Do public frontline managers involve frontline professionals in managerial decision-making in response to crisis? Evidence from Danish hospitals during COVID-19

ABSTRACT

This article investigates whether public frontline managers respond to crisis by involving frontline professionals more in managerial decision-making by utilizing a survey among physicians employed at public hospital departments before and during the COVID-19 crisis. The findings show a positive relationship between how affected the departments were by the crisis and how much the frontline professionals were involved in operational decisions about how to solve problems in the organization. Furthermore, exploratory analyses reveal a potential convex relationship between crisis affectedness and involvement in strategic decisions about the overall goals, direction, and priorities of the organization.

KEYWORDS:

• Crises
• frontline managers
• decision-making
• employee involvement
• COVID-19

Managers from across sectors, types of organizations, and organizational echelons are central actors during crises where decision-making processes become more uncertain and complex (e.g. Ansell, Sørensen, and Torfing 2021; Comfort and Okada 2013; Garretsen et al. 2022; Van der Wal 2020; Van Wart and Kapucu 2011). Recently, studies have been published that specifically investigate the role of public frontline managers during crises, as public frontline managers are managers situated at the frontline of public organizations where public services are delivered to users (e.g. Bertelsen, Lindholst, and Hansen 2022; Lund and Andersen 2022). However, these studies do not inform us on how public frontline managers make decisions in response to crises, where there are increased levels of complexity and uncertainty in decision-making.

This article combines insights from the literature on crisis management (e.g. Christensen, Lægreid, and Rykkja 2016; ‘; T Hart, Kouzmin, and Rosenthal 1993) with insights from the literature on employee involvement in decision-making (e.g. Kelman, Sanders, and Pandit 2016; Nicholson-Crotty, Nicholson-Crotty, and Fernandez 2017; Somech 2010), to investigate how public frontline managers make decisions in response to crises.

The main theoretical argument of the article is that the increased uncertainty and complexity tied to decision-making at the frontline of public organizations during crises mean that frontline managers will involve frontline professionals more in managerial decision-making to increase the amount of information available in these decision-making processes. Frontline professionals are frontline employees such as nurses, teachers, physicians, police officers, and caseworkers, who possess specialized theoretical knowledge and occupational norms that they have acquired through their education and work (e.g. Andersen and Pedersen 2012; Cecchini and Harrits 2022). The article ties this theoretical argument to two different types of managerial decisions, namely strategic decisions concerned with overall goals and priorities of the organization (e.g. Bryson 2004; Maynard-Moody and McClintock 1987; Poister et al. 2013), and operational decisions about how to solve problems in the organization (e.g. Jung 2018; March and Simon 1993; Simon 1997).

The article uses COVID-19 as an example of a crisis and tests the theoretical argument by utilizing a panel dataset with 338 observations from 169 physicians employed at 143 different hospital departments before and during the COVID-19 crisis. Findings from a two-way fixed effects analysis reveal a positive linear relationship between how affected the departments were by the COVID-19 crisis and how much more the managers would involve the physicians at the department in making operational decisions, but not in making strategic decisions. In addition, an exploratory analysis suggests that when it comes to strategic decisions, a potential convex relationship exists between how affected the departments were by the COVID-19 crisis and involvement in strategic decisions. That is, when it comes to strategic decisions, there is no difference in the level of involvement when comparing high and low degrees of crisis affectedness, however, it seems that at medium degrees of affectedness, managers might involve frontline professionals less in strategic decisions, compared to high and low degrees of affectedness. The article thus demonstrates that the extent to which public frontline managers involve frontline professionals more in managerial decision-making in response to crisis depends on both the type of decision and the degree to which the organization is affected by the crisis.

Crises and involvement of frontline professionals

The role of public frontline managers during crises

Crises are situations ‘[…] in which there is a perceived threat against the core values or life-sustaining functions of a social system that requires urgent remedial action in uncertain circumstances’ (Rosenthal, Charles, and T Hart 1989, 10; see also Christensen, Lægreid, and Rykkja 2016). Crises demand urgency to act coupled with uncertainty regarding the causes of the crisis, the best course of action to deal with the crisis, and the potential consequences of decisions (Christensen, Lægreid, and Rykkja 2016, 888; Weible et al. 2020, 228; see also Rosenthal, Charles, and T Hart 1989).

Despite of the central role of public frontline managers during crises (Ansell, Sørensen, and Torfing 2021; Bertelsen, Lindholst, and Hansen 2022; Lund and Andersen 2022; Van der Wal 2020), few studies explicitly investigate public frontline managers decision-making in response to crises. One exception is a qualitative study in the UK healthcare context which illustrates that crisis decision-making involves addressing information asymmetry initially for the individual organization and subsequently handling information ambiguity in joint decisions with other organizations (Phillips, Roehrich, and Kapletia 2021). The study informs us of the sequence of how managers deal with decisions during a crisis, but it does not inform us how public frontline managers make decisions in response to a crisis.

When exposed to a crisis, public frontline managers have to make decisions in direct response to the crisis and in response to decisions made at higher organizational and political levels in order to alleviate the negative effects of the crisis on the organization (e.g. Boin and Hart 2003; Boin and Lodge 2016). One way of thinking about managerial decisions is to divide them into strategic decisions and operational decisions. Strategic decisions are decisions concerned with overall goals, direction, and priorities of the organization (e.g. Bryson 2004; Maynard-Moody and McClintock 1987; Poister et al. 2013), whereas operational decisions are concerned with how to solve problems related to day-to-day activities in the organization (e.g. Jung 2018; March and Simon 1993; Simon 1997).

Both strategic and operational decision-making in response to crises is inherently uncertain, as crises calls for revised goals, direction, and priorities of the organization, while also creating unforeseen and new problems related to the day-to-day activities in the organization (e.g. Ansell, Sørensen, and Torfing 2021; Lund and Andersen 2022; Van der Wal 2020; Van Wart and Kapucu 2011). Therefore, managers at the frontline of public organizations cannot rely on standard procedures and incremental decision-making when responding to crises (e.g. Ansell, Sørensen, and Torfing 2021). The question, then, is how frontline managers in public organizations make these managerial decisions in response to crises.

Public frontline managers’ decision-making in response to crises

To develop theoretical expectations about how public frontline managers make decisions in response to crises, the article draws on the theoretical framework of bounded rationality to combine insights from the literature on crisis management with the literature on employee involvement in decision-making. Bounded rationality posits that even though decision-makers are intentionally rational, they are not able to reach optimal decisions because of a combination of the general limits of human cognition and the complexities and uncertainties of their environment (Jones 2003; Simon 1997).

According to the literature on bounded rationality, decision-makers facing conditions of increased complexity and uncertainty can, broadly speaking, choose between at least two decision-making strategies. One is to try to simplify the decision-making process and thereby reduce complexity, and the other is to search for more information to inform the decision (Bendor 2015; Kelman, Sanders, and Pandit 2016; Lindblom 1959, 1965; March and Simon 1993).

One way to simplify a decision-making process is to centralize it (e.g. Altamimi et al. 2023; Staw, Sandelands, and Dutton 1981, ‘; T Hart, Kouzmin, and Rosenthal 1993). Decisions in organizations can be centralized on two dimensions: the hierarchy of authority and the degree of employee involvement in decision-making. Hierarchy of authority relates to the structure of decision-making, as it denotes where in the organizational hierarchy decisions are executed, whereas degree of employee involvement is related to the process of decision-making, as it denotes how many individuals are involved in organizational decision-making (Andrews et al. 2007, 58; Hage and Aiken 1967). This article is concerned with the dimension of employee involvement, as the focus is on the decisions made by managers at the frontline of public organizations. In this sense, the structural dimension, the hierarchy of authority, is fixed.

The potential pros and cons for a manager to centralize the decision-making process by involving employees less in decision-making can be viewed along two dimensions. On one dimension, leaving individuals out of a decision-making process reduces transaction costs and, thus, the complexity of the decision-making process because the decision is taken by a smaller group of individuals, making it possible to speed up the process (e.g. Drennan, McConnell, and Stark 2015; Schomaker, Hack, and Mandry 2021, 1281; ‘; T Hart, Kouzmin, and Rosenthal 1993). On the other dimension, leaving individuals out of the decision-making process leads to a loss of information and knowledge from the individuals left out (e.g. Altamimi, Liu, and Jimenez 2023; Andrews et al. 2007; Ashmos, Duchon, and McDaniel 1998; Latham, Winters, and Locke 1994). There is thus a trade-off between speed and information, which might be central to the frontline manager when deciding whether to involve employees in decision-making. The important question in this article is whether public frontline managers respond to crisis by involving frontline professionals more or less in strategic and operational decisions, compared to the decision-making process that they have in place at the frontline of their organization in situations without crisis.

The need for specialized theoretical knowledge and information about how to make strategic and operational decisions when responding to crises might be so great that leaving employees out of the decision-making process will not be a useful strategy at the frontline of public organizations. That is, the general uncertainty and complexity tied to strategic and operational decision-making are amplified during a crisis (e.g. Ansell, Sørensen, and Torfing 2021; Christensen, Lægreid, and Rykkja 2016). Therefore, the potential increased decision-making speed gained by leaving employees out of the decision-making process does not make up for the knowledge and information lost by excluding these individuals from the process. The amount of uncertainty that arises because of the required amount of specialized theoretical knowledge and information is too great at the frontline of pulic organizations.

Considering this need for specialized theoretical knowledge, public frontline managers might therefore choose to involve their employees more in managerial decision-making to increase the amount of available knowledge and increase the number of individuals working on the decisions (Kelman, Sanders, and Pandit 2016; Lindblom 1965; March and Simon 1993).

Involving frontline professionals in managerial decision-making in response to crises

Because of the complexity in public service delivery, many public frontline employees are frontline professionals who possess specialized theoretical knowledge and occupational norms that they have acquired through their education and their work (e.g. Andersen and Pedersen 2012; Cecchini and Harrits 2022; Jakobsen et al. 2018). Frontline professionals often have more knowledge and information about how an organization functions and how work processes relate to the goals of the organization than frontline managers do (e.g. Andersen and Pedersen 2012; Ansell, Sørensen, and Torfing 2021; Brehm and Gates 1997; Cecchini and Harrits 2022). When frontline professionals are involved in decision-making, it enables them to share information about how the organization functions and what works with their managers (Boyne et al. 2004; Jakobsen et al. 2018; Pasha 2018; Staniok 2017). This information can be used by managers to solve problems related to day-to-day activities in the organization (operational decisions), but also in terms of informing decisions about the organization’s priorities, overall goals, and strategies (strategic decisions), as the information can help the managers understand how strategic decisions ultimately influence service delivery in the organization.

The literature on employee involvement in decision-making indicates that managers might be more inclined to accept the potential costs associated with involving employees when they face decisions that are risky or complex and require processing of many informational inputs (Kelman, Sanders, and Pandit 2016; Nicholson-Crotty, Nicholson-Crotty, and Fernandez 2017), such as both strategic and operational decisions during crises. Indeed, involving frontline professionals in decision-making makes it possible to ‘[…] seek and synthesize multiple sources of information and perspectives’ (Kelman, Sanders, and Pandit 2016, 466) because ‘[m]ultiplicity copes with the inevitability of omission and other errors in complex problem solving’ (Lindblom 1965, 151).

Summarizing the line of argument in the theory section, we might expect that when the uncertainty and complexity related to both operational and strategic decision-making increases at the frontlines of public organizations because of a crisis, frontline managers respond to this by involving frontline professionals more in these decisions. That is, we expect a positive linear relationship between how affected an organization is by a crisis and how much frontline managers involve frontline professionals in strategic and operational decisions. These expectations can be expressed in the following hypotheses:

H1:

There is a positive linear relationship between the extent to which a public organization is affected by a crisis, and how much the frontline managers will involve frontline professionals in operational decisions.

H2:

There is a positive linear relationship between the extent to which a public organization is affected by a crisis, and how much the frontline managers will involve frontline professionals in strategic decisions.

Research design

Empirical setting

To investigate the hypotheses, we require a setting where public frontline managers have been exposed to a crisis to varying degrees and, in response to this crisis, had to make both strategic and operational decisions. Such a setting existed at Danish hospital departments at the beginning of the COVID-19 crisis. On 13 March 2020, Denmark had 19 new daily confirmed COVID-19 cases per million people, compared to Italy’s 32, Sweden’s 11, Germany’s four, and the United Kingdom’s four. A month later, on April 13, Denmark had 40 new daily confirmed COVID-19 cases per million people, with Italy at 65, Sweden at 51, Germany at 47, and the United Kingdom at 66 (Dong, Du, and Gardner 2020; Our World in Data 2022). While Denmark was not an extreme case compared to other European countries at the beginning of the COVID-19 crisis, its various hospital departments, the frontline of the Danish public healthcare sector, were significantly affected, albeit to varying degrees, as explained below.

On 13 March 2020, the Danish Health Authority sent out overall guidelines on how the Danish healthcare sector was supposed to reduce activity at the hospital departments to accommodate the expected increase in the number of COVID-19 patients who would need hospital treatment (Danish Health Authority 2020). The overall message was that certain patients were not to be treated during the COVID-19 pandemic, including those who did not need treatment for acute or life-threatening conditions or with conditions for which delayed treatment would not cause increased risk of loss of mobility and where it was professionally justifiable to postpone treatment (Danish Health Authority 2020, 2). This was intended to free personnel and material, making it possible to undertake tasks related to patients with COVID-19 while also limiting the risk of the virus spreading in hospitals. Furthermore, some guidelines on how to prioritize a reduction in ambulant and surgical activity were included. However, while stressing the expectation that hospitals would adhere to the guidelines, it was also made clear that the hospitals were still responsible for eventually treating the patients they chose to postpone (Danish Health Authority 2020).

The overall guidelines were relatively broad, and frontline managers at the different hospital departments therefore had to respond to both the crisis of COVID-19 and the decisions made at the political level, requiring them to make strategic decisions about the direction, goals, and priorities of their departments during the crisis, as well as operational decisions in terms of solving new types of problems in relation to how they carried out day-to-day tasks at the departments.

Two issues are central to the empirical design of this study. The first is that the COVID-19 crisis affected hospital departments to varying degrees because different types of departments differ in, for instance, whether or not they deliver services to patients where it is professionally justifiable to postpone treatment, which creates variation in how much the departments were impacted by the COVID-19 crisis. This is exemplified in news articles in the Journal of the Danish Medical Association published in April and May 2020.

In one article from May 2020, clinical directors managing intensive and acute care departments at the beginning of the crisis highlighted the need for rapid decisions and changes. For example, they had to prepare three times the capacity for intensive care patients within 8–10 days (Journal of the Danish Medical Association (Ugeskrift for Læger) 2020a, 3). This exemplifies operational decisions in a department that was highly affected by the crisis as the decisions are concerned with the solving of problems related to how the department can handle day-to-day tasks during the crisis. Furthermore, decisions had to be made under conditions of uncertainty, exemplified with one of the clinical directors stating that ‘[e]specially in the first period there was a lot of uncertainty, we did not really know the job’ (Journal of the Danish Medical Association (Ugeskrift for Læger) 2020a, 6 – the author’s translation).

In another article from April 2020, the clinical director of an orthopaedic department mentioned that the department was minimally affected by the crisis. This allowed some physicians to take vacations, while others could focus on research and other tasks (Journal of the Danish Medical Association (Ugeskrift for Læger) 2020b). This example illustrates a strategic decision in a department that was less affected by the crisis as the decision is concerned with the overall priorities of the department during the COVID-19 crisis. Taken together, the above examples illustrate the opposite ends of the spectrum of how affected the different departments were by the COVID-19 crisis, but they also illustrate different types of decisions.

Naturally, we could also imagine strategic decisions in departments that were highly affected by the crisis and operational decisions in departments that were less affected. As an example of a strategic decision in a department highly affected by the crisis, we could imagine an intensive and acute care department where the clinical director had to decide on how the department were to prioritize between COVID-19 patients and regular patients. As an example of operational decisions in departments less affected by the crisis, we could imagine an orthopaedic department with a reduced patient intake due to COVID-19, where the clinical director chooses to spend time evaluating and changing the department’s work routines, such as how bed rounds are conducted.

The second issue that is central to the design is that 13 March 2020 serves as a cut-off point where the departments were hit extensively in terms of the strategic and operational decisions they had to make in response to the crisis and the decisions made at the political level. The case thereby seems suitable for testing whether, at the frontline of public organizations, the degree to which an organization is impacted by a crisis is related to public frontline managers’ involvement of frontline professionals in strategic and operational decisions.

Data

The aim of the data collection was to reach as many different Danish hospital departments, and thus public frontline managers, as possible. This article focuses on clinical directors, who are physicians and formal frontline managers of all physicians at a hospital department. Hospital departments are the frontline of the Danish public healthcare sector because this is where services are delivered to patients, and the degree of involvement among physicians at the department thus comes down to the degree to which the clinical directors involve the physicians in decision-making. Clinical directors have formal personnel responsibility for the physicians and have the ultimate responsibility for making strategic and operational managerial decisions at the hospital departments, thereby making them public frontline managers. A survey sent to all clinical directors at danish public hospital departments in 2020 showed that clinical directors on average manage 35 physicians, but there is a large degree of variation between departments (Pedersen, Thomsen, and Elbæk 2020).

Research has shown that managers tend to overestimate how much they behave in a certain way compared to how much their employees perceive that behaviour (see, e.g. Jacobsen and Andersen 2015). Following this logic, managers might overrate how much they involve frontline professionals in decision-making compared to how the frontline professionals perceive it, and this article therefore relies on data from a questionnaire distributed in February/March 2021 to the physicians who the clinical directors manage.

The clinical directors at Danish public hospital departments manage two types of physicians, namely chief physicians and junior physicians. The chief physicians in focus here are physicians who hold the title of chief physician but who are not formal managers with personnel responsibilities. However, chief physicians are responsible for the treatment of patients at the hospital departments. The junior physicians include physicians who have received their postgraduate degree in medicine but are either still in the process of educating themselves within a medical speciality or who have finished their education within a medical speciality but who are not employed as a chief physician. Importantly, both types of physicians are employed at public hospital departments as frontline professionals and are managed by the clinical director. Therefore, both groups were included in the sample to make sure that the results mirror the experience of both types of physicians employed by clinical directors.

To reach junior physicians and chief physicians working for as many different clinical directors as possible, the branch of the Danish Medical Association that represents the junior physicians and the branch that represents the chief physicians were contacted and asked to distribute a survey to some of their members. The branch that represents the junior physicians agreed to distribute the survey to those of their members who were listed as a union representative or as an alternate at the time of distribution (February/March 2021), meaning that it was sent to 695 junior physicians. The branch that represents the chief physicians agreed to distribute the survey to those of their members who were listed as a union representative or as an alternate at the time of the survey’s distribution and to those who were part of their member panel, meaning that it was sent to 332 chief physicians. Each branch received a report with aggregated descriptive statistics after the data collection was complete. In these reports, it was of course not possible to identify individual respondents or departments.

For the analyses conducted in this article, the focus is on those individuals who worked at the same department from January 2020 to July 2020 and who also answered the variables used in the main analyses. Missing values, duplicates, and respondents who were not working at an identifiable hospital or not working at a hospital department and chief physicians who stated that they were formal managers were removed¹.

The sample used in the main analyses consists of 169 individuals clustered within 143 hospital departments. The respondents were on average 49 years old, 52% women, and 56% chief physicians. Around 40% of the respondents were either a union representative or an alternate at their department in the period from January 2020 to July 2020. This implies that they might have a better idea of how the department functions and, thus, be able to answer how the clinical director in general involved frontline professionals across the department, which is what they are asked to do in the questionnaire (see Andersen and Pallesen 2008 for another example of the use of union representatives). Due to the way the data were collected, it is unfortunately not possible to make a clear investigation of the representativeness of the sample or analyse non-response bias. However, the distribution of respondents across the five Danish regions is relatively close to the distribution of clinical directors in 2019².

Dependent variables

In the survey, the respondents were asked about different periods in the first half of 2020 and to answer a set of questions aimed at capturing the level of employee involvement in strategic and operational decisions by the clinical director at the department in these periods. Previous studies investigating the effects of crises or extreme events have also asked respondents to recall such events, in periods covering up to two years (e.g. Zhang, Welch, and Miao 2018). This article asks respondents to recall periods from one year back and, furthermore, uses vignettes to make it easier for the respondents to recall the periods. It uses responses related to the period before COVID-19 – January 6 to March 16 (Vignette 1) – and the period during which the departments were asked to reduce hospital activity because of COVID-19 – March 16 to April 13 (Vignette 2). The vignettes can be seen in Table 1 below. The first vignette focuses on the period before COVID-19, and it is thus supposed to make the respondent recall the period before the departments were affected by the crisis. The second vignette describes the period right after the departments were affected by the COVID-19 crisis. By providing the respondents with descriptions of two clearly distinguishable periods, the use of the vignettes should help alleviate some potential concerns about recall bias.

Table 1. Vignettes.

Vignette 1 – period before COVID-19

When you answer the following questions, we will ask you to think back to the first 2.5 months of 2020, which was the period before the healthcare system was first told to limit activity due to the COVID-19 situation. To be clear, that was before the more radical changes began; that is, week 2 to week 12 (January 6 to March 16). To what extent was the group of junior physicians/chief physicians in the department where you were employed in the period January 6 to March 16, 2020, directly involved in the following tasks by the clinical director, during the same period? 1. The prioritization of the primary purposes of the department’s work 2. The setting of specific goals for the department 3. The planning of which tasks you as a department can handle 4. The organization of the work processes in the department 5. The identification of problems related to how you perform your tasks in the department 6. The identification of solutions to problems related to how you perform your tasks in the department 7. The implementation of solutions to problems related to how you perform your tasks in the department

Vignette 2 – period during COVID-19

When answering the following questions, we will ask you to think back to the transition period immediately after the healthcare system was first told to limit activity due to COVID-19, with the aim of being able to release staff and equipment due to the expectation of very great pressure on the healthcare system; that is, weeks 12 to 16 (March 16 to April 13, 2020). To what extent was the group of junior physicians/chief physicians in the department where you were employed in the period March 16 to April 13, 2020, directly involved in the following tasks by the clinical director, during the same period? 1. The prioritization of the primary purposes of the department’s work 2. The setting of specific goals for the department 3. The planning of which tasks you as a department can handle 4. The organization of the work processes in the department 5. The identification of problems related to how you perform your tasks in the department 6. The identification of solutions to problems related to how you perform your tasks in the department 7. The implementation of solutions to problems related to how you perform your tasks in the department

Note: The bold text indicates how the vignette differed between the surveys to junior physicians and chief physicians. All questions were measured on a 1–5 Likert scale with the possible answers ‘not at all’, ‘to a lesser degree’, ‘to some degree’, ‘to a high degree’, and ‘to a very high degree’.

To further alleviate concerns regarding potential recall bias, the respondents were asked at the end of the questionnaire how it had been for them to recall the periods presented to them in the vignettes. 161 respondents answered this question. Only 21.74% thought that it was hard or very hard to recall these periods, while 50.31% thought it was easy or very easy and 27.95% thought that it was neither hard nor easy.

In Table 1, the questions posed to the respondents after each of the vignettes aimed at capturing employee involvement in strategic and operational decisions can also be seen³. As mentioned, strategic decisions are here managerial decisions concerned with overall goals, direction, and priorities (e.g. Bryson 2004; Maynard-Moody and McClintock 1987; Poister et al. 2013). Questions 1 to 4 in Table 1 were formulated to capture involvement in strategic decision-making at the hospital departments. They were measured on a 1–5 Likert scale, with the possible answers ‘not at all’, ‘to a lesser degree’, ‘to some degree’, ‘to a high degree’, and ‘to a very high degree’.

As mentioned, operational decisions are here viewed as managerial decisions concerned with how to solve problems related to how the organization functions (e.g. Cyert and March 1992; Jung 2018; March and Simon 1993; Simon 1997). Problem solving can be seen as a three-step process: 1) identifying problems that need attention, 2) searching for solutions to these problems, and 3) implementing the most appropriate solutions (Cyert and March 1992; Holm 2018, 305). Questions 5 to 7 in Table 1 were formulated to capture involvement in such operational decision-making at the hospital departments and measured on the same Likert scale as Questions 1–4.

An explorative factor analysis with oblique Oblimin rotation was conducted. Oblique rotation was used because the two types of decision-making are related to some extent, in the sense that operational decision-making might focus on problems and decisions regarding how the organization deals with its day-to-day tasks that can be linked to the organization’s ability to pursue its overall goals and strategies (e.g. Cyert and March 1992; Holm 2018).

As can be seen in Table A1 in the appendix in the supplemental material, it is not possible to identify two factors when only including answers in response to Vignette 1, the period before COVID-19, while two clear factors emerge when the answers to Vignette 2, the period during COVID-19, are used in the factor analysis. Furthermore, the two factors also emerge when we use the pooled responses to the seven items. This might indicate that before the crisis, employee involvement in the two types of decisions was strongly correlated but that the crisis made the differences clear. This is interesting as it points towards some differences in involvement in the two types of decisions dependent on the crisis.

Considering the theoretical distinction between the two types of decisions and the fact that two clear dimensions materialize when we use the responses to the items tied to Vignette 2 and when we use the pooled responses, two indexes were created. One is for involvement in strategic decisions using responses to Questions 1–3, and one is for involvement in operational decisions using responses to Questions 5–7. Responses to Question 4 are left out of the indexes, even though the factor loading is relatively high on the second factor. The reason for leaving it out is that it was originally expected to be part of the index related to strategic decisions but loads higher on the factor with the questions tied to operational decisions. In hindsight, that makes sense, considering that it is more focused on an operational than a strategic decision-making process. Including Question 4 in the index on operational decisions does not affect the results.

The index for involvement in operational decisions has a standard deviation of 1.17, mean of 2.91, and Cronbach’s alpha of 0.97 in response to Vignette 1, and a standard deviation of 1.23, mean of 2.87, and Cronbach’s alpha of 0.97 in response to Vignette 2. The index for involvement in strategic decisions has a standard deviation of 1.20, mean of 2.43, and Cronbach’s alpha of 0.94 in response to Vignette 1, and a standard deviation of 1.21, mean of 2.49, and Cronbach’s alpha of 0.96 in response to Vignette 2.

Independent variable

To capture the independent variable – how affected the departments were by the COVID-19 crisis – answers to a question where respondents were asked to indicate how involved their department had been in handling the COVID-19 crisis are used. The question asks the respondents to state, on a scale from 0 to 100, how involved their department had been in handling COVID-19, with 0 being no involvement and 100 being the most involved department. This question was posed before the questions about involvement in decision-making to ensure that the vignettes related to the dependent variables did not influence how much they felt their department had been affected. The mean is 55.10, and the standard deviation 32.89. Thus, it seems that there is indeed a large degree of variation in how affected the departments were by the COVID-19 crisis, which is central to the design.

The answers to this question thus correspond to the period in focus in Vignette 2, which is the period in which the departments were asked to reduce hospital activity because of COVID-19 (see Table 1). In terms of the period before COVID-19, which matches Vignette 1 (see Table 1), all responses were coded 0 on the independent variable. It is assumed that none of the departments were extensively affected by COVID-19 at this point in time, since it was not until March 13 that they were asked to make changes in response to the crisis. This results in a panel data structure, meaning that there is data for n different entities observed at T different time periods (Stock and Watson 2015, 397). In this case, the panel dataset consists of data from the 169 physicians on the dependent and independent variables from the period before COVID-19 and from the period after the hospital departments had to adapt to the crisis of COVID-19, resulting in a total of 338 observations.

Statistical model and controls

The structure of the data makes it possible to investigate the hypotheses using respondent-level fixed effects regression in all the models to control for all unobserved and observed individual and departmental time-invariant confounding factors (Stock and Watson 2015, 403‒405). This is an important strength of the study considering that the departments differ in, for instance, size and the work they do, which might directly affect both how much they were affected by the COVID-19 crisis and how much the physicians were involved in decision-making. Importantly, this approach should also help alleviate some of the concerns regarding recall bias, in the sense that the fixed effects control for any individual-level, time-invariant factors within the respondent associated with recall bias.

Using respondent-level fixed effects also reduces the potential common method bias problem, which is caused by the fact that both the independent and dependent variables are measured in the same survey. Common method bias can be seen as a confounding variable that systematically influences both the independent and dependent variables measured in the same survey, which might either inflate or deflate the relationship between the variables (Jakobsen and Jensen 2015, 5). Using employee-level fixed effects controls for all this time-invariant bias as it controls for all unobserved and observed individual time-invariant confounding factors. The fear of time-variant common method bias due to, for instance, transient mood is less of a concern because the respondents were asked the questions at the same point in time. However, it is of course still important to acknowledge, first, that the method used here does not guarantee that there is not some bias present in the way that the independent variable is measured and, second, that it would have strengthened the results if it would have been possible to find an objective measure of crisis affectedness.

Because the respondents are presented with a vignette explaining the period they are supposed to think back to, a vignette/period dummy is also included to control for the potential effect from the vignette/period on all respondents, thus utilizing the variation from their experience of their department being affected by the crisis.

The models were estimated with cluster robust standard errors at the hospital level because the unit of interest is clinical directors who are nested within 26 hospitals, and these hospitals might systematically have been affected differently because of their size or location or might have chosen different ways of dealing with the crisis, which means that the clinical directors within hospitals might be more alike. However, considering that the 169 respondents are nested within 143 departments, a robustness test was also conducted where the standard errors were clustered at the hospital department level.

Results

Table 2 shows the models for the relationships between perceived crisis and perceived involvement in operational and strategic decisions. For ease of interpretation, all variables have been rescaled to range between 0 and 1. Beginning with the relationship between crisis and involvement in operational decisions, the expectation is, that there is a positive linear relationship between the extent to which a public organization is affected by a crisis, and how much the frontline managers will involve frontline professionals in operational decisions (H1). As can be seen in Model 1, Table 2, there is a positive linear and significant relationship between how affected the organization is by the crisis and the clinical directors’ involvement of frontline professionals in operational decisions, which supports Hypothesis 1. More specifically, the results show that going from 0 to 1 in terms of how affected the departments were by the COVID-19 crisis – that is, the full variation going from not being affected to being fully affected – corresponds to a nine-percentage point increase in involvement of the frontline professionals in operational decisions during the COVID-19 crisis compared to before the COVID-19 crisis.

Table 2. Respondent-level linear fixed effects regression estimates for the involvement of frontline professionals in operational decisions and in strategic decisions.

	Model 1 – Involvement in operational decisions	Model 2 – Involvement in strategic decisions
Affected by crisis	0.09** (0.030)	0.06 (0.043)
Period/vignette dummy	−0.06* (0.025)	−0.02 (0.029)
Constant	0.478*** (0.008)	0.357*** (0.006)
Observations	338	338
Respondents	169	169
R^2 – within	0.024	0.017
R^2 – between	0.028	0.021
R^2 – overall	0.022	0.014

Note: Hospital clustered standard errors in parentheses (26). Respondent fixed effects regression with period/vignette dummy. ⁺p < 0.1, *p <0.05, **p <0.01, ***p <0.001.

Turning to the second hypothesis (H2), the expectation is, that there is a positive linear relationship between the extent to which a public organization is affected by a crisis, and how much the frontline managers will involve frontline professionals in strategic decisions. As can be seen in Model 2, Table 2, there is no significant linear relationship between crisis and the clinical directors’ involvement of frontline professionals in strategic decisions, and therefore no support for the hypothesis. Thus, the results of this analysis show that the clinical directors involve frontline professionals more in operational decisions in response to crisis, but it does not seem that there is a positive linear relationship between crisis and the degree to which they involve frontline professionals when it comes to strategic decisions in response to crisis.

Several robustness tests were conducted, such as clustering standard errors at the department level, dropping respondents who indicated that it was hard for them to recall the two time periods, and dropping all respondents that produced duplicates. These tests did not affect the results significantly and are explained in detail in the appendix in the supplementary material, where the specific results of the robustness tests can also be found in Tables A2-A5.

Exploratory analysis

To further explore the relationships between how affected the departments were by the crisis and the clinical directors’ involvement of frontline professionals in operational and strategic decisions, exploratory analyses were made in which both the relationship between crisis affectedness and involvement of frontline professionals in operational decisions, and the relationship between crisis affectedness and involvement of frontline professionals in strategic decisions, were estimated with a squared term. Including a squared term makes it possible to test whether there might exist non-linear relationships between crisis affectedness and the clinical directors’ involvement of frontline professionals in strategic and operational decisions. Thus, when it comes to the relationship between crisis affectedness and involvement of frontline professionals in operational decisions, including the squared term makes it possible to test the robustness of the linear model presented in Table 2, Model 1. When it comes to the relationship between crisis affectedness and involvement of frontline professionals in strategic decisions, including the squared term makes it possible to explore whether there might exist a non-linear relationship.

The results of the exploratory analyses can be seen in Table A6 in the appendix in the supplementary material. The results in Model 1 in Table A6 reveal that there is no indication of a non-linear relationship between crisis affectedness and involvement of frontline professionals in operational decisions, supporting the linear model presented in Model 1, Table 2. However, the result in Model 2 in Table A6 indicates a convex relationship between how affected the organization is by the crisis and involvement of the frontline professionals in strategic decisions. To ease the interpretation of the convex relationship between how affected the organization is by the crisis and involvement of the frontline professionals in strategic decisions, the relationship is depicted in Figure 1.

Figure 1. Relationship between how affected the organization is by crisis and the involvement of frontline professionals in strategic decisions.

Note: The figure is based on Model 2 in Table A6.

More specifically, the relationship indicates that there is a negative relationship between how much the departments were affected by the COVID-19 crisis and involvement of the frontline professionals up until a turning point of around 0.5 of crisis affectedness. From the turning point, there is indications of a positive relationship between how much the departments were affected by the COVID-19 crisis and involvement of the frontline professionals in strategic decisions. However, the model also indicates that there is no difference in the level of involvement of frontline professionals in strategic decisions when comparing high degrees and low degrees of crisis affectedness, but it appears that at moderate degrees of crisis affectedness, the clinical directors may involve the frontline professionals less in strategic decisions compared to both high and low degrees of crisis affectedness.

In sum, this indicates that the relationship between how affected the departments were by the crisis and the clinical directors’ involvement of frontline professionals in decision-making was different depending on whether it was a strategic or an operational decision and depending on the degree of crisis affectedness. To test the robustness of the exploratory results presented in Table A6, the same robustness test was applied as those used to test the results in the analysis with linear fixed effects, and these are explained in detail in the appendix in the supplementary material. The tests did not affect the results significantly, and the results of the robustness tests can be seen in the appendix in the supplementary material, Tables A7-A10.

Concluding discussion

Findings and contributions

First, this article finds support for the expectation of a positive linear relationship between the extent to which a public organization is affected by a crisis and the degree to which frontline managers, here clinical directors, involve frontline professionals in operational decisions. That is, in situations without crisis, the frontline managers will have some decision-making process in place for making operational decisions, in which frontline professionals can be more or less involved, but as the organization is affected by crisis, the frontline managers will involve the frontline professionals more in operational decisions.

The finding that the public frontline managers in this study seem to involve frontline professionals more in operational decisions in response to the COVID-19 crisis has implications for the public management literature concerned with the role of public frontline managers during crises. The result seems to suggest that frontline managers in public organizations might trade off speed for more knowledge and information as the uncertainty introduced by a crisis calls for knowledge and information instead of speed, when it comes to operational decisions. In that way, the finding is aligned with the notion that public frontline managers might benefit from greater involvement of their employees, such as frontline professionals, in managerial decision-making during crises (Ansell, Sørensen, and Torfing 2021).

Second, the article did not find support for the expectation of a positive linear relationship between the extent to which a public organization is affected by a crisis and the degree to which frontline managers involve frontline professionals in strategic decisions. Instead, the article found indications of a convex relationship, which indicates that at medium degrees of crisis affectedness, managers might involve frontline professionals less in strategic decisions, compared to high and low degrees of crisis affectedness, but that there is no difference in the level of involvement in strategic decisions when comparing low and high degrees of crisis affectedness.

The fact that frontline managers might not make all decisions in the same way when it comes to involvement of frontline professionals in response to crisis is also apparent in interviews with clinical directors in news articles in the Journal of the Danish Medical Association (Ugeskrift for Læger) published between March and May 2020. One clinical director mentioned that some department decisions were open to discussion at the beginning of the crisis, while others were not (Journal of the Danish Medical Association (Ugeskrift for Læger) 2020a, 5). Similarly, another clinical director noted that uncertainty during the crisis necessitated involving more people in decisions, but time constraints limited discussion (Journal of the Danish Medical Association (Ugeskrift for Læger) 2020c, 4).

The results call for a revision of the theoretical argument of how frontline managers make strategic decisions in response to crisis. One possible theoretical explanation for the results regarding the impact of the crisis on involvement in strategic decisions could be its tendency to shift focus away from strategic decisions. Strategic decisions typically have a long-term focus, involving priorities and overall goals (e.g. Poister et al. 2013), whereas operational decisions deal with day-to-day issues. Therefore, the decreased involvement of frontline professionals in strategic decisions during a crisis compared to before a crisis might stem from the reduced emphasis on long-term planning during such times. This could be due to the crisis demanding intense focus on operational decisions or because higher organizational levels handle strategic decisions during crises, leaving fewer to be made at the frontline. However, this does not explain the U-shaped relationship depicted in Figure 1, which indicates that there might still be a relatively higher need to involve frontline professionals in strategic decisions when the organization is affected by a crisis to a high degree, compared to a medium degree of affectedness.

Thus, we might alternatively theorize that when it comes to strategic decisions during crises, the need for quick decisions regarding the direction, goals, and priorities outweighs the need for information when the organization is only somewhat affected by the crisis. However, at some point, the uncertainty due to an organization’s level of affectedness by the crisis might be so great that involving frontline professionals less in decision-making is no longer a useful strategy for the manager. Instead, the frontline managers might require information from frontline professionals to make overall priorities, set goals, and plan the tasks of the organization, to the same extent as when they are not affected by a crisis. This could explain the u-shaped relationship between crisis and involvement in strategic decisions at the frontline of public organizations.

In sum, this study contributes to the public management literature focused on the role of public frontline managers during crisis (e.g. Ansell, Sørensen, and Torfing 2021; Bertelsen, Lindholst, and Hansen 2022; Lund and Andersen 2022; Phillips, Roehrich, and Kapletia 2021; Van der Wal 2020) by showing that whether frontline managers involve frontline professionals more in managerial decision-making in response to crisis, depends on the type of managerial decision and the degree of crisis affectedness.

Limitations and future research

There are some limitations to this study that require further discussion which also points to avenues for future research. The empirical case is clinical directors’ involvement of physicians at Danish public hospital departments during the beginning of the COVID-19 crisis. Compared to the level of COVID-19 affectedness of other European countries in the beginning of 2020, Denmark was, as mentioned earlier, not an extreme case, and in that regard, the findings could also travel to other countries that were not affected to an extreme degree. In terms of the focus on clinical directors and physicians, the theoretical arguments are formulated to be applicable to all settings where public frontline managers must make decisions in response to crises and where the frontline employees are frontline professionals who deliver public services to users. Even so, COVID-19 is a rather special case, physicians are one of the most professionalized professions, and healthcare delivery is very complex (e.g. Ackroyd, Kirkpatrick, and Walker 2007; Andersen and Pedersen 2012) which potentially limits the generalizability of the results. Therefore, future research could and should try to replicate the findings of this study among other types of public frontline managers, professionals, and public organizations in relation to crises other than COVID-19, potentially in different countries.

Furthermore, because of necessary limitations in the way the data were collected, it was not possible to make a clear investigation of the representativeness of the sample or analyse non-response bias. This could potentially introduce some bias in the results of the study that readers should be aware of, and future studies should aim to make sure that their data collection allows for an explicit formal test of non-response bias and representativeness.

Additionally, the measure of how affected the hospital departments were by the COVID-19 crisis is a one-item question posed to the respondents about how involved the departments were in handling the crisis. Such a one-item measure might not capture all nuances in how affected departments were by the COVID-19 crisis. Future studies interested in how crises affect involvement of frontline professionals in decision-making, could seek to use more nuanced measures to capture how affected the organizations are by crises – for instance, by looking at magnitude, frequency, and scope of crises (see Zhang, Welch, and Miao 2018). Relatedly, it is important to note that there is a limitation in using a subjective measure of how affected the departments were by the crisis of COVID-19, compared to an objective measure, in terms of being able to claim causality even though the crisis was external to the organizations.

It is also important to explicitly discuss the limitations of the exploratory analysis and its results. First, given the exploratory nature of the analysis, it is important that future studies seek to reproduce these results to improve our confidence in them. In that way, the results of the exploratory analysis should be seen as a starting point for future research aimed at investigating whether the relationship between crisis and involvement of frontline professionals could be non-linear for some types of decisions, as suggested by the findings of this article. Second, these future studies should seek to obtain a larger sample size to increase statistical power, increasing the confidence in any non-linear relationships.

Besides the avenues for future research focused on overcoming the limitations of this study, the findings of the study also call for new research to further improve our understanding of decision-making by public frontline managers during crises. Based on the exploratory analysis, this study has offered two potential theoretical explanations as to why we see a u-shaped relationship between how affected an organization is by crisis and involvement of frontline professionals in strategic decisions. Future studies should seek to test these two explanations, and one way of doing this would be through in-depth interviews with public frontline managers, aimed at disentangling the mechanisms at play when they decide how to make strategic decisions in response to crises.

A qualitative study could also further our understanding of the theoretical mechanisms that drive frontline managers to involve frontline professionals in operational decisions in response to crisis. This article draws on insights from the literature on employee involvement in decision-making to theorize that this is due to the need for more information, knowledge, and processing capabilities because of the increased uncertainty in decision-making introduced by the crisis. In-depth interviews could substantiate and maybe add to these mechanisms in the context of crisis and thereby build on and further nuance the findings of this study.

Finally, the findings call for more scholarly and practical insights into the implications of varying levels of involvement of public frontline professionals in different types of decisions during crises. Employee involvement has shown potential benefits in situations without crisis, but outcomes can vary based on contextual and situational factors (e.g. Grissom 2012; Wagner and Gooding 1987). Understanding how involving frontline professionals more in operational decisions and less in strategic decisions affects a public organization’s ability to cope with crisis is essential. After all, we need to know whether a strategy where public frontline managers adapt the decision-making process during crises – prioritizing the involvement of frontline professionals for operational decisions and showing a complex response for strategic decisions – can be deemed good decision-making in response to crisis.

Acknowledgments

The author would like to thank Christian Bøtcher Jacobsen, Peter Bjerre Mortensen, Eva Knies, the public leadership section at Department of Political Science, Aarhus University, and the participants at the EGPA Permanent Study Group III, Public Personnel Policies, 2021 for helpful comments and suggestions on earlier drafts of this article.

Disclosure statement

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

Supplementary material

Supplemental data for this article can be accessed at https://doi.org/10.1080/14719037.2024.2371957

Notes

1. 24 initial observations were dropped because they were duplicates (that is, some respondents answered the survey twice, and one of these entries was dropped – see robustness tests in appendix in the supplementary material for elaboration). In the analysis, a robustness test was also conducted in which both entries from those who answered the survey twice were dropped to make sure that there was not a systematic bias from those who answered the survey twice.

2. The distribution in this sample is 38% from the Capital Region of Denmark, 22% from the Central Denmark Region, 8% from the North Denmark Region, 15% from the Region Zealand, and 17% from the Region of Southern Denmark. The distribution of clinical directors in 2019 across the five regions was 29% from the Capital Region of Denmark, 20% from the Central Denmark Region, 14% from the North Denmark Region, 17% from the Region Zealand, and 20% from the Region of Southern Denmark (Pedersen, Thomsen, and Elbæk 2020).

3. The survey was pilot tested through an interview with a clinical director and an interview with a chief physician. The interviews were conducted to ensure that the periods presented to the respondents in the vignettes were meaningful and that the questions that the physicians were asked in relation to whether the clinical director would involve them in decisions made sense, from the points of view of both the clinical director and the physicians. Overall, the questions, decisions, and periods made sense to the interviewees, and the primary change made in response to the interviews was to make it clearer that the respondents had to think about the clinical director of their own department.