (e.g., ministries, donors, public authorities, private enterprises, academics, higher education management, etc.). The German higher education system has increasingly established student selection procedures to find the best fit between study requirements and personal characteristics, such as abilities, skills, and interests. Particularly, since the dissolution of the German Central Office for the Allocation of Study Places (Zentralstelle für die Vergabe von Studienplätzen, ZVS), different selection procedures have been applied by different institutions (Schuler & Hell, 2008).

In the present article, we focus on a cooperative university of applied sciences for police and public administration in a German federal state. The higher education institution competes in its market segment with other higher education institutions for the best candidates (Marginson, 2013; Winston, 1999). As for many higher education institutions, preferences for specific and study program-related student selection signal an implicit desire for excellence and stratification comparable to that in elite universities aimed at being the favoured choice of applicants (Bloch et al., 2015). As such, student selection takes place within the context of a changing structure in an expanding but also selective German higher education system (Ertl, 2005; Müller & Klein, 2023).

At cooperative universities of applied sciences, stakeholders are relevant for student selection. In our specific case, the students are not necessarily selected under consideration of their capacity to pass a study program successfully, but they are, among other things, selected because they meet the criteria of their appointment authority or the training authority, which leads us to the overall selection procedure and other stakeholders’ preferences. Hence, the exact number of students admitted depends on the respective financial situation of the federal state and the number of police inspectors needed to replace those retiring (Terstiege & Krull, 2023).

Given the complex demands placed on police inspectors, an overall selection procedure resting purely on school report grades would not be sufficient, considering the discussions on their predictive validity (Vulperhorst et al., 2018). The special requirements of the police civil service and the number of applicants exceeding the number of study places underline the necessity of selection procedures (Rowold, 2015). Hence, the police inspector candidates undergo a complex multimethod selection procedure to ensure that the specific minimum requirements are appropriately met. In contrast to research universities or other universities of applied sciences, future police students do not apply at the educational institution but rather at the responsible state office. Thus, in our case, it is not the cooperative university but the state authority responsible for police education and training that sets the criteria for student selection. Furthermore, expectations regarding student selection are mainly influenced by the preferences of an additional stakeholder, namely the federal state’s Ministry of the Interior.

During the recruitment phase, police inspector candidates undergo a 3-day assessment based on the selection procedures. On the first and second day, a written recruitment test and examination are conducted by the police medical service. Applicants who meet the cognitive and health requirements are invited to the decentralised assessment centre in the premises of one of the 11 regional recruitment authorities. After the selection days, a post-selection and final application process take place. The successful applicants then begin the 3-year dual course of study for a Bachelor of Arts in Law Enforcement. After completing the study program, they are appointed as police inspectors.

In general, these selection procedures mainly reflect the preferences of the recruitment and training authorities and partly meet the demands of the higher education institution concerning the students’ academic abilities. Thus, the recruitment process comprises a mixture of medical examinations, non-academic assessments such as those of communication skills and professional behaviour (work sample tests), and knowledge-based assessments such as cognitive ability tests. Curriculum sampling tests, on the other hand, that mimic part of an academic program and have been shown to be strong predictors of performance in theoretical courses (Niessen et al., 2018), are not taken into consideration.

Stakeholder preferences tend to evolve when other actors influence the processes of student selection. For example, the number of selected candidates meeting the necessary criteria may be lower than the suggested recruitment figures of the Ministry due to the demographic changes Germany is going through (Terstiege & Krull, 2023), and this may reduce the size of the potential recruitment pool. In that case, the Ministry may intervene to recruit those candidates who nearly met the selection criteria but were not selected in the first round.

State of research and student selection theory

The literature on student selection processes is complex (Harman, 1994; Marnewick, 2012; van Ooijen van der Linden et al., 2017). Nursing, medicine, and psychology are considered disciplines at the forefront of student selection studies (Bore et al., 2009; Courneya et al., 2005; van Ooijen van der Linden et al., 2017; Vierula et al., 2020). However, research on student selection has become increasingly important for all fields of study (Ahola & Kokko, 2001). Research suggests a growing trend in harmonising selection procedures, study ability tests, assessment centres, and interviews among higher education institutions around the globe (Courneya et al., 2005; Mayer, 2013; Sladek et al., 2016).

A similar convergence can be found in the available evidence. Most studies have attempted to identify the factors that accurately predict students’ achievements (Boer & van Rijnsoever, 2021; Shen et al., 1995) and their future occupational practice or job performance (Ahola & Kokko, 2001; Bore et al., 2009). For example, some studies aimed to identify selection procedures that lead to the fewest prediction errors in student selection (Sladek et al., 2016; van Ooijen van der Linden et al., 2017). Consequently, research and practice using professional methods and admission procedures have led to increasingly complex instruments and identification of common standards (Berkowitz & Stern, 2018). However, the increasing convergence of the instruments also implies agreement regarding their weaknesses and the blind spots of the associated research.

Recent research shows that studies of student selection are ‘dominated by research on undergraduate medical students’ (Boer & van Rijnsoever, 2021, p. 12), a source of potential bias. Beyond this, researchers have only addressed the visible aspects of admission procedures. Most studies, therefore, focus on students who are successful in their applications (Sladek et al., 2016), no longer taking into consideration those students whose applications are not successful. Hence, researchers cannot test their hypotheses in a sample that includes selected and rejected students. The evidence for their conclusions is incomplete, leading to some doubt about the validity of the conclusions drawn (Sladek et al., 2016) and to a cumulative research gap. Accordingly, the implicit assumption about lower performance of unselected students relates to our first hypothesis:

H1:

Students whose applications are not successful in the first selection round, have a higher probability to drop out from the higher education institution.

In research and practice, student selection is an important area of analysis, with researchers offering a range of theoretical perspectives. While some studies have investigated attitudinal and motivational factors (Schuler & Hell, 2008), others have considered examination scores, socio-demographic, or procedural and organisational factors, requiring careful interpretation in some cases (Cerdeira et al., 2018; Vulperhorst et al., 2018).

Research on students’ academic success has a long tradition with rich retrospective data that have been carefully analysed over the years. The professionalisation and specialisation within the research field derived in part from overall developments in research methodology and methods. Hence, digitalisation, increased data availability, use of online tools, or new sophisticated instruments contribute to a better understanding of student selection and learning (Pohlenz et al., 2023; van Ooijen van der Linden et al., 2017; Vierula et al., 2020).

From our point of view, two broad research perspectives have emerged: the predictive validity of students’ capacities and skills, and the relationship between student selection procedures and academic success. These perspectives have become increasingly intertwined. Regarding research on the predictive validity of students’ capacities and skills, researchers have analysed personality and intelligence factors as explanatory factors and their influence on academic success (Backmann et al., 2019; Behling, 1998; Ferguson, 2003) in a bid to identify the most relevant factor(s) that predict(s) the best fit between higher education institutions and selected students. Some studies have explained the relationship between students’ competence, such as cognitive abilities, learning strategies, volitional and motivational competence aspects, and academic performance (Boer & van Rijnsoever, 2021). Other researchers, like Kappe and van der Flier, for example, used the Big Five personality factor model and found that the personality trait ‘conscientiousness’ is superior to other factors in predicting grade point average, length of time to graduation and specific performance criteria (Kappe & van der Flier, 2010). In fact, some recent studies have focused on aspects such as emotional intelligence (Pienimaa et al., 2021), reasoning (Vierula et al., 2020) and communication skills (Erozkan, 2013). The literature on student selection identifies a sector-specific usage of the factors. For example, selection based on cognitive factors and personality traits is frequently applied in medical schools. In contrast, skills-based selection is associated with ‘commercial and industrial sectors as well as the police, military and other government services’ (Bore et al., 2009, p. 1068). Researchers have highlighted the strengths and drawbacks of various selection procedures (Boer & van Rijnsoever, 2021; Bore et al., 2009).

Regarding the relationship between student selection procedures and academic success, academic success is generally considered a dependent variable that is explained by factors such as school grades, intelligence or personality test scores, or student performance in admission tests (Felinto de Farias Aires et al., 2018; Shen et al., 1995; Silva et al., 2020). Studies investigating such factors have also analysed the relationship between the methodological design of admission procedures, study progress, and students’ occupational performance after graduation (Powis et al., 1992). Based on the second research perspective, we propose the following hypotheses:

H2:

Students who have lower scores in the overall admission procedure, have a higher probability to drop out from the higher education institution.

H3:

The effect of the written recruitment test scores on dropout is stronger for those in the first selection group than for those in the second selection group (previously deselected students).

H4:

Students who perform better in secondary education in mathematics, have a lower probability to drop out from the higher education institution.

H5:

Students who perform better in secondary education in English, have a lower probability to drop out from the higher education institution.

Research design and methods

Typically, higher education institutions do not receive any detailed information about the students who are not selected but focus on the performance of the selected student group. In the present case, the federal state's Ministry of the Interior made an exception in 2015 and demanded that – because of state police staff shortage – students who did not pass the initial selection test were enrolled. Therefore, 241 additional applicants, who had initially been rejected, were enrolled and began their studies at the cooperative university of applied sciences. Together with the students from the first selection round (n = 1,596), we obtained an overall sample of 1,837 students. This setting provides unique insights into students who usually remain ‘hidden’ to empirical research (Sladek et al., 2016).

The present investigation is based on a quantitative analysis designed as a case study of a German cooperative university of applied sciences for public administration and police. To test the hypotheses presented above, we did not collect self-reported data. Instead, we relied on a process-generated and administrative dataset provided by the university. Although we do not think that an administrative dataset has the same quality as a scientific dataset, particularly for criteria like validity or flexibility (Figlio et al., 2016), we assumed that measurement errors in our dataset occurred randomly instead of having a dataset that was systematically biased (e.g., by social norms, external expectations, common methods etc.). When measurement errors are noticed, most false data entries would have been quickly corrected, because students want to avoid transcription errors in their performance records.

For the statistical analysis, we implemented binary logistic regression and measured our dependent variable as graduation (1) or dropout (0). With this type of measurement, we do not need to enter the extended discussions about using grades as measures of academic achievement which is contested for various reasons (Vulperhorst et al., 2018; Weatherton et al., 2021).

Our first independent variable, related to hypothesis H1, distinguishes between the first and the second selection of students measured with binary values of 1 (first selection) and 0 (second selection). The second independent variable is the centralised score of the overall admission procedure including all testing results leading to a mean value of 0. We decided to centralise the score, because the coefficients in logistic regression are scale-dependent. The centralised score of the overall selection procedure is linked to our hypothesis H2. For H3 we use an interaction term between the centralised score of the written recruitment test and the group variable of the first and second selection. Furthermore, to investigate H4 and H5, we measure the school grades in mathematics and English. The grades in secondary school range from 0 to 15 points, where 0 points indicate ‘insufficient’, and 15 points are considered as ‘very good’. To check the robustness of our results, we also included three control variables, namely gender, age and migration background (see Table 1).

Table 1. Descriptive statistics of all selected variables.

Variable	Values	Arith. Mean	Min.	Max.	Std. dev.	No. of obs.	In %
DV: graduation
	graduation = 1					1,602	87.2
	dropout = 0					235	12.8
IV: group
	first selection = 1					1,596	86.9
	second selection = 0					241	13.1
IV: score procedure (centralised)	higher values indicate better testing performance	0	−8.04	14.41	4.52	1,837
IV: score recruitment test (centralised)	higher values indicate better testing performance	0	−18.33	24.67	7.65	1,837
IV: mathematics	15 points (“very good”) to 0 points (“insufficient”)	7.57	0	15	2.94	1,837
IV: English	15 points (“very good”) to 0 points (“insufficient”)	7.70	1	15	2.50	1,837
CV: gender
	female = 1					647	35.2
	male = 0					1,190	64.8
CV: age		23.30	18.72	40.02	3.57	1,837
CV: migration background
	yes = 1					209	11.4
	no = 0					1,628	88.6

Source: own dataset (DV = dependent variable, IV = independent variables, CV = control variable)

Analysis and results

To test our derived hypotheses, we ran a binary logistic regression (see Table 2). According to our coding, the coefficients for the group variable, the testing scores and their interaction term which, relating to H1, H2 and H3, all reveal the predicted sign. The positive coefficients indicate that being in the second selection and having lower test scores increases the probability of dropping out. However, the coefficient for the selection round is not significant, and hence, H1 can not be confirmed. In contrast, the coefficients for the overall score of the admission procedure and the interaction of admission test score and the group variable are significant. The results for the overall admission procedure reveal that students who pass the three admission days successfully have a higher probability of graduating (H2). Considering H3, the positive effect of testing scores on graduation is stronger for those students in the first selection because the testing score reflects a larger positive performance spectrum. In contrast, the testing scores of the second selection contain the narrow end below the critical testing values for the initial selection round which blurs the lines between success and failure in relation to testing results. The results are significant and confirm our hypotheses H2 and H3.

Table 2. Results of binary logistic regression.

Variables	Regression coeff.	Std. error	p-Value	Odds ratio, Exp(B)
group (0 = second selection)	0.012	0.243	.961	1.012
score overall procedure	0.046*	0.021	.031	1.047
group*recruitment testing score	0.043***	0.011	.001	1.044
grade mathematics	0.123***	0.027	.001	1.131
grade English	−0.015	0.031	.639	0.986
gender (0 = male)	0.316*	0.161	.050	1.372
age	−0.013	0.020	.521	0.987
migration background (0 = no)	−0.045	0.212	.831	0.956
constant	1.389*	0.608	.022	4.013
Nagelkerkes r-squared	0.065

Note: The dependent variable is student success (0 = dropout), and additional model statistics are: log likelihood = −670.041; AIC = 1358.083, BIC = 1407.726; omnibus-test chi-square = 64.946, df = 8, p < 0.001; Hosmer-Lemeshow-test chi-square = 5.353, df = 8, p = .719; * p < .05, ** p < .01, *** p < .001.

For testing our fourth and fifth hypotheses, we included the variables of the school grades in mathematics and English. While the coefficient for English grades neither reveals the expected sign nor a meaningful coefficient, the result for grades in mathematics shows the expected sign and is significant. Hence, better grades in mathematics increase the probability to graduate.

As Table 2 also illustrates, our results are robust under consideration of selected control variables. We analysed the independent variables by taking into account gender (0 = male), age and migration background (0 = no migration background). Only gender reveals a significant positive coefficient (p = 0.0495) indicating that male students have a higher probability to drop out. This is in accordance with results from existing studies which show that under specific conditions female students perform better (Barrow et al., 2009; Richardson & Woodley, 2003).

To further check the robustness of our model, we investigated our statistical results to identify possible violations such as multicollinearity. The results show that the independent variables do not correlate with values higher than r = 0.553, which is the bivariate correlation between the two centralised scores for the admission procedure and the group. In a detailed analysis of the independent variables, we calculated the Variance Inflation Factors (VIF). The values range from 1.05 to 1.16 and do not indicate multicollinearity.

Considering the explanatory power of our model, pseudo r-squared reveals comparatively low values (Nagelkerkes r-squared = 0.065 and Cox and Snell r-squared = 0.035). Nevertheless, the omnibus test is significant indicating that the model offers an improvement in comparison to a null model. However, the low explanatory power fits well to the overall idea of our research problem, because the de-selection of students may be considered as a gradual process, which means that for example in our case, those students who are at the top of the de-selected students presumably do not differ very much in their performance from those students who belong to the bottom group of the selected students.

Beyond this, the investigation of the organisational perspective requires a shift from individual to aggregate data. Hence, we calculated the dropout and graduation rates for the students in our sample (Table 3). Although the second selection led to an overall increase of 201 students with successful graduation in terms of absolute numbers (from 1,401 to 1,602 graduates), the results revealed a decrease in the relative success of the higher education institution from 12% to 13% dropout quota. This is because the second selection reveals a higher dropout quota (17%) than the first selection (12%). The results show a significant statistical association between selected or de-selected students and graduation or dropout (Fisher’s exact test, p = 0.039). The results support our assumption about the organisational performance dilemma to some extent. An increase in absolute output corresponds to a decrease in relative output (ceteris paribus).

Table 3. Graduation and dropout rates in the first and second selection.

	1^st selection		2^nd selection		all
	n	in %	n	in %	n	in %
Graduation	1,401	88	201	83	1,602	87
Dropout	195	12	40	17	235	13
All	1,596	100	241	100	1,837	100

Source: own dataset

Regardless of the declining relative performance of the organisation, the results from Table 3 also indicate that a previously deselected group of 241 students from which 83% succeeded, were excluded in the initial selection round. Hence, the organisational performance dilemma and the comparatively low dropout rate of 13% reflect the partly paradoxical and diverging demands the cooperative university of applied sciences is confronted with. This underlines the relevance of reflecting student selection procedures in a broader organisational context.

Discussion of results

Considering our specific case, the additional selection of police officer candidates may result in a decrease in organisational performance when based on reduced graduation rates or increased dropout rates resulting in an organisational performance dilemma (reduced relative output). However, this conclusion is only valid under ceteris paribus assumptions. The organisation can respond in different ways, either by lowering or by maintaining professional standards. Lowering of professional standards could lead to constant dropout rates, while the maintenance of standards could increase the dropout rates. The situation becomes more nuanced when we for example integrate considerations about the role and access to and provision of resources in our line of argumentation (Salancik & Pfeffer, 1974). Hence, it might be relevant if there are incentives or sanctions for the higher education institution to reduce the dropout rates. If this is not the case, the higher education institution may resist the Ministry’s preferences for more graduates. In sum, our research results add two perspectives to the existing debate: first, the organisational performance dilemma, which we assume is generalisable for other educational organisations, and second, a demand for further integration of organisation theory into conceptualisations of student selection.

Generalising the organisational performance dilemma

Although our investigation only considered a single cooperative university, the empirical results seem to be generalisable to other higher education institutions (Müller-Benedict & Gaens, 2020). For example, higher education institutions in Germany enjoy a comparatively high degree of autonomy (Wolter, 2004) implying that the principal state ministries have only legal oversight (Rechtsaufsicht). However, debates on the role of target agreements, introduction of competitive models, and orientation towards educational demands have considered these aspects as equivalents to the functional oversight (Fachaufsicht) of the state ministry. Although higher education institutions in Germany enjoy autonomy, they need to interact with ministries in capacity planning. Similarly, we may assume the same for private higher education institutions if their main sponsors opt for an extension of existing enrolment capacities. Beyond this, we assume that the organisational performance dilemma and its possible effects on organisational legitimacy may matter not only for higher education institutions but also for educational institutions that have restricted influence on capacity planning, such as primary, secondary or vocational schools, and nursing schools. For organisations where an increase in input units does not automatically lead to an increase in output units, the demand for increased output may lead to an organisational performance dilemma and legitimacy pressures (ceteris paribus).

Organisations can face the dilemma in one of two ways: on the one hand, an organisation can choose to not respond to the new selection conditions, e.g., not adjust their study programs which presumably leads to increased dropout rates. On the other hand, an organisation can adjust its internal processes to the enlarged input to maintain the organisational performance level. These adjustments can be made by either increasing the capacity of the organisation, if resources are granted externally, or by reducing quality standards internally. All options create legitimacy pressures and may even jeopardise the organisation’s legitimacy.

Integration of organisation theory in the research on student selection

In addition to existing concepts of student selection and organisational performance, we suggest integrating a complementary theoretical perspective, because selection procedures unfold their influence in specific organisational settings and institutional contexts. In accordance with recent literature, organisational factors, internal processes, and stakeholders’ views are becoming increasingly relevant and complement views on selection procedures (Neumann et al., 2021; Niessen et al., 2018). Therefore, we suggest combining student selection theory with organisation theory.

An institutional organisation theory perspective on selection procedures emphasises that these procedures also reflect accepted norms and organisational culture revealing ‘great symbolic power’ (Hoffman & Lowitzki, 2005, p. 470). They balance the external demands of stakeholders and internal demands for the best fitting students to ‘protect the status quo’ (Razack et al., 2014, p. 177). For example, there is evidence of a positive influence of selection procedures on student performance and satisfaction, well-being, study time, and reduced dropout rates (Edwards et al., 2012), but some higher education institutions may apply selection procedures for other reasons (Kotzee & Martin, 2013). Contrastingly, other studies have indicated that poor selection procedures may result in an increased administrative burden on the university and inappropriate use of limited resources for lecturers and students, and the institution might be regarded as a misguided financial investment on the part of public or private organisations (Oude Egbrink & Schuwirth, 2016). Hence, processes of student selection do not seem to be independent from contextual factors, such as administrative structures, stakeholders’ expectations and preferences, or formal as well as informal rules.

Conclusion

In our paper, we combined research on student selection with organisation theory and assumed that the performance of higher education institutions may be significantly influenced by internal and external expectations. Hence, the present paper provides a case study of a cooperative university for applied sciences that is confronted with changing external demands regarding student output. We analysed the data of police candidate students taking up their studies in the year 2015, when the ministry, as one of the main stakeholders, decided to enrol additional students to meet the demands for increased numbers of police inspectors.

Herein, we derived five hypotheses: students who do not belong to the first selection round (H1) and students with lower testing scores in the overall admission procedure (H2) have a higher probability to drop out from the higher education institution. Furthermore, we hypothesised an interaction effect between the selection round and the written recruitment assuming that being in the second selection and having lower recruitment test scores have a higher probability to drop out from the higher education institution (H3). Finally, students with better grades in secondary education in mathematics and English have a lower probability to drop out from the higher education institution (H4 and H5).

In our findings the coefficients for H2 and H3 are significant and indicate the correct sign. The results reveal a positive effect of testing scores on graduation for the students selected in the first round indicating that the predictive power of the admission test diminishes gradually with lower testing scores. Furthermore, our results supported H4 implying that better school grades in mathematics reduce the probability to drop out.

Beyond this, we investigated the graduation rates of the first and the second selection. Our results showed that meeting the external expectations by increasing the number of recruited students led to a marginal decline in graduation rates or an increase in dropout rates, respectively. Hence, although the overall performance of the organisation increases in terms of the number of students, its relative performance decreases. This may result in an organisational performance dilemma that has the potential to undermine the organisation’s legitimacy due to reduced relative performance measures, in line with the general demand to broaden the view about selection processes in research (Arnold, 2015; Neumann et al., 2021). Thus, selection procedures are embedded in a broader organisational context and unfold their impact in institutional environments, surrounded by various internal and external expectations.

The implications of our research are twofold. First, the study underlines the necessity for a stronger coupling of external stakeholders’ preferences, organisations’ internal standards regarding study programs, and the curriculum. The findings offer additional perspectives in addressing increasingly diverse groups of students and on how far stakeholders need to reform organisational support structures or to adjust their expectations regarding organisational output. Second, the results widen our view on the interactions of selection procedures and the organisational context in which internal and external expectations for massification may result in a dilemma for the organisation because students who fail, contribute to an increased dropout rate that could be misunderstood by external actors as a reduction in organisational performance.

Our results suggest further research questions, such as how higher education institutions strike a balance between internal and external demands regarding student output; how selection processes and admission instruments are fine-tuned or embedded in their institutional environments (Arnold, 2015; van Ooijen van der Linden et al., 2017); and how they are loosely or tightly coupled with the inner organisational structures, cultures, processes, and demands.

Acknowledgements

We would like to thank the two anonymous reviewers and the editor for their constructive comments. Special thanks go to “Reviewer 1”, who supported us with helpful suggestions that contributed significantly to the development of our work.

Disclosure statement

No potential conflict of interest was reported by the author(s).