It ’ s about time! Temporal dynamics and longitudinal research designs in public administration

Many of the fundamental research questions in public administration relate to individual-or organization-level temporal dynamics, including the impact of public sector reforms, (in)stability of public policies and organizations, development of public service motivation, or the workplace socialization of public employees. However, theoretical, methodological, and empirical public administration scholar-ship continues to take time and temporal dynamics insufficiently seriously. This constitutes a major shortcoming within the profession and implies that we are yet to unlock the transformative potential of longitudinal research. Building on the recent development of novel research infrastructures that can support the study of temporal dynamics of — and within — public organizations, this Symposium pushes for a “ longitudinal turn ” in the study of public administration. We maintain that more concerted efforts to apply a temporal lens to our research endeavors are critical to theorize, empirically assess, and understand public administrations as well as the bureaucrats employed within them.


INTRODUCTION
Two decades ago, Baumgartner and Jones (2002: p. 6) lamented that "one of the truly great failings of the policy sciences has been the inability to produce reliable longitudinal studies".Contemporaneously, Gill and Meier (2000: p. 157) argued that public administration research had "fallen notably behind research in related fields in terms of methodological sophistication" particularly with respect to the use of time-series analyses.Twenty years later, descriptive case studies and cross-sectional research designs continue to dominate the study of public administration (Pandey, 2017;Ritz et al., 2016;Stritch, 2017).While it is not uncommon for research papers to conclude with a call for applying a longitudinal perspective to the phenomenon under analysis, this rarely occurs in practice.Moreover, longitudinal research designs and methodological approaches are still not uniformly established within the canon of public administration research methods.The Handbook of Research Methods in Public Administration, Management and Policy (Vigoda-Gadot & Vishdi, 2020) dedicates only one chapter to one specific longitudinal research approach (i.e., difference-indifferences models), whereas Van Thiel's (2022) Research Methods in Public Administration and Public Management engages in a broader discussion of the utility of cohort, panel, and trend studies. 1  This observation is particularly disconcerting since many fundamental public administration research questions relate to individual-or organization-level temporal dynamics, or involve causality as an issue of temporal ordering.One can think, for example, about the impact of public sector reforms on organizational outputs and performance (Christensen & Laegreid, 2007;Levine, 1978Levine, , 1979;;Wynen et al., 2019), the development of public service motivation or organizational citizenship behavior (Chen et al., 2023;Kim, 2021;Miller-Mor-Attias & Vigoda-Gadot, 2022), the socialization of public sector employees (Moyson et al., 2018;Murdoch et al., 2019), the life cycle of public organizations (MacCarthaigh et al., 2012), or the short-and long-term impact of representative bureaucracies on public policies (Ding et al., 2021;Gershenson et al., 2022).The persistent failure in much public administration research to take time and temporal dynamics seriously thus constitutes a major hindrance to the development of the profession.It "abstracts away from the temporal flow of much of [public] organizational life" (Langley et al., 2013: p. 4), and hampers "investigations into substantive questions of interest to practitioners and academics" (Gill & Meier, 2000: p. 157).
Clearly, there are a number of reasons for the slow progress of public administration scholarship from its static, cross-sectional focus toward a more dynamic, longitudinal approach.Compared to most cross-sectional work, establishing longitudinal datasets generally requires a substantial investment of time and resources.The incentives and demands of modern academic careers, however, do little to reward longer-term approaches to data-gathering and infrastructure development.There are considerable career risks involved for researchers who invest in projects that can take years before useable data (and publications) emerge.Moreover, resources are limited and access to funding has long been a major hindrance for public administration scholars, thus constraining the possibilities to develop necessary research infrastructures (Gill & Meier, 2000).Even when time and resources are not the main obstacle, other challenges remain.At a conceptual level, for instance, what policy is and what defines a public organization may change over time, making any basis for comparison more arduous.Furthermore, developing longitudinal datasets at the individual level faces complications due to attrition (as employees retire or move) and lack of repeated and/or sufficiently frequent access, among others (Stritch, 2017).
Notwithstanding these constraints, recent years have witnessed the creation of several new research infrastructures to support longitudinal perspectives on public policy, public organizations, and bureaucrats.Building on these developments, this Symposium challenges the persistent dominance of the static status-quo.While there have been singular attempts to bring temporal concepts into play (which we review below), we argue that now is the time to engage in more concerted efforts to address this long-neglected aspect of our discipline and award more attention to the transformative potential of longitudinal research.We also wish to inspire colleagues to collaborate on building research infrastructures that can facilitate scientific inquiry into the longitudinal dimensions of public administration(s).Static approaches to the field can obviously offer important insights, but they are inherently unable to model dynamic relationships, account for time dependence, or address time-related factors that may be a causal factor of interest in their own right.Hence, it is critical to apply a temporal lens to research endeavors in order to advance our ability to theorize, empirically assess, and understand public administrations, their evolution and reform.
In the remainder of this introduction, we first define and discuss the key concepts of time, temporal dynamics, and longitudinal approaches.Then, we review extant strands of longitudinal research in public administration, before turning attention to the resources and research infrastructures required for a longitudinal turn in public administration scholarship.We thereby also highlight what we can (and cannot) learn in this direction from neighboring disciplines.Finally, we offer a brief preview of the contributions included in this Symposium.

KEY CONCEPTS: TIME, TEMPORAL DYNAMICS, AND LONGITUDINAL APPROACHES
We start by clarifying the three central concepts of time, temporal dynamics, and longitudinal approaches.While each of these has been used extensively in academic publications, different scholars rely on distinct definitions, interpretations, or understandings.Hence, to avoid confusion, a common understanding is required.
We follow Ancona et al. (2001: p. 513) in defining the concept of time as "a non-spatial continuum in which events occur in apparently irreversible succession from the past through the present to the future."While this continuum is often viewed as involving a linear progression with time moving "forward" (i.e., in one direction), this need not necessarily be the case.Public budgets, for instance, follow a well-defined set of stages every year (Rubin & Bartle, 2023), the development of legislation typically requires passing a pre-set series of stages (Willems & Beyers, 2023), while the "policy cycle" describes the various stages of public policy development (Howlett et al., 2009).Similarly, many phenomenaincluding the evolution of public sector organizationsfollow a developmental pattern across their "life cycle" (MacCarthaigh et al., 2012).Our conceptualization of time allows for such non-linear progressions.Bearing in mind the Greek distinction between Chronos (formal time) and Kairos (events), it also allows for time to be conceptualized as event-based (Ancona et al., 2001).One could, for instance, think of happenings including elections, pandemics, earthquakes, or terror attacks as event-based time in which specific occurrences repeat in a predictable (i.e., regular) or unpredictable (irregular) pattern. 2  When we think about time as defined above, we can start to develop a "temporal" understanding of public administrations, their staff, and their actions (Ancona et al., 2001;Fleischer, 2013;Goetz, 2014;Goetz & Meyer-Sahling, 2009).For instance, mapping activities and actions onto the time continuum naturally brings up temporal concepts such as the scheduling of events at certain points in time (e.g., during agenda-setting), the allocation of time to activities (e.g., short versus long deadlines), as well as activities' frequency, rate of incidence, duration, and so on. 3When we allow for change to take place over time, even more temporal concepts come into playsuch as the rate of change (e.g., rapid or gradual), its timing (e.g., immediate or delayed), and the potential for (a) synchronicity (i.e., multiple events occurring at the same time). 4With such concepts in hand, we can also dig deeper into the temporal relationships involved-such as between an event occurring at a particular point in time and subsequent behavioral outcomes, including bureaucratic representation, interpersonal trust, or organizational legitimacy (Methot et al., 2017;Murdoch et al., 2022).
This discussion brings us to our second concept, namely, temporal dynamics.We define temporal dynamics as the nature of the development in the value of a given variable over time, that is, the specification of its time-path.This goes beyond the concept of change, which "merely" refers to the process of value A turning into value B over time without specifying the characteristics of this development.Although documenting change is often of relevance in and of itself, a focus on temporal dynamics requires researchers to give due attention to when and how change occurs (Ancona et al., 2001;Methot et al., 2017;Murdoch et al., 2022).First, one should account for the rate of change, which can be fast, slow, linear, exponential, or stepwise.Second, one should establish the magnitude of change observed at different points in time, which may include distinguishing between the short-and long-term impacts of an event or intervention as well as the assessment of potential anticipation effects.Finally, one should appraise the presence and length of leads (such as pre-trends) and lags (i.e., the amount of time between events and the onset of its observable impact on some outcome of interest).
Finally, taking time seriously from a conceptual and theoretical perspective requires taking it equally seriously from a methodological and empirical perspective.This brings us to the third concept, namely longitudinal research designs.As mentioned, this concept has been stretched to cover a wide range of approaches.Following Menard (1991), we impose three key features.First, the data must cover two or more periods.Second, the data must measure the same concepts at multiple points in time (e.g., public service motivation for individual i at times t and t + x).Third, the central intention of the analysis should be to use these data to compare between and/or among the periods.This by construction excludes studies linking concept A at time t to concept B at time t + x (e.g., Chen et al., 2023;Kim, 2021).Although such designs impose a temporal ordering on the data collection and analysis, they cannot capture changes and temporal dynamics in a given outcome of interest over time.
Note that these three basic criteria do not specify the unit of observation.In a strict interpretation, one might argue that longitudinal research designs require studying the exact same set of cases from one period to the next (whether individual employees, divisions within a ministry, local authorities, or countries).This would rule out, for instance, repeated cross-sectional survey data (Fernandez et al., 2015;Yackee & Yackee, 2021) or socalled "a posteriori" comparisons (de S a e Silva & de Oliveira, 2023; Montero & Baiocchi, 2022) since different samples are studied at different points in time.It would likewise exclude "pseudo-panels," which aggregate individuals within groups to calculate intra-group means for comparison over time (Bertelli et al., 2015;Boon et al., 2023;Fernandez et al., 2015;Jensen et al., 2018;Oberfield, 2014).Pseudo-panels generally do not refer to the exact same set of cases due to potential shifts in each group's composition across survey waves.Although we do not categorically rule pseudo-panels out from our conception of longitudinal research designs, we maintain that two critical conditions should be met to enable valid inferences when using pseudo-panels to study organization-level temporal dynamics.First, the groups must be based on relevant, clearly defined, and stable organizational units (e.g., ministries, departments, or local authorities).Second, the data employed to create the pseudo-panels should be representative at this relevant organizational level-as is the case, for instance, with the stratified random samples of the US Federal Employee Viewpoint Survey (Office of Personnel Management, 2020).

Longitudinal public administration scholarship: An overview
Of course, it would be misleading to suggest that public administration and policy scholars do not engage with issues of time and temporality.In fact, considerable literature studies in public administration and public management look at macrolevel temporal dynamics-including studies on the historical development of public administration and reform traditions (Hood, 1998;Painter & Peters, 2010), the long-standing practice of programevaluation studies (Bingham & Bowen, 1994), and the vast literature looking into policy diffusion across time and space (de S a e Silva & de Oliveira, 2023; Shipan & Volden, 2012).Nonetheless, analyses of changes over time and temporal dynamics at the micro and meso levels-that is, taking individual public bureaucrats and formal public sector organizations, respectively, as the unit of analysis-are considerably less common.
A decade ago, for instance, MacCarthaigh et al. (2012: p. 846) noted that "the academic literature on organizational change in the public sector offers several ideas and perspectives for explaining the birth, survival, and death of public organizations.Nevertheless, there remains considerable room for supporting these ideas with analyses of actual longitudinal continuity and change."Recent work has sought to close this gap, but witnessed a particular flourishing of work on organizational termination (Greasley & Hanretty, 2016;James et al., 2016;Park, 2013;Witteloostuijn et al., 2018; for an important exception, see Peters & Hogwood, 1988).Naturally, organizational termination constitutes only the end point of the organizational life cycle.A narrow focus on this final stage implies that much remains unknown about, for instance, the birth of public organizations at specific points in time or their growth/decline and reform across their life span.This is slowly changing.Lichtmannegger and Bach (2023), for instance, analyze organizational reforms within a single (Austrian) Ministry over the period 1986-2015.In this Symposium, Kleizen and MacCarthaigh (2023) contribute to this development by studying the determinants of intermediary life-cycle events for 634 Irish public sector organizations covering the 1922-2022 period.
Such dynamic research endeavors at the organizational level often stand or fall with access to detailed historical information.The Structure and Organization of Governments project (SOG-PRO; www.sog-pro.eu)clearly illustrates the potential of such data collection efforts.SOG-PRO brought together detailed annual information about structural developments in the central governments of France, Germany, the Netherlands, and the UK between 1980 and 2015.This triggered a flurry of longitudinal research into, for instance, the association between political turnover and ministerial portfolio name changes (Yesilkagit et al., 2022), the scale of diversity and innovation within ministries (Bertels & Schulze-Gabrechten, 2021), or the autonomy of ministers to engage in such reshaping (Kuipers et al., 2021).Fleischer et al. (2023) in this Symposium contribute to this research stream by looking into the political factors affecting such structural developments.
At the individual public employee level, panel data and longitudinal research designs have most frequently been used to study the relationship between employment in the public sector and public service motivation.This literature illustrates that the direction of causality can run both ways.That is, public service motivation may induce self-selection effects to enter public employment (Georgellis et al., 2011;Kjeldsen & Jacobsen, 2013).Yet, the specificities of public sector employment may also affect the development of public service motivation over time (Braender & Andersen, 2013;Georgellis & Tabvuma, 2010;Jensen et al., 2020;Kjeldsen & Jacobsen, 2013).A second branch of individual-level longitudinal research relies on relatively short-term panel datasets from field, survey, or natural experiments to study the impact on public employees of leadership styles (Bro & Jensen, 2020;Jakobsen et al., 2023;Jensen et al., 2019;Nielsen et al., 2019) or major reforms within public organizations (Geys et al., 2020;Geys, Connolly, et al., 2023;Geys, Laegreid, et al., 2023;Murdoch et al., 2019).While well-designed (quasi-)experimental studies allow for strong causal inferences with regard to the presence/absence of change over time, their short time periods naturally come with limitations from the perspective of temporal dynamics.For instance, they generally do not allow assessment of the speed of change, appraising short-versus long-term impacts, or verifying the presence and length of any impact lags.
Finally, and contributing to the theory of representative bureaucracy, scholars of education have recently started using short panel datasets to study whether and when exposure to same-race teachers is "beneficial to underrepresented minority students" (Gershenson et al., 2022: p. 3).While generating useful insights, such studies can usually only observe student outcomes at specific points in time, which is likely to underestimate representative bureaucracy effects for a number of reasons.It not only concentrates on short-term effects, but also misses any residual impact from past same-race teachers for a given student, potential role model effects (from other teachers not teaching the class), policy changes implemented to remove same-race biases, and contagion effects between teachers of different races (or genders). 5

From theory to practice: Resources for longitudinal research
It is clear that applying a time dimension to public administration questions requires the availability of longitudinal data. 6Within political science, the development of such datasets has a long history.As a result, the discipline now has access to a wide range of databases that capture various aspects of political systems and those working and living within them for often very substantial numbers of years and countries.Examples include the WhoGov (Nyrup & Bramwell, 2020; https://politicscentre.nuffield.ox.ac.uk/whogov-dataset) and Who Governs? (Bértoa & Enyedi, 2021; https://whogoverns.eu)databases.They cover detailed information about cabinet members in 177 countries from 1966 to 2016, and the composition of European governments from 1948 to 2020, respectively.Similarly, Skaaning et al. (2015) supply a lexical index of electoral democracy in independent states between 1800 and 2013 (https://doi.org/10.7910/DVN/29106),while the Quality of Government dataset documents government quality and transparency since the mid-1980s (Teorell et al., 2023; https://www.gu.se/en/quality-government/ qog-data).The Correlates of War database brings together data on intra-and interstate conflicts dating back to 1816 (Sarkees & Wayman, 2010; https://correlatesofwar.org).Most countries also regularly field National Election Studies containing detailed questions about inhabitants' political behaviors and attitudes-sometimes containing overlapping panels for subsets of individual respondents in consecutive survey waves (Geys et al., 2021(Geys et al., , 2022)).
Moving to a lower level of aggregation, cohort studies that track the development of subpopulations over time have achieved considerable status in, for instance, political science, sociology, and demography.One early example is Wisconsin Longitudinal Study (WLS), which has been following a random sample of 10,317 high-school graduates (and parts of their families) at irregular intervals since 1957 (https://wls.wisc.edu).This study inspired, for instance, the Michigan Youth-Parent Socialization Panel Study-which has followed 1,669 high-school seniors and their parents between 1965 and 1997-and more recent comparative birth cohort studies (e.g., Growing up in Digital Europe; https://www.eurocohort.eu).Household panel surveys have also emerged, which track a set of households-and the individuals within them-over often considerable lengths of time.Examples include the British Household Panel Study (BHPS), the European Community Household Panel (ECHP), and the German Socio-Economic Panel (SOEP).Recently, the Comparative Panel File project harmonized data from seven countries' household panel surveys (Australia, Germany, Russia, South Korea, Switzerland, the UK, and the USA), thereby capturing "2.7 million observations from 360,000 respondents, covering the period from 1968 to 2019 and up to 40 panel waves per respondent" (Turek et al., 2021: p. 505).Furthermore, in the context of ever-greater creation and retention of administrative data by governments, record linkage studies-which involve linking multiple administrative records for the same individual over time-have become popular means to study individuallevel behavior in a number of disciplines (Bratsberg et al., 2022;Briggs et al., 2021;Fiva et al., 2021;Grytten et al., 2013).
Public administration scholarship has long struggled to establish a similar longitudinal research infrastructure.In recent years, however, numerous initiatives have started to develop.At the organizational level, several datasets capture the evolution of administrative systems and the organizations within them.Seeking to develop common classifications for organizational life cycles in the public sector at national level, MacCarthaigh and Roness (2012) published a special issue on the theme of "Mapping Public Sector Organizations."Contributions included studies of Ireland from 1922 to 2010 (MacCarthaigh, 2012), Norway from 1947 to 2011 (Rolland & Roness, 2012), Estonia from 1990 to 2010 (Sarapuu, 2012), Lithuania from 1990 to 2010 (Nakrošis & Budraitis, 2012), and Hungary from 2002to 2009(Hajnal, 2012).In similar vein, the SOG-PRO project mentioned previously maps and analyzes structural changes in the central governments of France, Germany, the Netherlands, and the UK annually between 1980 and 2015.Task-focused databases have also been developed, such as Jordana et al.'s (2011) database of global regulatory agencies (48 countries and 15 sectors) between 1966 and 2007.
At the individual level, however, longitudinal databases on public employees remain uncommon (beyond the administrative data collected by governments). 7 Admittedly, individual-level surveys of the attitudes, views, and experiences of civil servants are conducted in many countries-often capturing the same concepts from one year to the next over long periods of time.Examples include the American State Administrators Project (1964-2018;Yackee & Yackee, 2021), US Federal Employee Viewpoint Survey (2002-2021;Fernandez et al., 2015), Norwegian Administration Surveys (1976-2016;Christensen et al., 2018), German Political-Administrative Elite survey (2005-2021;Ebinger et al., 2022), and three large-scale surveys among European Commission staff in 2008, 2014, and 2018(Kassim et al., 2013;;Murdoch et al., 2022).Annual or multi-annual surveys among all civil service staff members have also been conducted in, for instance, Australia (Australian Public Service Employee Census), Canada (Canada Public Service Employee Survey), New Zealand (Te Taunaki Public Service Census), and the UK (UK Civil Service People Survey).These surveys are usually designed as repeated cross-sections without unique identifiers for individual respondents.Fortunately, recent methodological contributions have made it possible to extract individual-level panel datasets from some of these datasets for at least a subset of the surveyed respondents (Geys, 2023;Murdoch et al., 2019).
Our survey of these resources in this introduction to the Symposium is not designed to be comprehensive, and there are undoubtedly other (trans)national projects underway (including in other languages) that engage meaningfully with temporal dynamics and longitudinal data.These novel datasets and methodological tools offer exciting new possibilities for researchers to study both public sector organizations and their personnel from a longitudinal perspective.The contributions to this Symposium illustrate how these innovative resources and research infrastructures, as well as a temporal mindset at the conceptual, theoretical, methodological, and empirical levels, can create a fertile breeding ground for a "longitudinal turn" in the study of public administration.

Overview of contributions to the Symposium
In this final section, we briefly introduce the seven contributions included in this Symposium.We structured these articles to move from the macro level (i.e., public authorities), via the meso level (i.e., public organizations), to the micro level (i.e., staff in public organizations).This intends to illustrate both the breadth and depth of the potential afforded by research endeavors applying a temporal lens in public administration scholarship.
Starting at the macro level, Elston et al. (2023) study the implications of recent developments toward increased interjurisdictional cooperation using a panel dataset at the level of English local councils.Exploiting the staggered rollout of cooperation agreements across England over time, they apply a stacked differencein-differences estimator to 11 years of performance data on council-level tax administrations.This provides a unique opportunity to compare how these administrations' collection rates and costs developed before and after entering into a cooperation agreement.The main findings show that collaboration caused a reduction in the tax collection rate and does not lower public expenditures on this public service.They attribute this surprising result to a lack of interdependence in the provision of this particular public service, implying there was no "problem" for interjurisdictional cooperation to "solve."This observation strongly suggests that jumping on the latest reform wave just because "everyone else is doing it" may generate at best limited benefits and could carry important risks for the involved public authorities.
Moving to the meso level, Kleizen and MacCarthaigh (2023) contribute to the growing literature on the causes and consequences of major events in the life cycle of public organizations.In contrast to the majority of previous research conducted in this field, they delve into the determinants of both intermediary and termination lifecycle events linked to shifts in the tasks, policy domain, legal form, and formal structure of public organizations.This is feasible due to an in-depth mapping exercise that allowed constructing timelines of the entire life cycle of over 600 state agencies in the Irish national public administration over the 1922-2021 period.Applying survival analysis to this unique new dataset, a first major finding is that an organizational reform today could contribute to additional reforms in the future.This is consistent with theoretical predictions that public organizations may become "stuck" in persistent reform cycles (Brunsson, 2006;Pollitt, 2007), which public sector managers should take into account before starting an organizational reform program.Additionally, their contribution extends prior work on the determinants of reform by suggesting that factors such as political turnover and legal form can also affect a public organization's likelihood of facing reform.Fleischer et al. (2023) are likewise interested in when and how the organizational structure of public sector organizations changes.In a similar vein to Kleizen and MacCarthaigh (2023), their analysis builds on an in-depth mapping of the timelines of ministerial units in four European democracies over the 1980-2013 period.However, their empirical analysis takes a complementary approach to that of Kleizen and MacCarthaigh (2023) by focusing on key political determinants of the duration of ministerial departments.The results show that cabinets' ideological turnover and extremism as well as the level of ideological polarization in the national parliament significantly affect the probability that ministerial departments are terminated (or, at least, very comprehensively reformed).Crucially, however, their findings also highlight that the exact definition and operationalization of the "structural duration" under investigation can matter for the inferences drawn from the analysis.This highlights that future scholars should pay special care in defining, measuring, and coding organizational changes (or lack thereof) in order to draw valid and robust inferences.Boon et al. (2023) also work at the organizational level, but their analysis employs individual-level data from repeated cross-sectional surveys aggregated up to the level of the public organizations within which these individuals work (in this case, US agencies).The creation of this pseudo-panel (see Bertelli et al., 2015) allows them to take an explicitly dynamic approach to the study of how reputational threats affect the rigidity of public organizations (defined as a work environment that "discourages employee involvement, innovation, creativity, flexibility and empowerment"; Boon et al., 2023: p. 17).Using dynamic panel data models over the 1998-2010 time period, the main results illustrate that negative reputation shocks as well as increased inter-temporal variability in organizational reputations are linked to statistically significantly higher levels of organizational rigidity.This finding reveals the risk of a vicious cycle whereby public sector managers respond to negative environments by increasing the level of formalization in their work environment.Unfortunately, however, these actions may well work to enhance the negative perception of an inflexible, unresponsive, and "fossilized" public sector (Barton, 1979;Moynihan & Ingraham, 2010;Rainey & Steinbauer, 1999).
Moving down to the micro level, Geys (2023) presents a novel methodology to extract individual-level panel datasets from repeated cross-sectional surveys among the same respondent pool (first applied in Murdoch et al., 2019).The approach builds on the fact that most sociodemographic background characteristics of individuals do not change over time (such as birthplace), change at a fixed pace (such as aging), or reflect permanent elements in a person's life history (such as career history).Geys (2023) illustrates when and how specific combinations of such characteristics can be used to reveal the repeated presence of the same respondent(s) across repeated surveys among the same target population.This approach offers two key benefits to researchers and practitioners.On the one hand, the resulting panel datasets provide the opportunity to engage in longitudinal analyses at the individual level, which extends the possible insights that can be gained from the original datasets.On the other hand, the approach offers a tool that can help researchers and practitioners make decisions regarding any re-coding and/or withholding of specific information in order to maintain respondent anonymity while creating Public Release datasets.
Geys, Laegreid et al. ( 2023) apply the methodological approach mentioned above to the Norwegian Administration Surveys fielded in 2006 and 2016 (Christensen et al., 2018) in order to study the impact on Norwegian civil servants of the July 22, 2011 terrorist attack in Oslo.The dataset covers 186 individuals observed before and after this terrorist event and thus captures their perceptions of key aspects of their work environment at both points in time.The main findings from a two-group, twoperiod difference-in-differences analysis illustrate that having been in very close proximity to a terror attack strengthens the internal cohesion of the involved public organizations, while also increasing the attention the civil servants under attack award to the signals of their political leaders.The authors interpret this as evidence that external threats to one's workplace act as a bonding device for the affected employees, but in a public sector setting can also lead politicians to gain more influence over the decision-making of civil servants.The latter can have particularly important implications for democratic governance.
Finally, Yackee (2023) evaluates whether and how ideological extremism in the political environment of public organizations affects the desire to grow and expand among the leaders of these organizations.Her analysis is based on a unique combination of 11 surveys fielded between 1964 and 2018 among all US state agency leaders (N > 10,000 respondents).This allows the empirical analysis to exploit extensive variation across both space and time in US state agencies' ideological environment.A first finding from the analysis is consistent with Niskanen's (1971) view of budget-maximizing bureaucrats and shows that US agency leaders tend to express a strong preference for their agency to expand in terms of its programs and services.Crucially, however, heightened political extremism works to curb these inherent growth aspirations, particularly when the presence of a divided government at the state level creates additional uncertainty about the position of elected leaders.Taken together, the results suggest a connection between the extremism of political leaders and the restraint of public officials that-in the absence of an explicitly longitudinal approach to this question-has heretofore remained unrecognized.

ENDNOTES
1 The concept of longitudinal research is a broad church with a myriad of understandings permeating the literature.We return in detail to this conceptual discussion in the next section.
2 One can also differentiate between "objective time" as an objectively determined order and "subjective time" as expressed in individual or organizational perceptions and expectations (Ancona et al., 2001;Shipp & Jansen, 2021).
3 This links, for instance, to Kingdon's (1995) "windows of opportunity" concept as well as the importance of "time tactics" raised by Pollitt (2008).The latter concept refers to the strategic use of time by involved actors, such as politicians engaging in delaying tactics (e.g., when setting up a committee to investigate a certain matter) or speeding up decision processes (e.g., by temporarily suspending preexisting rules).
4 This relates to several well-worn concepts taught within the policy sciences to explain change, or its absence, over time.Alongside Baumgartner and Jones' (1993) "punctuated equilibrium" lie the much-cited concepts of "paradigm shifts" (Hall, 1993) and "path dependency" (Pierson, 2000).In effect, as illustrated by Kickert and van der Meer (2011), the historical institutionalist approach and its associated concepts of "critical junctures" and "path dependence" have been a source of considerable inspiration in public administration research.Yet, the temporal dimension that is inherent to them is rarely identified.