Abstract
Motivated reasoning theory is a psychological theory that reads that policymakers interpret evidence in ways that fit their preferences rather than assessing it neutrally. The theory is increasingly used to explain policy processes as part of a behavioural approach to public administration, but it has limitations. As psychological research relies on experiments, the question remains what role motivated reasoning plays in real-world policy processes. Based on ethnographic observations collected during the planning phase of a large infrastructure project, this study confirms that motivated reasoning explains how people interpret information. However, it also shows that peoples’ context has a great impact on their reasoning. Ultimately, we suggest that a focus on time and real-world context is essential in understanding processes of reasoning, for which methodological diversification is needed.
People are inclined to interpret information in light of existing attitudes, rather than approach it neutrally. They read it in such a way that it confirms their attitudes, or are critical of it when it does not. Conflicts caused by differentiating views can be better understood by looking at the attitudes that inform these views. Discussions that might seem aimless at first might have secondary functions such as building trust amongst participants.
Keywords
Introduction
It is becoming increasingly popular to study public administration through a behavioural lens, drawing on psychological theories and behavioural economics to explain decision-making processes (Grimmelikhuijsen et al., 2017; Kasdan, 2020). A widely used example of such a theory is motivated reasoning theory. Motivated reasoning implies that policymakers are subconsciously inclined to accept information that confirms their attitudes and matches personal experiences more easily than contending information (Bækgaard and Serritzlew, 2016; Christensen et al., 2018; Gigerenzer and Gaissmaier, 2011; Thaler and Sunstein, 2009).
One of the particularities of behavioural (public administration) research is its reliance on experimental methods (Moynihan, 2018). This means that behaviour is studied in a stylized experimental setting, rather than a real-world context. However, research in both public administration (Jones and Baumgartner, 2012; Lindblom, 1979; Simon, 1976) and psychology (Gigerenzer and Brighton, 2009; Hertwig and Grüne-Yanoff, 2017) shows that this real-world context is especially important in understanding decision-making. In other words, that policymakers engage in motivated reasoning during an experiment does not tell us the extent to which they will do so in real-life settings, nor how the influence of motivated reasoning mechanisms holds up against external influences.
This paper reflects on the explanatory power of motivated reasoning in a real-world policy context. It uses concepts from motivated reasoning theory to analyse one year of ethnographic observations collected during the planning phase of a large infrastructure project, following a programme management team developing and then interpreting a report monitoring the progress and predicting the output of a set of projects. The contribution of this paper is threefold. First, this approach allows us to analyse if and how policymakers engage in motivated reasoning in a real-world policy context. Second, it shows how motivated reasoning affects the course of a policy process. Third, it reflects on the added value of studying mechanisms such as motivated reasoning through observations in real-world contexts.
The paper will first expand on theories of the impact of motivated reasoning on decision-making and the impact of context on decision-making in policy processes. Then, we explain how we collected and analysed our data. In the Results section we observe that, in accordance with motivated reasoning theory, policymakers are likely to perceive and use the information in reports selectively depending on prior attitudes. The decisiveness with which they do so makes policy process come across as illogical and incoherent at times. Ultimately, policymakers’ interpretation of information only changes under contextual pressure.
Engaging with information in policy processes: motivated reasoning in a complex context
In this paper, we study motivated reasoning by observing the use and interpretation of a piece of information used in a real-world policy process. More specifically, we follow policymakers’ reactions to a report that is introduced as evidence for the expected output and progress of an infrastructure programme. This kind of evidence generally serves to ‘inform the development and implementation of policy’ by evaluating ‘the effectiveness of policy options to inform decisions on what policy action to take’ (Sanderson, 2002, 4) and ensure the quality of policy (Nutley et al., 2007).
However, the policy process seldomly is so orderly and linear that it allows such a direct application of evidence (Cairney, 2016; Gerrits, 2012). Policymakers operate in networks of organizations in which customs, (organizational) interests, sensemaking and habits play as much of a role as the evidence contained in reports (Bevir and Rhodes, 2010; Halpin, 2011; Jones and Baumgartner, 2012; Thornton and Ocasio, 2008). In this complex reality, policymakers are expected to rely on a multitude of rules of thumb or ‘heuristics’ when making decisions. One of these heuristics is motivated reasoning (e.g. Nørgaard, 2018; Linde and Vis, 2017).
Motivated reasoning theory is based on the idea that people are inclined to reason towards certain goals (e.g. Kunda, 1990). This drives them to subconsciously evaluate information such as a piece of evidence in a way that confirms their prior attitudes and beliefs. It is brought about by the way in which people ‘affectively tag’ concepts (Fazio et al., 1986). An affective tag is an emotive association with a concept and determines one’s attitude towards said concept. Affective tags can be positive or negative and strong or weak depending on the experience one has had with that concept (Lodge and Taber, 2000). If the affective tags connected to a concept are weak, this leads an individual to be almost non-attitudinal towards a concept. If they are strong, they provide the basis for strong attitudes (Bargh et al., 1992; Lodge and Taber, 2013).
According to Taber and Lodge (2006), motivated reasoning has the following concrete effects on people's reasoning:
People will exhibit a prior attitude effect, meaning that they will evaluate arguments that support their attitude as being stronger than arguments that oppose their attitude. In addition, people will spend more time critically evaluating arguments opposing their attitude, which is called disconfirmation bias. Furthermore, they will be inclined to seek out information confirming their attitude, which is called confirmation bias.
1
Over the course of a process, these biases and effects will lead to attitude polarization, meaning that people's attitudes become more extreme over time.
Motivated reasoning is believed to transpire in situations of information overload or high complexity (Walgrave and Dejaeghere, 2017). Long and complex policy processes—such as the infrastructure policy processes that we observed—are likely to be such contexts. Policy is increasingly made in collaborative arrangements (Ansell and Gash, 2007), where policymakers represent organizations which themselves are complex constellations of different, seemingly independently moving parts (Orton and Weick, 1990) that can have different, conflicting goals and their own routines, rules, values and habits (Thornton and Ocasio, 2008). This results in a situation in which a piece of evidence is likely to have different meanings to different actors in a policy process, and a level of complexity that probably incites motivated reasoning.
Case and methods
To study processes of reasoning in a real-world policy context, we employed an ethnographic research design based on Strauss and Corbin’s (1998) grounded theory research strategy. Our ethnographic design had two distinct advantages over other approaches. First, it enabled us to witness changes in the use and interpretation of evidence over time. Second, an ethnographic design allowed us to move beyond how people summarize and idealize their practice in interviews (Czarniawska, 1997) and instead focus on practice as a messy day-to-day experience (Vagle, 2016, 58). In the remainder of this section, we introduce the setting in which we conducted one year of observations and explain how we analysed the data collected there.
Case
Between April 2018 and April 2019, we were able to observe a series of meetings of the programme management team of a highway infrastructure programme in the Netherlands. 2 The Netherlands is ‘commonly regarded as one of the strongholds of policy analysis’ (Van Nispen and Scholten, 2015) with a longstanding tradition of integrating analyses into policy processes. Dutch law requires infrastructure policy processes to be informed by a wide variety of studies (Ministerie van Infrastructuur en Milieu, 2016), making management team meetings a likely place to encounter many instances of evidence interpretation.
The programme management team we observed had been charged with managing the progress of the infrastructure programme. The team consisted of 10 representatives of all organizations that contributed to the programme financially. All team members were civil servants. Five of them were employed by the national government and five by regional or local governments. The group reported to a so-called political ‘steering committee’. This committee consisted of a group of high-level civil servants and aldermen who were politically responsible for the progress of the project.
The programme management team typically met every other week. An average meeting lasted about 3 h. Over the year, the team discussed everything from complex traffic studies to the layout of the project website and from budget mutations worth millions of euros to whether cycling would be a fun team-building activity. We used theoretical sampling as defined by Strauss and Corbin (1998: 211–212) to select the discussions surrounding a monitoring report as the focus of our analysis. Discussion of this report occurred frequently throughout the process. Our initial coding revealed that when the monitoring report was on the meeting agenda, discussions were usually elaborate and interpretations of the content of the report differed significantly amongst meeting participants. This led us to expect that focusing our analysis on discussions of this report would provide a wide range of interpretations whilst preventing variance in the type of information discussed from influencing observed reactions.
Data collection
We observed discussions on the monitoring report during a total of 44 meetings over the course of a year, adding up to 116 h of observations. Our primary way of gathering data was through field notes. Because the setting was often quite static, the primary focus of these fieldnotes was on what people said and how they interacted. We paid attention to tone of voice and non-verbal communication. In addition, our field notes contained details on setting. Aiming for saturation (Schwartz-Shea and Yanow, 2009: 67), we continued observing meetings until we were confident that no major new patterns would occur.
We had no role in the process other than to observe it. This meant that we could maintain a relative distance from the process, or a ‘disengaged position’ (Ybema and Kamsteeg, 2009). To ensure that we understood the contents of the meetings, we engaged in ‘member checking’ in small conversations with participants during meeting breaks (Schwartz-Shea and Yanow, 2009: 62). In addition to the meetings, we also had access to preliminary versions of the monitoring report, as well as a variety of other documents. These were not part of our analysis, but were necessary to understand the content of the processes we observed. The people we observed were promised anonymity. Therefore, non-essential details about the project as well as the people working on it have been altered in this article.
Coding strategy
As our data comprises a substantial number of meetings taking place over the course of several months, the main purpose of our coding strategy was to create a summarizing overview of the many discussions of the monitoring report we witnessed. The theory of motivated reasoning is centred around the idea that people have positive or negative attitudes (affective tags) towards concepts that influence how they initially interpret and respond to phenomena when they (re-)encounter them. Therefore, we decided to make reactions the focal point of our coding. Because we were present at discussions during all stages of the process of making the monitoring report, reaction-focussed coding allowed us to deduce people's prior attitudes from the positions they initially took and link these to their interpretation of the monitoring report in subsequent discussions. Ultimately, our coding resulted in a schematic overview of these discussions in the form of reaction patterns. In the Results section of this article, we then link these patterns to the four behaviours that motivated reasoning theory expects to occur (prior attitude effect, disconfirmation and confirmation bias and attitude polarization) to see if and how policymakers engage in motivated reasoning and reflect on the explanatory power of motivated reasoning theory in a real-world policy context.
We coded the data in NVivo 12. Coding took place in several steps. The first step was the line-based open coding of a number of meetings to generate a first set of reactions. Next, these codes were used during theoretical sampling to identify the discussions surrounding the monitoring report as a rich site to study evidence interpretation. Further discussions of this report were identified and coded in a second round of open coding.
The open coding stage resulted in 56 different reaction types. These were reviewed in a third round of coding, in which codes generated during the open coding stage were grouped in thematic categories that had explanatory power in our specific case (axial coding, see Strauss and Corbin, 1998: 123–127). Some of these were clearly linked to an emotive state, such as fear; others related to a broader concept, such as trust. The categories generated in the third round of coding are displayed in Table 1. Counting the number of times that each axial code occurred in each meeting allowed us to make displays of frequently occurring codes over time (based on Miles and Huberman, 1994). These reaction pattern displays formed the basis of our analysis, which is presented in the next section of this article.
Reaction categories after axial coding, with examples from our data.
Because field notes were written in Dutch, which is only spoken by one of the authors, coding was done in an iterative manner. Lars Dorren coded the data, but both authors worked on the coding scheme at every stage of the process. This meant that the authors discussed codes and their applicability to translated samples from fieldnotes and decisions during the axial coding stage were made jointly.
Results
Our study centres around the reasoning of policymakers concerning a monitoring report. The monitoring report was meant to monitor the progress of a group of short-term measures which were part of an infrastructure programme and was frequently discussed by the programme management team (hereafter referred to as the ‘team’) responsible for these short-term measures. The goal of these short-term measures was to ensure that traffic would keep flowing smoothly on a particular stretch of highway as the team worked on longer-term measures for the same trajectory. The measures included in the programme were quite diverse, covering everything from reconfiguring highway exits and entrances to a campaign targeting local businesses to motivate their employees to use different modes of transportation. Some measures received local media attention, but none of them seemed to give rise to much public controversy.
To monitor the progress and expected effects of these measures, the report relied on a reduction of the number of cars using the highway during rush hour, hereafter referred to as ‘traffic reductions’, as the central indicator. Some of the short-term measures were managed by members of the team, but most of them were managed by staff members who were not part of the team. The majority of these staff members were employed by the regional or local governments involved in the programme.
The report central to our study was made by an external analyst, who regularly sent updated versions of the report to the team. The analyst did not appear to have much of an opinion about the different policies that their work would be monitoring. Rather than advising on content, they spent most of their time with the team discussing the complexities of measuring and predicting the impact of policies. Outcomes were presented neutrally, without passing much judgement.
Analysing our coding, we found substantial differences between reaction patterns displayed by regional government representatives (RGR) and national government representatives (NGR). These two distinct reaction patterns are displayed in Figures 1 and 2. Figure 1 presents the type (indicated by the different colours) and the number (indicated by the height of the bar) of reactions displayed by NGR; Figure 2 presents reactions as displayed by RGR. Each number on the horizontal axis corresponds to a single meeting.
Coded reactions to the monitoring report by national government representatives, in absolute numbers, per meeting.
Coded reactions to the monitoring report by regional government representatives, in absolute numbers, per meeting.
Focussing on explaining the differences in reactions between RGR and NGR, we will proceed as follows. First, we will provide a descriptive overview of team members’ attitudes towards the monitoring report and the policy process in general at the beginning of the observation period. Second, we will explain how these attitudes shaped team members’ reactions to the monitoring report. Third, we describe how throughout the policy process, contextual pressures came to increasingly shape interpretations of the report. Under the influence of time constraints and political pressure, we observe how attitudes and interpretations – even those held firmly – eventually change.
The start of the process – policymakers’ goals and attitudes
While it is difficult to determine the true start of a process of thought and interpretation, the monitoring report was first discussed in a meeting in April 2018. Back then, we found the process in relatively calm waters. The previous meeting with the political steering committee of the programme had taken place several months prior and the next one was to happen months later. Many projects in the programme were still at an early stage of development. In this absence of political pressure and managerial urgency, team members expressed the attitudes informing their initial reactions to the monitoring report.
In contrast to later stages of the process, attitudes still appeared largely similar among team members. What would be the focal point of much controversy later – the status of the measurements in the report – was now still being discussed in an atmosphere of mutual agreement. For instance, reactions to the question of how to predict and measure policy impact were characterized by great enthusiasm for adopting ‘innovative’ or ‘out of the box’ ways of measuring, but not so much for traffic reductions as an indicator specifically. In meeting 3 in Figures 1 and 2, the programme manager (PM) even enthusiastically argued to eliminate the idea of using traffic reductions as the main indicator for output altogether, calling it an ‘old-fashioned’ concept. The absence of truly negative reactions in this first meeting also indicates that no one in the group had an openly negative attitude towards the basic idea of monitoring projects or making projections of project outcomes.
In addition to sharing a non-negative attitude towards monitoring, the team shared a commitment to deliver a well thought-out, well-performing programme. Team members’ positions with regards to how to reach this goal differed, though. This difference became apparent during the meetings following the first meeting in April 2018. By then, the analyst had started interviewing project staff members to gather input for the monitoring report. Based on these interviews, the PM indicated that they were worried about the results that the projects would yield (see meeting 12 in Figures 1 and 2). The predicted amount of traffic reduction appeared to vary widely, depending on who one asked. The PM explained that the number 3250 was ‘echoing through the halls’, but they were alarmed by the fact they had also heard estimates far below this number. Team members from both groups of policymakers expressed that they were glad that the PM had shared this information and began to consider an appropriate response.
As we will see later, the consensus amongst team members would not last. The primary reason for this was that team members had different visions of how to reach the goal of delivering a well-performing project. NGR wanted to meet with sub-project staff often and tried to impose specific styles of progress reporting on project partners, stressing that the assurance that staff were working hard was not enough and that they wanted ‘hard numbers and facts’ about their progress. NGR feared that they could not adequately monitor the progress of the short-term projects, which would in turn put the sub-projects at risk of being delayed. The fears of NGR were amplified by the fact that their home organization would ultimately be held accountable for the success of the project. RGR, on the other hand, relativized the importance of frequent meetings with project staff. They often indicated that they would rather see the project staff ‘just [keep] working on projects’ instead of having to spend time on frequent progress reports.
In sum, the beginning of the process was characterized by calm and open discussions. Team members shared a commitment to deliver a good result, were open to a variety of methods when it came to monitoring projects and had a similar interpretation of the report. However, we also observed a difference between RGR and NGR when it came to their attitude towards project management. This difference would become the central driver of the conflict to come.
Evaluating the evidence in the monitoring report
Over the next few months, the team discussed several versions of the monitoring report. The analyst worked on the report continuously and occasionally sent an updated version to the team for feedback and discussion. Starting in October, we began to observe differences in the ways that team members responded to these preliminary versions of the report. During meetings, a re-occurring pattern emerged. Supported by other NGR, the PM would raise a concern about whether the projects would in fact be able to generate 3250 traffic reductions, which RGR would then relativize. The PM's fears were amplified during a meeting in which all members of the project staff were invited to present their progress (meeting 7 in Figures 1 and 2). The staff members gave presentations that were not very concrete. The phrases ‘preliminary stage’ and ‘premature’ featured in most presentations, which were met with a great deal of worry from the PM. In their view, what was an estimated result earlier on had become a target, which they feared the staff members would not be able to make. Any predicted outcome below 3250 was now seen as unsatisfactory, rather than just the newest estimate.
Contrary to the reactions of NGR, RGR did not appear worried. Where they initially supported the PM's concerns about the project outcome, they now downplayed their fears and expressed concern about the new status of 3250 as a target. These concerns were partly organizational–political. For example, one RGR stated that half a year ago, he told his executive to invest in certain projects, and he did not want to now suddenly have to say that these same projects were ineffective. If he were to do that, this representative argued, it would have financial consequences.
The concerns of the RGR also had to do with the methodology of the monitoring report. In a particularly heated discussion, we observed one RGR refusing to talk about projected project results because ‘right now, there is no result. We are working on the result!’ In the remainder of that meeting, the PM tried to assuage the representatives by stating that perhaps they had been a bit too pessimistic; some projects were doing very well and might even overperform. This sparked a methodological discussion, in which regional representatives made remarks such as, ‘but where does that come from? Out of nowhere!’ The PM could not really answer this, except to reply that the number came from ‘under the analyst's hood’. 3 By the end of the meeting, the PM concluded the discussion by accepting that it might be best to report the progress of the sub-projects to the steering committee in terms of process, but not include the projected traffic reductions.
In contrast to what one might expect, similar discussions were observed in meetings 21, 23 and 27. The PM appeared to forget the outcomes of previous meetings and had to be reminded repeatedly that they had agreed to not report expected outcomes to the steering committee. Despite team members’ attempts to downplay the outcomes of the study, we found that the PM repeatedly insisted on informing the steering committee on the outcomes of the report, arguing they would want ‘hard facts, measurable targets’. In meeting 27, RGR even started to receive support from NGR, with NGR claiming that it would cause unnecessary panic to report projected outcomes to the steering committee. Like the ones before it, this meeting was concluded by postponing the decision on what to do with the monitoring report.
These types of discussions were exemplary for this stage of the project, and can clearly be linked to motivated reasoning theory. Even though all team members were talking about the same report, they treated the information in it differently. The PM took the information in the preliminary versions of the report at face value and used it to argue for more managerial control. RGR, who were in favour of letting the project staff ‘just work on their projects’, questioned the report's methodology and wondered why they suddenly had to deal with a target that seemed to come out of nowhere to them. At the beginning of the process, when the specifics of the monitoring report were not yet known to policymakers, there was no substantial disagreement. Now, we can see how team members’ reactions to the evidence in the report differ in accordance with their attitude towards project management. In addition, the re-occurrence of this kind of discussion fits the idea that policymakers are motivated reasoners. An archetypical rational actor, guided by reason rather than motivated reasoning, might be expected to remember that, for example, in meeting 21, the team debated and then decided to only report on progress and not outcomes. In contrast, we observed how this debate was repeated in meeting 23 and then again in meeting 27 before its conclusion stuck.
Constant re-evaluation of evidence and changing attitudes
After a period of recurring discussions on the exact meaning of 3250 as an outcome during which the attitudes of team members seemed relatively fixed, we started observing shifts in the ways in which team members interpreted the monitoring report. These shifts occurred when the context in which the report was discussed changed. The atmosphere of the meetings we observed gradually changed over the course of the year. In April, meetings would last about 2 h and would unfold at a leisurely pace. After the summer, team meetings always lasted at least 3 and occasionally over 4 h. They were now filled with heated discussions. Staff members were regularly called in to discuss their project's progress, and team members appeared to be rushing to get their work done before the end of the year. Additionally, the next meeting with the political steering committee was scheduled to take place by the end of January 2019, meaning that the urgency of the question of how to interpret the monitoring report increased.
The main change in attitudes that occurred was that, rather than relativizing the predictions in the monitoring report and calling them ‘estimates’, RGR started referring to them as ‘targets’ in discussions with project staff. RGR were now regularly observed pushing staff members to explain how their work would result in a certain amount of traffic reductions. It even occurred that a staff member who asked where this target came from was told that it was ‘simply a result of the analyst's work’, an argument which was met with scepticism by RGR when the PM had made it during an earlier meeting. The alleged target number was also used as input for a government study that would inform decisions about the long-term measures that were part of the programme. This further cemented its status as a target to make, as now longer-term policy choices were made based on the idea that the projects would result in 3250 traffic reductions.
The new use of the information in the monitoring report as a target by RGR also changed the team's discussion about the interpretation of the contents of the report. At the end of November, the group received a final version of the report and debated how to communicate it to the steering committee. What was striking this time was that the reactions coded as expressing fear could now be linked to RGR (see meetings 35 and 36 in Figures 1 and 2). RGR started becoming concerned about the costs of the project in relation to its expected results. Or, as one representative stated: ‘If we give 3250 people a […] gift card and tell them not to use the road again, it would cost us less. How am I going to explain this to my [political executive]?’ RGR's comments surprised the PM, who wondered why during the past year, these concerns had not been brought up by anyone but the PM themselves. Not even attempting to answer that question, the group hastily started drawing up proposals on how to report the output of the monitoring report to the steering committee, with ‘3250’ now seen as a target by NGR and RGR.
Analysis: the impact of motivated reasoning
Throughout the process we followed, we saw the number ‘3250’ evolve from a number ‘echoing through the halls’ to a hard target worrying RGR, without the monitoring report itself undergoing fundamental changes. We have shown that – like motivated reasoning theory suggests – the interpretation of information is initially influenced by prior attitudes. Team members who wanted to closely monitor the progress of the project staff took information from the monitoring report at face value, whilst team members who wanted the staff to ‘just do their job’ continuously questioned the validity of measurements and downplayed the outcomes of the monitoring report. In other words, both groups of team members focussed on those aspects of the report that confirmed their position, showing what motivated reasoning theory calls a prior attitude effect.
Throughout the process we observed, the exact status of ‘traffic reductions’ as an indicator – whether it was a hard target or a soft estimate – remained unclear and was frequently questioned by team members. At the same time, the amount of traffic reduction predicted in the monitoring report was regularly used as if it were an indisputable fact. This usage either depended on someone's attitude towards project management or on contextual conditions. For example, team members who would first relativize the contents of the monitoring report because of their more relaxed approach to project management later portrayed the projected traffic reductions as an indisputable fact to motivate project managers. This behaviour corresponds to what motivated reasoning theory calls confirmation and disconfirmation bias, where people are inclined to, respectively, seek out or interpret information in ways supporting their position and more critically evaluate arguments countering their position.
Exactly how context dependent the way in which people reason is, is illustrated by the fact that after initially opposing and relativizing the contents of the report, RGR later used it to calculate the costs for each traffic reduction and wondered how they could convince their superiors that these costs were justifiable. In situations where there was little contextual pressure, team members reasoned more freely and prior attitudes played a more substantial role in their reactions. Contextual pressure caused team members to reconsider their positions or made it so that other attitudes came to play a more important role. The fact that an individual displayed certain attitudes at the beginning of a process did not mean that they would then consistently hold on to those attitudes throughout the process.
Furthermore, the observation that interpretations of the monitoring report were attitude driven explains why this policy process, like many other processes, feels messy to an outside observer. Analysing a policy process by focussing on people's initial reactions lays bare a policy process that is far less logical or strategic than one might initially expect. In our observations, team members had discussions only to forget the outcome and repeat the same debate weeks later and interpreted the monitoring report in mutually exclusive ways. This article shows that using concepts from motivated reasoning as analytical tools, one is able to account for the erratic ways in which this policy process unfolded.
Conclusions and discussion
This study set out to explore the explanatory power of motivated reasoning theory in a real-world policy context. It shows that, as motivated reasoning theory suggests, prior attitudes largely determine how people initially interpret information. It also shows that reasoning is an impulse-driven process, starting with a spontaneous response based on prior rather than a ‘neutral’ re-reading of the information each time it is encountered. The extent up to which people's initial attitudes play a role depends on people's context. We observed how, for example, interpretations change under the pressure of a deadline, or when people want to use information to motivate project staff members. As a result, single individuals used multiple, incommensurable interpretations of the outcomes of a report throughout the policy process we observed.
For both the practice and the study of public administration, this study has two main implications. First, it suggests that knowledge of people's attitudes is key in understanding policy processes. This means that when studying or caught in a process that seems to be in a deadlock, explicating the attitudes informing people's positions will help better understand or overcome this deadlock. Second, the study draws attention to the latent functions of knowledge use. It might be tempting to conclude that this study shows that deadlines help people overcome differences of opinion. However, previous research has found disagreement amongst process participants to increase the quality of reasoning (Klar, 2014). Also, this conclusion would too easily pass over the fact that before overcoming their differences, team members first spent the better part of 8 months familiarizing themselves with each other and their positions. Precisely because policy processes are characterized by spontaneity and ad-hoc decisions rather than rational bargaining, one needs to allow decision-makers some room for discussions that seem repetitive or aimless to the outside observer.
In addition to the two main implications, the study provides grounds to re-evaluate the way we study policy processes. It shows that the context in which processes of reasoning take place is an important influence on the way in which these processes unfold. It is safe to expect that motivated reasoning will occur, but the exact way in which it does depends on the specificities of the context that people find themselves in. This means that conclusions arrived at by observing behaviour in artificially constructed settings does not necessarily lead to conclusions that are able to predict the course of real-world policy processes.
Also, the study draws attention to the fact that time is an important factor in understanding processes of reasoning. Would we have observed this particular policy process for a shorter period of time, we would probably not have witnessed the changes in people's positions, nor would we have been able to explain what caused people to change their position. All in all, this study leads to the methodological recommendation that real-world reasoning can best be understood by methods that follow a process over time, rather than analysing a fixed snapshot of that process.
Footnotes
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: this work was supported by the Flemish Government (Research Centre Governance Innovation 2016–2020).
