(Re)framing and the confirmation bias in sensemaking in the design process

(1)

(RE)FRAMING AND THE

CONFIRMATION BIAS IN

SENSEMAKING IN THE

DESIGN PROCESS

Winnie Brouns (11418737) 23-6-2017 Final version

MSc. in Business Administration – Entrepreneurship and Innovation Track, UvA Supervisor: Dr. W. van der Aa

(2)

1

Statement of originality

This document is written by Student Winnie Brouns who declares to take full

responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and

that no sources other than those mentioned in the text and its references have

been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision

of completion of the work, not for the contents.

(3)

2

Abstract

Sensemaking has shown to be very important on an individual, group or organizational level for innovation performance. However, the cognitive processes in sensemaking are largely a black box, with only the outcomes visible. This thesis aims to provide more insight in the preservation and reframing behavior in sensemaking in the design process and the role that confirmation bias has in this. Hypothesized was that providing a frame, would lead to preservation of this frame and in turn to lower reframing propensity. This effect would be stronger for a frame that conforms to the general consensus, than a frame that is opposite to the general consensus. This was researched by executing an adapted version of the Stanford Design Thinking Experiment, which is able to measure preservation and reframing behavior of 30 master Business Administration students. Although, none of the hypotheses could be confirmed based on significant results, some patterns that resemble the expected outcomes could be identified. It was found that the preservation of a frame was higher in case of a frame based on general consensus than when the given frame was opposite to the general consensus or when no frame was provided. In addition, the total reframing score was lowest in case of a frame confirming to the general consensus was provided, followed by the opposite frame and the highest scores when no frame was provided, reflecting the expected results. The preservation of a frame is regarded as an effective strategy by testing and adapting it based on new data, both by the Data/Frame theory and confirmation bias literature. However, these results point towards a view in literature, that preservation of a frame can lead to accepting confirming data too easily and explaining away disconfirming data. This may lead to inaccurate frames and in the end suboptimal design outcomes that are not accepted by customers and stakeholders. Although these findings are not supported by conclusive evidence, they do provide valuable implications for future research.

(4)

3

Introduction

Over the past years, design thinking increasingly has received attention. Design thinking uses designer’s tool for human-centered innovation. In order to create a solution that will be accepted by the stakeholders, it is important to integrate collected information in the design process. This process is called information synthesis or sensemaking. The synthesis of information collected, e.g. user research data, is a decisive factor of what external knowledge will be included in the design outcome. Furthermore, it is found that creating a good fit between external information and the internal design process is a main driver behind market success of a design outcome (Hey et al., 2008). Although it is an important step for a successful design outcome, this importance has not been represented by the presence in literature. Much has been written about diverging techniques in the design process, such as brainstorming. However, the convergent topic of information synthesis has received considerably less attention (Gumienny, 2011). This underrepresentation in literature in combination with the fact that information synthesis is a complex and cognitively extremely challenging part of the design process, leads to information synthesis largely being a “black box” with only the design outcome visible. Information synthesis is thus largely a non-visible and intangible process, including large amounts of tacit knowledge (Gumienny, 2011). Although synthesis is also performed in teams, large parts of the synthesis process are performed privately, e.g. in own head or on scratch paper. This makes it challenging to define the cognitive processes behind the information synthesis process (Kolko, 2010).

Klein and his colleagues attempted to clarify the sensemaking process and create a framework: the Data/Frame model (Sieck, Klein, et al., 2007). They identified that people start with a frame. Klein et al. (2006b) describe a frame as: “an explanatory structure that defines entities by describing their relationship to other entities.” From there, the frame can be preserved and optionally be elaborated with further information. The other possibility is to reframe. Reframing is defined as the disassembling of an existing problem solving frame set and reassembling a new frame set, possibly leading to a new problem solution (Plattner et al., 2012). An article of Klein et al. (2006b) indicates that people tend to early recognize a frame and preserve it by explaining away data that do not match the frame. There has been contradicting findings in literature about the effect of reframing and preservation. Some believe that the tendency to preserve leads to suboptimal outcomes in the end. Klein et al. (2006b) believes that early consideration of a frame and preserving it, helps to do more efficient research and search for information. Although the mentioned article does not recognize this as a negative phenomenon, but merely a more efficient way of data gathering, similarities with cognitive heuristics (and the possibly negative effects of these) can be indicated.

(7)

6 It has been shown that in information processing cognitive heuristics indeed can play a role. Cognitive heuristics are described by Hallihan et al. (2012) as “…unconscious rules utilized to enhance the efficiency of information processing and are possible antecedents of cognitive biases.”. The hypothesis confirmation bias, or confirmation bias, is one of the most prevalent cognitive biases in information processing (Nickerson, 1998). This confirmation bias might have an effect on the information processing in the sensemaking stage of the design process and thus affect the design outcome in the end.

Although some clear similarities can be identified between findings of the Data/Frame theory and the confirmation bias, literature has not succeeded to clearly indicate the presence of confirmation bias in the sensemaking process and the relationship between the choice of preserving or reframing and the confirmation bias. Contribution to insights in the cognitive processes behind the sensemaking process can be very relevant and help organizations to more effectively execute the sensemaking process. Insights in the confirmation bias in sensemaking can increase the awareness of this bias and potentially help to develop mitigating strategies to overcome negative effects. The subject is very limited described in literature and thus leaves a research gap.

Subsequently, this thesis aims to provide more insight in the preservation and reframing behavior in the sensemaking process and the role that confirmation bias has in this. This will be done by building on literature about confirmation bias (in general and in the design process) and literature about the sensemaking process. By comparing the literature, an attempt is made to find similarities and translate some of the confirmation bias literature to sensemaking in the design process to create testable hypotheses. To test these hypotheses an experiment, based on the Stanford Design Thinking Experiment is performed. The research question answered in this thesis is: What effect does the early provision of a frame in the sensemaking process (of the design process) have on the propensity to preserve a frame and the propensity to reframe and what role does the confirmation bias play in this?

(8)

7

Literature review

Information synthesis in the design process

In the design process several steps are taken in order to come up with innovations. The design process starts with collection of information that contributes to the understanding of users’ needs, stakeholders’ interests and possible limitations of the solution. In order to create a solution that will be accepted by the stakeholders, it is important to integrate the information in the design process and extract meaning from the data (Gumienny et al., 2011). While this is done throughout the design process, this is mostly done in the moment between research and definition (Kolko, 2010b). The process described here is called information synthesis and is further explained by the following definition:

Jon Kolko (2010): “Synthesis is an abductive sensemaking process. Through efforts of data manipulation, organization, pruning, and filtering, designers produce information and knowledge.” Information synthesis has shown to be a cognitively extremely challenging process. The act of filtering, organizing and sensemaking of ambiguous information is complicated and exhausting (Gumienny et al., 2011). Also, literature about information synthesis is quite fragmented and different terms are used across articles. Information synthesis is also described with terms as framing or sensemaking. Sensemaking is described as: “a process that is both personal and shared, takes place over a longer period of time and that is heavily dependent on a perspective or point of view” (Kolko, 2010b). In line with this Dervin (2003) (as cited by Kolko, 2010b) sees sensemaking as “a process that is personal and contingent on experience, that substantiates learning, that takes place continually and forever, and is fundamentally based on each participant’s perspective or point of view.” This implies that sensemaking is a subjective rather than an objective process and heavily based on unique experiences, emotions and history of the designers. Consequently, the outcome of the sensemaking process is still very unpredictable and will vary when different designers are asked to solve the same problem. In addition, this leads to the uniqueness in the designer’s ability to reframe and to empathize. In other words, the ability to consider situations or problems from different perspectives and to make logical inferences from a new viewpoint, is a unique skill of design (Kolko, 2010b).

Sensemaking and creativity and innovation in organizations

Sensemaking is an essential part of the design process and innovation, especially since the design process has a human-centered approach. Sensemaking allows to make sense of the information (about customers, stakeholders, etc.) that is collected, create new understanding and identify new

(9)

8 opportunities. It is a decisive point in the design process, since it determines the quality of external knowledge that will be included in the design outcome. Creating a good fit between external information and the internal design process, is one of the main determinants of market success (Gumienny et al., 2011). The focus of this thesis is mainly sensemaking on an individual or project level. However, the sensemaking process also has an effect on creativity and innovation on an organizational level. Although the link of sensemaking to creativity and innovation does not have a large literature base to draw from, literature indicates some interesting implications of sensemaking for creativity and innovation in organizations.

Drazin et al. (1999) looked at the effect of sensemaking process on creativity from a multilevel perspective. They propose that the creative process is based on episodes of sensemaking. These are initiated as a result of crises that arise in projects within organizations. With every episode of sensemaking a belief structure is created by negotiations, which is retained until the next crisis. In this belief structure there is a relative balance between technical and managerial belief structures which determines the extent to which each side engages in creativity. Then the power balance shifts, other actors come into play, which through sensemaking create new belief structures to drive the creative process.

Dougherty et al. (2000) have provided evidence of the importance of sensemaking for innovation in organizations. In their study, they compared more and less innovative firms and found differences in the way employees framed market and technology knowledge and the linked products and businesses. Employees in more innovative firms, evaluated themselves as engaging in knowledge and business practices directed at solving problems for customers by ongoing relationships with them. They engaged in collective sensemaking by interactively working on a problem with a shared understanding of the goal. Employees in less innovative firms, however, treated market and technology knowledge as separate factors of production. As a result of this, the sensemaking process was siloed and lacking a frame that encourages collective sensemaking.

The above findings show that sensemaking is an essential part for enabling creativity and innovation in organizations. An important factor in this is the collective sensemaking activities after crises or uncertainty which create new belief structures. The creation of these new belief structures can lead to novel understanding and new ways of doing business and in this way stimulate innovation. Also, collective sensemaking across different practices with a shared understanding of the goal, contributes to the innovative nature of firms.

(10)

9

Data/frame theory of sensemaking

Although sensemaking is still little understood, Klein and colleagues attempted to clarify the sensemaking process (Sieck, Klein et al., 2007). In cooperation with the U.S. Army Research Institute for the Behavioral and Social Sciences they conducted a series of experiments over a period of three years. This cooperation was found to be useful since army leaders often end up in situation which are unfamiliar or uncertain in which they must quickly make sense out of the situation. In the first year, Information Operations officers completed a series of scenarios in which the participant’s comments were categorized and coded. Also, real-life navigational incidents of sensemaking were studied. Year 2 included Cognitive Task Analysis data collection methods, series of methods and tools used to gain in depth access to mental processes that underlie performance of tasks. Year 3 was mainly focused on testing the differences of novices and experts in sensemaking.

Based on these results and review of literature, they gained insights in the sensemaking process. They defined sensemaking as: “…the process of fitting data into a frame, and fitting a frame around the data”. Generally, people will try to make sense of data by constructing or finding a story to account for the data. Their repertoire of stories, which can based on individual perspectives, viewpoints and previous experience, affects the interpretation and which data they consider. This repertoire of stories is here described as frames. Klein et al. (2007) describes a frame as: “an explanatory structure that defines entities by describing their relationship to other entities” and “... a frame is a structure for accounting for the data and guiding the search for more data”. With the last sentence they describes that a frame works two ways: 1) frames shape and define the relevant data, and 2) acquisition of data changes the frame itself. Frames and data are thus working together to find an explanation or solution.

The results allowed to identify six sensemaking activities: elaborating the frame, questioning the frame, preserving the frame, comparing frames, seeking a frame, and reframing. Based on these findings a framework was created. This framework is the Data/Frame model and is a dominant theory of the sensemaking process. Figure 1 shows this model.

(11)

10 Figure 1 The Data/Frame Theory of sensemaking (Klein, 2006b)

In the design process, designers bring their individual frames to the design problems. These frames can range from simply an understanding of “the way the world works” to more sophisticated perspectives about functionality and performance based on prior experience and expertise (Plattner et al., 2012). Thus, the sensemaking process never starts from zero. A frame can then be questioned to see if the explanations it provides are actually valid. If this is the case the frame can be preserved and elaborated, to add extra data or find new relationships for example. When the frame does not seem valid, it can be reframed. The initial frame is rejected and a new and better frame is searched. Alternative frames can be compared to find the most accurate or a new frame can be sought. The next paragraphs describe these two elements, preservation and reframing, in more detail.

Preserving the frame

In the case of preserving a frame, the initial frame is accepted. The frame is perceived as (likely to be) true or relevant. As described before, the choice for acceptance or deviation from a frame is not necessarily negative or positive. In decision making it is sometimes advised that one should avoid early consideration of a hypothesis and keep an open mind in order to avoid fixation error (Heuer, 1999). However, in the Data/Frame theory it is believed that rapid recognition of a frame helps to gather information more efficiently. Also, by having more specific expectancies, which can be violated by anomalies, allows for adjustment and reframing (Klein et al, 2006). In addition, it was found that

(12)

11 commitment to a frame enables it to be tested effectively and learn from inadequacies, which is absent from an open-minded approach (Rudolph et al., 2003 (as cited by Klein, 2006)). This is based on an experiment executed by Rudolph and colleagues. Anesthesiologists were faced with a “garden path” problem, which is a setup that suggests one hypothesis, following contrary cues that point to a different hypothesis. Results showed that the participants that confirmed to the suggested hypothesis and jumped to an early conclusion, performed the worst. Participants that kept an open mind and refused to speculate were mediocre, while participants that jumped to an early speculation and then deliberately test it, performed the best (Klein, 2006). The initial hypothesis helped them to seek data that would be diagnostic, while an open mind was in this case a passive mode of receiving data without testing them too much. This suggests that methods that help decision makers to keep an open mind, can be counterproductive.

However, on the same time contradicting arguments can be found in literature. Klein et al. (2007) indicate that if the match between a set of data and a frame is more plausible than the match to any other frame, we tend to accept the first frame as the likely explanation. In this process, an individual attaches to anchors that are provided within the frame. Klein et al. (2007b) describes an anchor as follows: “The process of decision making usually begins with an initial judgement of the situation being faced. This initial value, or anchor, may be evoked by the formulation of the problem, the partial computation of available information, or in uncertain situations, seemingly trivial factors may have profound effects...” As described in this definition, some anchors may inappropriately receive too much attention. When attaching too much credibility to inaccurate anchors in this frame, it results in interpreting other data to fit in this inaccurate frame. An inaccurate anchor that comes early in the sensemaking process, has a larger effect on the outcome than one later in the process (Klein, 2007). In addition to this, they state that people tend to preserve a frame by explaining away data not matching the frame.

Reframing

A French sociologist and anthropologist named Bruno Latour stated: “To design is always to redesign.” (Latour, 2008). Just as described before, Latour believes that the design process never starts from zero. The designer starts with a frame and then reframes into “something more lively, more commercial, more usable, more user friendly, more acceptable, more sustainable and so on, depending on the various constraints to which the project has to answer.” This points towards a view that redesign and reframing are a central component of the design process. Reframing is defined as “the disassembling of an existing problem solving frame set and reassembling a new frame set, possibly leading to a new problem solution.” (Plattner et al., 2012). A designer’s propensity to reframe, their willingness to shift

(13)

12 existing problem-solving frame set based on the introduction of new information, is an important determinant of the design outcome. When the new information is important for shifting the frame, reframing is an appropriate activity. However, in the case of information not being important or relevant, reframing of the problem may not lead to a better outcome. It is the task of the designer to decide to which extent the new information leads to reframing (Plattner et al., 2012). Klein et al. (2006) adds to this by stating that frames provide certain expectations. In case of violation of these expectations, the designer may start to question the accuracy. In turn this may lead to reframing. The decision to reframe by introduction of new information is thus the responsibility of the individual. Consequently, this suggests that in the process of redesign certain cognitive capabilities can have an impact on the design outcome. However, these cognitive abilities still are largely non-defined. Plattner et al. (2012) assessed the cognitive profile of students in design teams based on the eight Wilde cognitive modes. Next, the students participated in an experiment in which reframing behavior was measured. None of the cognitive modes showed a significant correlation with individual reframing behavior. If indeed “all design is redesign” the ability to reframe (on the appropriate moments), involving breaking apart existing frames and reassemble new frames, is a critical component of the design process (Plattner et al., 2012). Insights in cognitive processes related to this could be very valuable.

To summarize, table 1 gives an overview of the most important literature findings about the subject of preservation and reframing in the sensemaking process.

(14)

13 Table 1 Literature overview of preservation and reframing

Positive Negative

Preservation Commitment to a frame enables it to be tested effectively and learn from inadequacies (Rudolph et al., 2003 (as cited by Klein et al., 2006))

Avoid early consideration of a hypothesis and keep an open mind in order to avoid fixation error (Heuer, 1999)

Rapid recognition of a frame helps to gather information more efficiently. Also, by having more specific expectancies, which can be violated by anomalies, allows for adjustment and reframing (Klein et al, 2006)

When attaching too much credibility to inaccurate anchors in the frame, it results in interpreting other data to fit in this inaccurate frame. An inaccurate anchor early in the sensemaking process, has a larger effect on the outcome (Klein, 2006)

Klein et al. (2006) state that people tend to preserve a frame by explaining away data not matching the frame

Reframing A designer’s propensity to reframe, their willingness to shift existing problem-solving frame set based on the introduction of new information, is an important determinant of the design outcome (Plattner et al., 2012)

In the case of information not being important or relevant, reframing of the problem may not lead to a better outcome (Plattner et al., 2012)

When the new information is important for shifting the frame, reframing is an appropriate activity (Plattner et al., 2012)

Neutral: Klein et al. (2006) states that frames provide certain expectations. In case of violation of these expectations, the designer may start to question the accuracy. In turn this may lead to reframing

(15)

14 Work from Klein and colleagues and the empirical findings of Rudolph et al. lead to the conclusion that early consideration of a frame does not have to be a negative thing. However, it is important to stay critical and test, adjust, or reject the frame if it is not accurate (as seen in the outcomes of the experiment of Rudolph and colleagues). In most cases the first frame will not be the most accurate frame and thus a designer’s propensity to reframe, their willingness to shift existing problem-solving frame set based on the introduction of new information, is an important determinant of the design outcome. However, as has been shown, providing credibility to (inaccurate) anchors, especially those introduced early in the sensemaking process, can lead to preservation of a frame and to explaining away data that do not match the frame. This is a wrong way to preserve a frame and can lead to suboptimal design outcomes.

As indicated above, Klein and colleagues believes that preservation of a frame is rather an efficient way to test the frame and collect new data. However, similarities with cognitive heuristics (and the possibly negative effects of these) can be indicated. In the next paragraphs the confirmation bias and implications for sensemaking will be discussed.

(16)

15

What is Cognitive Heuristics?

Cognitive heuristics are described by Hallihan et al. (2012) as “…unconscious rules utilized to enhance the efficiency of information processing and are possible antecedents of cognitive biases.” Cognitive heuristics are observations of natural occurrences within the human cognition, which are used without conscious intent and are not explicitly learned. Cognitive heuristics can actually lead to efficient information processing and minimize cognitive effort and thus can be beneficial. However, since using heuristics is not a conscious process, often people are unaware of how heuristics can lead to cognitive biases and irrational judgements (Hallihan & Shu, 2013). Heuristics have shown to influence a diverse set of complex decision making tasks. Based on the fact that decision making is an integral part of the design process, at least some reliance on cognitive heuristics is expected. When cognitive heuristics are relied on during the design process, it could lead to cognitive biases in the information processing in the design process. Numerous cognitive biases are described in literature. For this thesis, the focus will be on the hypothesis confirmation bias, which is considered as one of the most prevalent biases in human reasoning (Nickerson, 1998).

What is the hypothesis confirmation bias?

Confirmation bias refers to a tendency to seek out evidence, or interpret evidence in such a way, that is consistent with pre-existing beliefs, at the expense of considering belief inconsistent information (Hallihan, et al., 2012). In simple words, the hypothesis confirmation bias is a tendency to confirm rather than disconfirm a hypothesis or belief. This leads individuals to fail to consider the diagnostic value of supportive data and accept confirming data too easily in order to confirm the hypothesis. On the other hand, they misinterpret and fail to accept disconfirming data and based on this adjust beliefs (Nickerson, 1998). This leads to a bias towards the present hypothesis or belief while this belief may be inaccurate.

Hallihan and Shu (2013) further argue that confirmation bias may be a product of the availability bias. The availability bias is described as “an overreliance on the most available information in memory when making judgements”. The most available information is then supportive data, since this is easier to think of than disconfirming data. When individuals rely on the availability bias in the evaluation of design beliefs (and the most available data is supportive), they also may be prone to having a confirmatory bias.

However, the hypothesis confirmation bias does not automatically lead to poor information processing outcomes. It can be part of pragmatic and adaptable decision making strategies (Hallihan & Shu, 2013; Friedrich, 1993). It can serve a goal for the decision maker in terms of efficiency, practicality, etc. As an

(17)

16 example, the confirmation bias can provide a framework which makes searching for decision-relevant information more efficient than a random search (Hallihan & Shu, 2013; Klayman and Ha, 1987). However, when the confirmation bias leads to erroneous conclusions, it leads to suboptimal conclusions and design outcomes in the end.

Confirmation bias in design

Since decision making and information processing are a key components of the design process and the confirmation bias has been identified in nearly all domains it has been studied (Hallihan & Shu, 2013), it is expected that confirmation bias to some extent is also present in the design process. However, only limited research has been describing this bias specifically in the design process. An article of Hallihan et al. (2012) describes an experiment which reveals observations of the confirmation bias in the concept generation phase of the design process. The participants were 30 engineering students from a fourth year design course were split in nine groups and were given a design task. Group discussion were transcribed and qualitatively coded. The results showed that participants were more likely to make statements that confirmed beliefs than to make statements that disconfirmed beliefs: 83% of statements were coded confirming while only 17% were disconfirming. This shows strong evidence towards the tendency to confirm rather than to disconfirm. Some interesting qualitative findings were found as well, which are summarized below:

Ignoring the facts

Belief perseverance in the light of contradicting evidence was observed. Participants showed misinterpretation or ignorance of relevant information. Confirmation bias can lead to design fixation or unwillingness to compromise on design ideas.

Confirming analogies

A biological analogy was provided to inspire participants for solutions to a design problem. Participants used this analogy, but often failed to see that the analogy was inappropriately applied to their solutions.

Seeking validation

Participants were asking affirming questions to validate the strengths of concepts. This leads to suboptimal outcomes, since these questions are not constructive to improve the concept. Questions that address flaws in the concepts would lead to improved concepts.

(18)

17 Confidence

The participant’s perceived confidence regarding their knowledge about the design problem or analogy had an effect on the confirmation bias. High confidence led to more resistance to disconfirming evidence. The authors could link this to other literature of Nickerson (1998) which describes that reliance on confirmation bias has been found to increase confidence. The direction of this relationship is not clear based on these observations.

Disconfirmation

Disconfirmation was observed in moments when participants accepted evidence contradicting a design belief and when identifying potential limitations of own ideas. These disconfirmation instances were mostly associated with perceived lack of confidence in the design belief. Lack of confidence in their ideas leads to self-criticism and hesitation to defend own ideas.

Design criticism

It was observed that participants often were hesitant to criticize ideas of other participants. Reasons for this could include low self-confidence, courtesy, etc. In addition to this, criticism expressed was vague and thus easier to dismiss for the designer. Designer often only recognized a flaw when critiqued multiple times and specifically pointed out. These findings were again linked to other research that has shown that designers may fail to see the shortcomings of their own solutions (Silverman & Mezher, 1992) or fail to attend to the feedback of others (Busby, 1999).

The experiment of Hallihan provides evidence of presence of the confirmation bias in the concept generation phase of the design process. Participants showed strong tendency to confirm existing beliefs instead of disconfirming beliefs. In the next paragraphs, an attempt is made to identify similar phenomena in sensemaking phase of the design process.

Linking the Data/frame theory and the confirmation bias

Looking at the characteristics of the Data/frame theory, large similarities with the concept of confirmation (bias) and disconfirmation can be identified. Table 2 makes a comparison between the two theories described by Klein and Hallihan and highlight the most important similarities. Similarities can be identified between the disconfirmation of a belief, as previously defined, and the act of questioning a frame with the result of reframing the existing frame or creating a new one (reframing cycle). In both cases, the belief or frame is questioned, inconsistencies are detected, disconfirming data is accepted and limitations of the current belief or frame can lead to adaptation of the belief (reframing) or switching to a new belief (seeking a new frame). The same goes for confirmation of a

(19)

18 belief and the preservation cycle. For the confirmation bias theory, it is found that failing to accept disconfirming data can be present. Also, in the Data/frame theory they identified that people tend to preserve a frame by explaining away data that do not match the frame. Both theories also agree, that preservation or confirmation of a frame or belief, can lead to efficient and adaptable strategies, and can make the search for relevant information more efficient.

Since there are large similarities and the available literature on confirmation bias in sensemaking is very low, in the next paragraphs an attempt is made to argue in what extent the theory about confirmation bias can be translated to sensemaking.

Table 2 Literature overview of the confirmation bias in general and in the design process Data/frame theory (Klein et al.) Confirmation bias (Hallihan et al.)

Frame Belief or hypothesis

Preservation of frame: people tend to preserve a frame by explaining away data not matching the frame

Confirmation bias: accept confirming data too easy, misinterpret disconfirming data and fail to accept disconfirming data

Reframing of frame: Frame is questioned, inconsistencies are detected, disconfirming data is accepted, and the frame is adapted (reframing) or a new frame is sought

Disconfirmation of belief or hypothesis: Belief is questioned, inconsistencies are detected, disconfirming data is accepted and the belief is adapted or switched to a new belief

It was found that commitment to a frame enables it to be tested effectively and learn from inadequacies

It can be part of pragmatic and adaptable decision making strategies

Rapid recognition of frame helps to gather information more efficiently

Confirmation bias can provide a framework which makes searching for decision-relevant information more efficient than a random search

(20)

19

Confirmation bias and sensemaking

Literature that describes the confirmation bias in the information synthesis phase is very limited. In this paragraph the goal is to describe literature about the confirmation bias or similar phenomena. Klein et al. (2006) is one of the few articles that discusses the confirmation bias in sensemaking specifically. The authors argue that certain actions that might seem like a confirmation bias, can actually have a different purpose. They say that people do not engage in simple mental processes of confirming or disconfirming a hypothesis. A cognitive task analysis of real-world decision making showed that skilled decision makers shift into an active mode of elaborating a competing frame once they detect the possibility that the current frame is inaccurate. Instead of a confirmation bias, this may actually be an action of using a frame to guide information seeking.

The Data/Frame theory indeed believes that rapid recognition of a frame helps to gather information more efficiently. Also, it was found that commitment to a frame enables it to be tested effectively and adjustment and learning based on found anomalies (Rudolph et al., 2003 (as cited by Klein et al., 2006)). If this indeed is the case, there is no grounded reason to believe that preservation of a frame has a negative effect on the sensemaking outcome. However, the Data/frame theory also points out some seemingly inefficient effects, including attaching too much credibility to inaccurate anchors and explaining disconfirming data away.

In addition to this, large similarities can be indicated between the confirmation bias theory and the Data/frame theory. Research has showed that the confirmation bias has been recognized in decision making processes in almost every field including the design process and can actually have negative effects on information processing. It seems logical that some degree of confirmation bias will also be present in the sensemaking process. In congruence with the Data/Frame theory, the work of Hallihan and colleagues have shown that indeed confirmation bias could lead to more effective information seeking, frame testing and frame adjustment. However, in other cases this bias, by providing too much credibility to confirming data while disregarding disconfirming evidence, it can lead to unjustified decisions.

Based on these findings, the conclusion is drawn that whether preserving a frame has a negative or positive on the outcome of the sensemaking process, is totally dependent on how the individual subsequently acts upon it. When preserving a frame, testing it elaborately and based on judgement of data, adapt or reframe accordingly, this may be an efficient strategy to find the right frame. However, when a frame is preserved and the individual shows confirmation bias, by failing to consider the diagnostic value of supportive data and accept confirming data too easily while misinterpreting and

(21)

20 refusing disconfirming data, this is expected to have a negative effect. It can lead to inaccurate conclusions and suboptimal design outcomes, by not creating a good fit with the external knowledge. However, these expectations are found in other contexts and therefore, research is needed to confirm or disconfirm these expectations. The fact that there is still a clear gap in literature, results in the aim of this thesis being to clarify part of this research gap. This leads to the research question as described before: What effect does the early provision of a frame in the sensemaking process (of the design process) have on the propensity to preserve a frame and the propensity to reframe and what role does confirmation bias play in this? Based on findings in literature five hypotheses were created (displayed in the conceptual framework in figure 2):

H1: Early provision of a frame leads to higher propensity to preserve the given frame

As indicated, in the case of a match between a set of data and a frame being more plausible than the match to any other frame, the first frame is accepted as likely explanation (Klein, 2006). Also, complying to an anchor that comes early in the process has shown to have more impact. When preserving the provided frame, it is expected to lead to hypothesis 2.

H2: Preservation of given the frame leads to lower reframing propensity

When accepting and thus preserving a frame, people tend to explain away data that does not match the frame. When comparing literature about sensemaking and the confirmation bias, similarities could be found. Confirmation bias literature suggests that people are less critical to information that is preference consistent (matching the frame) than preference inconsistent. Also it is found that in the concept generation phase of the design process, people ignored relevant information resulting in believe perseverance. Thus, preservation of a frame is expected to lead to lower reframing propensity when presented with new information.

H3: Preservation of frame is lower when the frame provided is wrong compared to a provided frame that is right.

As indicated in the previous hypotheses, it is expected that early provision of a frame leads to higher confirmation to this frame and in turn lower reframing behavior when presented with new information. However, literature indicated when expectations are violated, this may lead to questioning of the frame. Providing a frame that confirms to general beliefs (here called “right frame”) is more likely to be accepted, than a frame that is the complete opposite of the general beliefs (called

(22)

21 “wrong frame”). When deviating from what is expected, people may start to question the accuracy of the frame and in turn this could learn to lower preservation of the given frame.

H4: Preservation of frame is higher when the provided frame is faulty compared to the situation where no frame is provided.

Although the preservation of the frame may be lower for a “wrong” frame than a “right” frame, still it is expected that preservation of the frame is still present and thus higher than in the case if no frame is provided. This is based on the earlier indicated finding that people can attach too much credibility to an inaccurate anchor and in turn fit other data elements into this inaccuracy, especially early in the sensemaking process. In addition, confirmation bias literature also adds to this by describing that people have shown to fail to accept and misinterpret disconfirming evidence and adjust beliefs based on this, accept confirming evidence too easily and fail to consider the diagnostic value of it (Nickerson, 1998).

Figure 2 Conceptual framework

Provision of frame Reframing propensity

Preservation of frame propensity

+

_-

-

H1 _H2 H3 H4

(23)

22

Data and Methods

Stanford Design Thinking Exercise

The Stanford Design Thinking Exercise (SDTE) is a tool developed to measure reframing behavior (Plattner et al., 2012). SDTE is found to be a robust measurement for reframing behavior on both an individual and group level. The advantage of this method is that it consists of a closed-form assessment tool that eliminates many of confounding variables and that can be easily quantified. The exercise is also convenient since it can be completed in less than 1 hour. The exercise is an episodic case in which the goal is to design an urban public share bicycle, a bicycle suitable for an urban environment and shared by the general public. This subject is chosen because it was expected to be familiar to participants, but not particularly polarizing. The “public shared” part was added because this was expected to be a less familiar concept and in this way stimulate innovative thinking (Kress et al., 2012). In this exercise, translation of a mental frame into observable behavior is achieved by letting participants indicate their ranking of design options. In four subsequent rounds, every round new options are introduced and participants are asked to make a new ranking with their top-10. Based on the amount of change in their ranking throughout the whole task, a reframing score can be calculated.

The experiment

The experiment used is an adapted version of the Stanford Design Thinking Exercise (STDE). Similar as the original exercise, the experiment starts with a description of the case, including the role of the participant in the case (designer in a bicycle-manufacturing firm) and the aim to design an urban public share bicycle. Based on the original SDTE, the experiment is designed to be episodic with a series of rounds where new information is introduced and new choice structuring (or reframing) can be measured. The experiment consists of four rounds. The following steps are taken:

1. Before starting the experiment, a short questionnaire is provided, to gain some general information about the participant. The requested information includes age, gender, nationality, Business Administration track and previous design experience.

2. On the first page of the experiment, a description of the case study and instructions are provided. In the first round (Initial set) 10 initial design choices are presented. The participant is instructed to rank those design choices from 1 (most important) to 10 (least important) on answer sheet 1.

3. The second round (Stimulus set) offers five additional design options. Again, the participant is asked to make a top 10 (of the 10 options from the previous rounds + 5 new options) and rank them on importance. 5 design options have to be excluded.

(24)

23 4. The third round (Final set) again introduces five new design options. Also now, the participant has to make a top 10 (of the 15 options from the previous rounds + 5 new options). This means that 10 design options have to be excluded.

5. The fourth round (Open set) offers the opportunity for the participant to provide their own design options. They are allowed to create five new options. However, they can also create fewer or no option at all. Again, the top 10 (of the 20 options from previous rounds + 0-5 new options) should be chosen and ranked on importance. This means that 10 to 15 options have to be excluded.

The outcome of the exercise are four top-10 rankings of design criteria. Based on a scoring system, the reframing behavior can be quantified.

Qualitative element

The original SDTE is a completely numerical measurement. This means you can only identify the amount of change in the chosen ranking throughout the rounds. To get more insight in the motivation behind the chosen rankings, a qualitative element is included. The aim of this qualitative element is to find out on what motivations and themes their initial frame is based, and how this has changed during the experiment.

The qualitative element consists of the question: “Please explain why you chose for the design of the bicycle (or certain design criteria) you indicated on answer sheet 1” on answer sheet 2 which should be filled in in round 1 after indicating the ranking. This question is repeated at the end of the experiment in round 4.

For the sheets of the whole experiment refer to appendix 1.

Participants

The participant group consists of 30 Business Administration master students at the University of Amsterdam. The experiment is a between-subject design with the participants randomly assigned to a condition.

Measures

Independent variable: Provision of frame

The independent variable “provision of frame” is the stimulation factor. The participants were equally distributed among three conditions. The three conditions are: condition A: no provision of frame; condition B: provision of frame with right information and condition C: provision of frame with wrong

(25)

24 information. The difference in the task between the three conditions is the first page or round 1 of the task. The rest of the task is identical in all conditions.

Condition A

The participants of condition A make the original STDE task (apart from the changes described earlier). This means that on the first page there is a description of the case with a list of ten design options coded A till J. The first page (round 1) is equal to the page of round 1 in the sheets as displayed in appendix 1.

Condition B

In condition B, on the first page again a description of the case and the ten design options are shown. However now the importance of the design options according to team members is indicated. This is virtual and is determined based on the consensus ranking as found in Plattner et al. (2012) (see appendix 2). The consensus ranking is based on 95 respondent who rated each item in order of importance on a five-point Likert scale. The importance according team members is indicated for every design option and visualized by a bar. This creates a frame that confirms to general beliefs (here called “right” frame). For the first page of the experiment of this condition refer to appendix 3.

Condition C

The first page of condition C again consists of a description and the ten design options. Again, the importance of the design options according to team members is shown. However, now the importance based on the consensus ranking is reversed. Thus, in this condition a frame is provided that is opposite to the general beliefs (here called “wrong” frame), by reversing the importance score. Appendix 4 included the first page of the condition C experiment.

Mediator: Preservation of frame propensity

This variable represent the degree to which the participant preserves the given frame. This is measured by the extent to which the chosen ranking in the first round matches to the ranking based on the importance according to team members (the provided frame) in the experiment. This is measured in the same way as the reframing score. Since reframing behavior is the opposite of preservation of a frame, this can be measured by opposite scores. Therefore, the variables that are used for the analysis is: reframing score compared to right frame and reframing score compared to wrong frame. If the reframing score is low, the preservation of a frame is high. In this case, the chosen ranking will be compared to the given frame, and the reframing score will be calculated.

(26)

25 Dependent variable: Reframing propensity

The reframing propensity measures to what extent the participant shows reframing behavior. Reframing behavior is quantified by calculating the reframing score of the whole experiment. This calculation is based on the calculation method as described in the original STDE task (Plattner et al., 2012) and is described later.

Control variables

To control for confounding variables, some general demographic control variables are included: age, gender and nationality. In addition, some indicators of previous experience and background are included: the Business Administration track and previous design experience.

Analysis

Calculating the reframing score

For calculating the reframing score, the calculation method of the original exercise was used (Plattner et al, 2012). When the experiment is completed, answer sheet 1 contains four top-10 rankings (one from every round). Scoring is achieved by counting the amount of change from one round to another. Change is calculated by measuring the number of spaces on the ranking that a design option has moved from one round to another. This means that a design option that has moved from the third place to the 6th_{pace, that there is a change of 3 (no sign). Design options that are introduced as new options,}

are seen as coming from the bottom of the top-10 list. This means that they are treated as they are on the 11th_{place. When scoring them on the second place for example, there is a change of 9. Design}

options that have not been moved or are removed from the list, have a change of 0. To compute the total reframing score, the reframing scores of every round are added. An excel sheet is programmed to automatically calculate the reframing score.

Calculating the preservation of frame propensity

The preservation of given frame is the extent to which the chosen ranking is confirming to the given frame. This could also be described as showing low reframing behavior compared to the given frame. This can be expressed as a reframing score and calculated as described before. For condition B the given frame (“right” frame) was compared to the actual chosen ranking and a reframing score was calculated. For condition C, the chosen ranking was compared to the given “wrong” frame and again a reframing score was calculated. Condition A functioned as a control condition for the other two conditions, comparing the ranking in round one with either the “right” or “wrong” frame.

(27)

26

Statistical analysis strategy

Preliminary steps

Excel and SPSS software was used to analyze the data. All data was checked for missing data with a frequency test. No data was missing. Next descriptive analysis was run to get an overview of the participant characteristics. In addition, a correlation analysis was run. The results are shown in table 3. It shows that gender correlates with Business Administration track with a correlation coefficient of r=-0,41 and p<0,05. Reframing score compared to right frame shows to have a strong correlation with reframing score compared to wrong frame (r=-0,94 and p<0,01).

Table 3 Mean, Standard deviation and Correlations of Study Variables

M SD 1 2 3 4 5 6 7 8 1. Provision of frame 2.00 .83 2. Gender 1.63 .49 -.17 3. Age 25.20 2.61 .10 -.18 4. Nationality 1.87 1.20 .07 -.03 .23 5. Business Administration Track 5.17 2.09 .34 -.41* -.02 -.05 6. Previous design experience 1.43 .51 -.17 .25 -.15 -.07 -.33 7. Reframing score compared

to right frame

22.00 9.73 -.02 .26 .15 -.09 -.11 -.27 8. Reframing score compared

to wrong frame

41.13 8.8 .09 -.29 -.10 .09 ,17 .27 -94** 9. Reframing score total 62.80 20.58 -.04 .22 -.07 -.17 ,19

-.17

.11 -.05

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

Furthermore, the dataset was tested for skewness, kurtosis and normality. Values between -2 and +2 for skewness and kurtosis are considered to be acceptable in order to prove normal univariate distribution (George & Mallery, 2010). Nationality had higher value for skewness (2,87) and kurtosis (11,36). This is most probably related to the fact that most of the participants had either a Dutch or European nationality. Previous design experience had a kurtosis score of -2,06. Reframing score compared to wrong score had a skewness score of -2,00 and a kurtosis score of 4,13. The rest of the variables had a score within the boundaries. However, nationality and previous design experience are only control variables and since participant are randomly assigned to groups this is difficult to manage. Comparing means

First of all, the means of preservation of frame propensity and reframing propensity are compared between the different conditions, to see if there is a difference. For the preservation of frame

(28)

27 propensity, the reframing scores compared to the given frame of condition B and C were both compared to condition A (control condition). Since in condition B and C, different frames are given, these condition can only be compared to the control condition instead of to each other. To compare two group means, the independent T-test was found to be appropriate for the data characteristics (McCrum-Gardner, 2008).

For the reframing propensity, the measured outcome is the same in all three conditions. Thus, the means of all three conditions can be compared to each other. To compare means of one variable across three groups the One way ANOVA was selected as an appropriate tool (McCrum-Gardner, 2008). Testing the whole model

The proposed hypothetical framework is a mediation model. To test the complete model, the PROCESS macro written by Andrew F. Hayes was used. Model 4 is was chosen for the theoretical model in this thesis since it is suited for simple mediation models.

Analysis of qualitative element

The written text of the qualitative element was digitally transcribed. These transcripts were analysed by the qualitative data analysis software NVivo. Based on the mentioned motivations for the chosen rankings, eight categories were identified:

 Comfort: The comfort that the bike provides to the rider

 Marketing and appearance: The way the marketing of the bike is handled and how the bike looks

 Quality and durability: The quality and durability of the bike and the materials used  Safety: The safety and security it provides to the rider

 Sustainability: The sustainable element of the bike and sustainable experience for the rider  Technical functionality: Technical features that determine the cycling functionality

 User focussed: The degree to which the bike is suitable for different users in different situations and the adaptability to different users/situations

 User friendliness: The degree to which the bike is easy to use for the rider

Next, groups of words or sentences of the transcripts were coded to the category in which they belong to with the help of NVivo. The outcome measure is the number of references in each category.

(29)

28

Results

Participant characteristics

The characteristics of the 30 participants are displayed in table 4. As can been seen more female participants than male participants were part of the experiment. However, the ratio male/female is fairly good distributed. Also no major differences in the mean age can be observed between the three conditions. With regards to nationality of the participants, most have a Dutch or European nationality. The distribution between the different Business Administration tracks is not even. The most participants were enrolled in the Entrepreneurship and Innovation track. The amount of participants having previous design experience was slightly higher than participants that did not have design experience so far.

Table 4 Participant characteristics

Condition A Condition B Condition C Total

N 10 10 10 30 Gender Male 3 3 5 11 Female 7 7 5 19 Age (years) 24,9 ± 2,6 25,2 ± 2,5 25,5 ± 3,0 25,2 ± 2,6 Nationality Dutch 5 4 4 13 European 5 2 5 12 Asian 3 1 4 African 1 1 Business Administration track Strategy 2 2 4 Marketing 1 1 2 Leadership and Management 1 1 Entrepreneurship and Innovation 6 6 5 17 Digital Business 2 4 6 Previous design experience Yes 5 5 7 17 No 5 5 3 13

Preservation of given frame

Reframing score compared to the right frame

Only in condition B, the consensus ranking (right frame) was indicated. Then a reframing score was calculated based on how much the ranking in round 1 deviates from the ranking of the right frame. Although in condition A this ranking is not shown, this reframing score was also calculated. In this way condition A functions as a control and enables to indicate the effect of condition B (showing the right frame). An independent T-test was used to test for differences. Table 5 indicates the mean reframing

(30)

29 score compared to the right frame for both conditions. A lower reframing score means higher preservation of the right frame.

Table 5 Mean and standard deviation

Condition N Mean Std. Deviation

A 10 24,00 11,392

B 10 18,40 9,924

As can been seen in table 5, the reframing score in condition B is lower (18,40 ± 9,92) than condition A (24,00 ± 11,39). This means that the preservation of the given frame is higher in condition B. However table 6 indicates that this difference is not large enough to be significant (F= 0,28, Sig 2-tailed= 0,26). Table 6 Independent T-test scores

F T Df ΔSE Sig. Sig. (2-tailed

Reframing score

0,277 1,172 2 4,778 0,605 0,256

Reframing score compared to the wrong frame

In condition C the opposite of the consensus ranking (wrong frame) was shown. Again, a reframing score was calculated based on how much the ranking in round 1 deviates from the ranking of the wrong frame. Also here, condition A was used as a control condition and the same calculation was done. An independent T-test was run to compare the two conditions. Table 7 indicates the mean reframing score compared to the wrong frame for both conditions. A lower reframing score means higher preservation of the wrong frame.

A 10 39,40 10,287

(31)

30 Table 7 shows a minor difference between the two categories with condition A showing a slightly lower reframing score (39,40 ±10,29) than condition C (41,20 ± 5,90). This means slightly lower preservation of the given frame in condition C. However, this difference is very small and as table 8 shows this difference is not significant (F=2,28, Sig 2 tailed=0,64).

Table 8 Independent T-test scores

F T Df ΔSE Sig. Sig. (2-tailed)

Reframing score

2,276 -,480 18 3,751 ,149 ,637

Total reframing score

Next the total reframing score (of round 2, 3 and 4) was compared between the three different conditions. One-way ANOVA was used to test for differences between the conditions. Table 9 shows the means of reframing scores for all conditions.

A 10 65,50 23,950

B 10 59,20 22,953

C 10 63,70 15,535

Total 30 62,80 20,584

Table 9 shows the highest reframing score in condition A (65,50 ± 23,95), followed by condition C (63,70 ± 15,54) and condition B having the lowest reframing score (59,20 ± 22,95). This is in line with the expectations expressed in the hypotheses. However, when looking at table 10 the reframing score did not differ significantly between the different categories (F= 0,24, p=0,79). Comparing these scores with the original SDTE, similar results are found. For the two stimulus sizes that were used in the experiment, the following scores were found: 63,2 ± 11,0 (N=30) and 64,4 ± 13,3 (N=25) (Plattner et al., 2012). Both scores are quite similar to the mean score of the control condition (65,5 ± 23,95).

(32)

31 Table 10 One-way ANOVA scores

Sum of Squares Df Mean Square F Sig.

Reframing score 210,600 2 105,300 ,235 ,792

Error 12076,200 27 447,267

Total 12286,800 29

Test complete model with PROCESS

To test the complete model, the PROCESS application is used. Model 4 is suited for the theoretical model in this thesis since it should be used for simple mediation models (like the theoretical model). Figure 3 shows the model again. The different relationships in the model are indicated with letters (a1,

b1, c1’).

These relationships can then compute different relationships:  Direct effect= c1’

 Indirect effect= a1b1

 Total effect (c1) c1’ + a1b1

The analysis is run two times: one time for the condition with the right frame (Condition A vs. B) and one time for the condition with the wrong frame (Condition A vs. C)

Right frame (Condition A vs. Condition B)

Table 11 shows that none of the relationships (a1,b1, c1’) are significant. Also, the direct, indirect effect

and the total effect in table 12 do not show a significant effect. This means that for the right frame none of the hypotheses are confirmed.

Provision of frame Reframing propensity

Preservation of frame propensity

a

1

b

1

c

1‘

(33)

32 Table 11 Scores of relationships a1, b1 and c1’

Consequent

Preservation of frame Reframing score

Antecedent Coeff. SE P Coeff. SE P

Provision of frame A1 0,0664 2,28 0,977 C1’ -2,482 5,07 0,630 Preservation of frame - - - B1 -0,0567 0,46 0,904 Constant I1 7,715 23,41 0,743 I2 37,367 52,1 0,481 R2= 0,239 R2= 0,199 F(6, 23)= 1,204 , p= 0,340 F(7, 22)=0,779, p=0,612

Table 12 Scores of direct, indirect and total effect

Effect SE P

Direct effect C1’ -2,482 5,07 0,630

Total effect C1 -2,485 4,96 0,621

Boot SE

Indirect effect A1B1 -0,004 1,77

Wrong frame (Condition A vs. Condition B)

Again, none of the relationships (a1,b1, c1’) are significant (table 13). Also, the direct, indirect effect

and the total effect again do not show a significant effect (table 14). This means that for the wrong frame none of the hypotheses are confirmed.

Consequent

Preservation of frame Reframing score

Antecedent Coeff. SE P Coeff. SE P

Provision of frame A1 3,834 4,09 0,366 C1’ -3,119 11,42 0,790 Preservation of frame - - - B1 -0,043 0,75 0,956 constant I1 77,911 25,66 0,0096* I2 76,627 90,60 0,414 R2= 0,351 R2= 0,237 F(6, 13)= 1,169 , p= 0,380 F(7, 12)=0,533, p=0,794

Effect SE P

Direct effect C1’ -3,119 11,42 0,790

Total effect C1 -3,282 10,62 0,762

Boot SE

(34)

33

Qualitative element

In order to get some insights in how participants changes their frame and on what reasons their choices are based, the qualitative element was introduced in round 1 and 4. Participants have written a motivation for the design they have chosen. Based on the motivations, categories of arguments were made: comfort, marketing and appearance, quality and durability, safety, sustainability, technical functionality, user focused and user friendliness.

In appendix 5 there is a complete table which indicates per subject and per round how often a code was mentioned in the text. The condition was indicated by a letter (A, B or C) followed by the participant number and the indication of round 1 or round 4. Safety was mentioned most as an argument for their choice of design, followed by technical functionality reasons and user friendliness purposes.

In the next section, figure 4 will show some charts of the distribution of the number of references per category. The next paragraphs will compare charts of the different conditions and the two rounds to see if some patterns can be found in differences in the reasoning behind their choice.

(35)

34 Total

Round 1 Round 4

All conditions

(36)

35 Condition B

Condition C

Figure 3 Qualitative results

All participants

The first three charts are based on all participants, and show a subset of round 1 and round 4. Some shifts can be observed from round 1 to round 4. User friendliness and comfort show a large increase from round 1 to round 4, while technical functionality and marketing and appearance seem to decrease.

Round 1 of all conditions

Next, the distributions of references in round 1 are compared among all categories. The idea is that effects of a given frame may be also present in the explanation. Condition A has relatively more focus on marketing and appearance and safety. Condition B has a relative large section in the technical functionalities and less focus on the marketing and functionalities as condition A. Looking at the

(Re)framing and the confirmation bias in sensemaking in the design process