Accountability for Digital Dreamers: Patterns of Failed Digitalization Initiatives

To contribute to digitalization and accountability research, this study adopted a pattern arising from failure due to weak accountability that was initially identified in Great Britain. This was done to investigate if the pattern reappeared in digitalization initiatives at the Swedish municipal level. Attempting to answer this, the present study structured a survey sent to every municipality in Sweden, resulting in a response rate of 40.4%. It was not possible to statistically claim that the pattern repeated itself in the chosen context, making this study’s main contribution to stress that there might be a pattern as an effect due to weak accountability, without any knowledge of how this pattern presents itself.


Accountability for Digital Dreamers: Patterns of Failed Digitalization Initiatives
Digitalization initiatives are being implemented worldwide to various extents in pursuit of the value they can bring (Winkler & Zinsmeister, 2019). Digitalization is driven by the emergence of new digital technologies, the aim of which is increased efficiency (Bloomberg, 2018). However, the downside of the potential value is excessive cost and a substantial risk of failure (Orji, 2019;Wade & Shan, 2018). According to the International Data Corporation (IDC), Europe is estimated to spend $1.2 trillion in 2022 on digitalization, with 13% coming from the public sector (IDC, 2022) a fail rate of 87% (Wade & Shan, 2018). A failed digitalization initiative implies a range of problems such as a failure in adherence to the budget or schedule or missed targeted value goals (Almarabeh & Abu Ali, 2010;Anthopoulos et al, 2016;Sundberg, 2019). To combat this, scholars have worked for decades to identify critical success factors (Poon & Wagner, 2001;Zahedi, 1987)-processes that must function and be implemented well to ensure digital success, and which vary between initiatives depending on existing conditions (Winkler & Zinsmeister, 2019).
One critical success factor that has been identified is having a supportive management team that is held accountable (Gunawong & Gao, 2017). According to Gunawong and Gao (2017), this team creates a driving force, ensuring that the values the initiative pursues are realized (Burga & Rezania, requirement that officials answer to stakeholders on the disposal of their powers and duties" (Sharma, 2008). Even though there is a lack of consensus on the term, there is a lot of research implying that accountability should be at heart of public democracies (Almqvist et al., 2013;Guerin et al., 2018;Thomas, 1998). The opposite of good accountability is corruption, which Johnston (2006) defines as "The abuse of public roles or resources for private benefits," which is a definition that has been widely accepted (Ray, 2012). Pina et al. (2007) have identified digitalization as a tool to foster citizens' participation and interest in politics, leading to greater integration with the government. Moreover, digitalization may also enhance the pivotal factors of accountability and transparency, as it has a favorable effect on the link between people, businesses, and information. Ray (2012) divides e-government into two comprehensive categories. The first asserts that the use of digital technologies can enhance activities and services, with an emphasis on customer service (Heeks, 2003), while the second highlights digital adoption to reform the government, making it more transparent, efficient, and accountable (Ray, 2012). Ray (2012) posits that e-government is viewed as a positive approach for governments to generate trust in citizens, create new ways for stakeholders to address challenges, and improve accountability (Clift, 2003), even though the second definition lacks scientific verification (Ray, 2012). Gray & Kaufmann (1998) conducted a study where they surveyed 150 high-level officials, who ranked corruption as the most significant obstacle that impedes their success process. This empirical evidence, along with the definition of corruption from Johnston (2006), the high percentage of failure noted by Wade & Shan (2020), and the amount of money expended on digitalization (IDC, 2022) underscores the importance of understanding the question of public accountability.
The general analytical framework is anchored on four aspects noted in earlier research: accountability, digitalization initiatives, failed digitalization initiatives, and a lack of accountability, as illustrated in Table 1 below. These concepts were operationalized as survey questions in the collection of empirical material and were utilized conceptually, theoretically, and empirically in the design of the study and the analysis of the result.
In accordance with previous studies, Guerin et al. (2018) argued accountability to be a crucial element of democratic governments because it provides citizens with insight into whether elected officials are acting in their best interests. The researchers further suggest that proper implementation of accountability "can increase the trustworthiness and legitimacy of the state in the eyes of the The lack of clarity on who was accountable Q6 Guerin et al. (2018) public," while failing to uphold accountability can increase the risk of failure. According to Guerin et al. (2018), the symptoms of lacking accountability include confusion over who is responsible; inadequate or non-existent consequences for poor performance; lack of transparency; "and, most tellingly, the same pattern of failure occurring repeatedly." The pattern is presented in the study by Guerin et al. (2018) as follows: • When problems or issues surfaced, ministers and civil servants started blaming each other, which gave the appearance that there were no consequences for those involved. • When problems arise, involved actors start pointing toward the historic failures of others, instead of addressing their problems and possibilities. • Even when it is clear who is responsible, it is possible for them to avoid explaining their approach, dodge consequences, and remain in office. • Lacking communication between involved actors with related problems such as different expectations regarding what they were trying to achieve. • Problems with measuring goal achievement.
To contribute to existing knowledge regarding accountability and digitalization initiatives, the present study will utilize the pattern identified by Guerin et al. (2018) as its starting position. The five points will then serve as a guide for empirical data collection and analysis. These points will also be used as an analytic framework which is elucidated, operationalized, and presented in Table 2.

METHoDoLoGy AND EMPIRICAL MATERIAL
For this study, a deductive quantitative research approach was deemed appropriate since the research topic is built upon prior conclusions made in existing literature (Guerin et al., 2018). Furthermore, the empirical material required to replicate this study had to be general in nature and was collected through a national survey. This method was chosen because it allowed for the use of statistical calculations

Evaluate goal achievement
To what extent has the value of the digitalization initiative intended to deliver been achieved?
Q13 Guerin et al. (2018) such as mean and standard deviation to determine the significance of the survey results. No further calculations or analysis was conducted using these values. To gather empirical data, a survey was structured in accordance with the pattern factors outlined above and presented in Table 2. The questions in the survey were divided into two categories, experience-based and opinion-based, as recommended by Gurdur., El-khoury, & Törngren, (2019). The first six questions in the survey, referred to as experience-based questions in this study, were not designed to measure the pattern identified by Guerin et al. (2018), but to contribute to the broader understanding of accountability in a digital environment and shed light on important questions regarding the role of accountability in public sector digitalization initiatives.
The Likert scale used in this study for the answer options regarding the pattern factors consisted of five fixed choices: Yes, on all; Yes, on most; Yes, but only occasionally; No, never; I don't know. A 5-point Likert scale was chosen based on the findings of Kho (2018), arguing that higher scales yield lower-quality data. Each fixed option was then assigned a value ranging from 1 to 4, with "Yes, on all" being assigned the value of 4, "Yes, on most" the value of 3, and so on, until "No, never," which was assigned the value of 1. The answer "I don't know" was not assigned any value, as respondents choosing this option did not take a stand toward the questions. The resulting scores were then used to calculate the mean and standard deviation of the responses.
To determine statistical significance at a 95% confidence level, the value of the mean subtracted by two standard deviations must result in a value above 2.5. In contrast, if only one standard deviation is used, the resulting equation would indicate a tendency toward what the question is asking ( MIT News, 2012).
Further, including an option that allowed the respondents to not take a stand on the survey question was based on the idea that it would yield a higher response rate. The questionnaire also used two open follow-up questions to the pattern dimensions in question number 9 and 11 intending to generate qualitative data. These two were assigned a follow-up question because these dimensions were most recurring in previous literature. The purpose of the open-ended questions was further to investigate whether the respondents felt that the content of the question was something that would change future conditions (see Q9 and Q10) or whether it led to the digitalization initiatives failing (see Q11 and Q12). Since research suggests that many digitalization initiatives fail (Wade & Shan, 2018), this was considered a good opportunity to provide a deeper insight into whether weak accountability could be a contributing factor. The respondents were kept anonymous to maintain an environment of lower social anxiety and higher self-esteem as reports indicated it would (Joinson, 1999). When the survey was finished, it was emailed to all Swedish municipalities with a presentation of the subject, information about anonymity, and the level of desire that the respondent sought a business development manager or similar. The categorization of the respondents' positions was later carried out using MIS: Standard for Swedish occupational classification (SCB: MIS 2012:1 SSYK2012). The email addresses used were found in a compilation of contact details carried out by SKR (Swedish Association of Local Authorities and Regions). The survey was sent mid-March 2022, and two reminders followed the same week, which gathered a response rate of 40.3%, or 117 out of 290 municipalities.
The only conclusion that could be drawn regarding the nonresponses was that some of the municipalities do not respond to surveys unless they originate from other municipalities or authorities. This conclusion was based on 17 responses informing us about this policy. Even though only 17 such responses were received, an assessment was made that more municipalities might have chosen not to respond for the same reason. Since the survey was anonymous it is difficult to draw more conclusions about the response loss without speculation.
The survey mostly used closed response options with a five-point Likert scale. Our limitation to only include different scales of "yes" as answers, instead of the traditional Yes, Neutral, No approach (Kho, 2018), was based on the purpose of this study. We wanted to examine to what extent the pattern was repeated, which means that different degrees of the answer "no," which the traditional approach advocates, fall outside of the present study´s purpose.
A shortcoming of the Likert scale is that it opens up for interpretation and allows the respondents to answer according to how they define, for example, "most" or "occasional," which could differ from respondent to respondent. The terms "most" and "occasional" were also judged to vary depending on how many digitalization initiatives one has participated in-occasional in number is judged to be more for a respondent who has participated in numerous digitalization initiatives compared to a respondent who has only participated in a few. However, investigating the extent to which the respondents experienced the pattern was not assessed as a factor affecting the validity of the method. The answer given by the respondents is how they experienced what the survey asked for, which was the purpose. It can also be argued that the collected data has higher reliability as a result of including the answer "I don't know," since it prevents respondents from guessing in the absence of knowledge or experience within that question. However, it also allows the respondents to bypass the question if they do not want, or have the time, to answer.
When compiling the respondents' answers to survey questions 1 and 2, there were some unexpected complications. Question 1 allowed the respondents to indicate their position within the municipality, but despite requesting a business developer or similar to answer the survey, only 15% of the respondents indicated this position. According to the Standard for Swedish occupational classification (SSYK), however, business developers and business managers have work areas encompassing similar tasks, which means that this is unlikely to invalidate the result. Also worth mentioning is that there is no clear academic path to becoming, for example, a business developer or a digital strategist, which means that the variation among the professions listed does not necessarily mean a lack of knowledge within the requested subject. An overall analysis of the results showed that all questions about the pattern were experienced by the respondents to a varying extent; although, in most cases they indicated that they experienced them to a lesser extent.
Since a variety of ways was given regarding the question asking the respondents how long they had worked at their position, the question could have been worded better. Some responses were, in years, or months, or indicated approximately when they started, for example, "since 2017". The latter example is open to subjective assessment, creating an imprecise, possibly inaccurate, number for how long the respondents worked in their position.
Some of the questions can be considered leading questions, which potentially makes parts of the result exposed to bias. Therefore, it is possible that the survey design and the wording of the questions could have been improved. We claim that it is difficult to study a pattern completely avoiding leading questions.

RESULT AND ANALySIS
Most of the respondents were digital strategists, IT managers, other managers, or business developers. The survey also received responses from economists and a small number who did not want to state their position. The average time of employment was 4.5 years, with an equal-sized standard deviation, and answers ranged between 3 months and 26 years. The next result was gathered from information about how many digital initiatives the respondents had participated in. This question had fixed answer options resulting in "30 or more" being the most answered option with 32.5%. Only 1.5% of the respondents had not participated in a digital initiative.
The next survey question asked the respondents if it had been clear who was accountable during these digital initiatives. The question resulted in 12.8% on "Yes, in all," 75.2% on "Yes, in most of them," and only 0.9% on "No, never." The next question asked the respondents to indicate how many of these digital initiatives they considered a failure. This question included a definition of "failed digital initiative" (see Table 1, definition number 3), to minimize variation due to interpretation of the term. The responses showed that almost 24% considered 0 of the digital initiatives they had participated in to be considered a failure. However, 3% considered 50 of the digital initiatives they had participated in as failed scenarios.
The next question in the questionnaire asked if the respondents had experienced problems due to weak accountability in digital initiatives. The results were that 2% answered "Yes, on all"; 28% answered "Yes, on most"; 60% answered "Yes, but only occasionally"; and 8.5% answered "No, never" (see Figure 1). The survey results are particularly noteworthy in this regard because most of the respondents were aware of who was accountable during most of their initiatives. But from this question, it seems like problems arise because of weak accountability, even though the respondents knew who was accountable.
The respondents were then asked if they had experienced if the person accountable blamed others for problems. This question generated 13% responding "Yes, on most"; 55% "Yes, but only occasionally"; 22% "No, never"; and 10% responding "I don´t know." This question resulted in a mean of 1.89 with a standard deviation of 0.63.
The following question asked if the respondents had experienced that the accountable referred to previous failures of others, rather than taking hold of their setbacks. To this question "Yes, but only occasionally" and "No, never" resulted in a 41.9% and 43.5% response rate, respectively. This question resulted in a mean of 1.61 with a standard deviation of 0.53. The next question asked if the respondents had experienced the person responsible avoiding consequences in cases of failed digitalization initiatives (see Figure 3). This question also had an open-ended question, asking the respondents who said "Yes" to further explain how this was done. This question resulted in two small groups answering "Yes, on all" and "Yes, on the most" with 2.6% and 5.1%, respectively. The question resulted in 24% answering "Yes, but only occasionally" and 41% answering "No, never." The mean calculation resulted in a value of 1.58 with a standard deviation of 0.77. Interestingly, this question also collected 27% answering to "I do not know," which was significantly higher than the same answer to other questions.
According to the responses to the open question that followed (presented in Figure 3), avoidance was mostly done by blaming a colleague or another actor involved (see quotes 1-4). Another approach mentioned by the respondents was that the accountable handed over the responsibility to someone else, a different colleague or department (see quotes 5-7): 1. "The supplier is usually blamed." 2. "Usually by blaming a lack of resources or other areas of focus." 3. "Ducking and pointing at someone else." 4. "Not being accountable for the result but blaming others." 5. "Left responsibility to a new employee." 6. "Moved responsibility to another person or department." 7. Handed over responsibility to a colleague and started another project." While other respondents stated that dodging consequences was done by simply moving on (see quotes 8 and 9), which according to the respondents might be easy since both the public sector and digitalization initiatives work that way (see quotes 10-12).
8. "You may not have learned from the failed project to be better in the future, but rather just moved on to the next project." 9. "Continues to work as before and refuses to take the cost of the initiative." 10. "In the public sector, problems/failures are swept under the rug." 11. "There are not immediately any consequences linked to failed digital ventures." 12. "Digitalization initiatives are rarely questioned, regardless of whether they are successful or not.
Thus, it is quite easy to avoid consequences." Most respondents pointed to the fact that those responsible were not held accountable or asked to explain their approach, but simply continued with the next project. Our interpretation is that the respondents had experienced the pattern described by McCrae et al. (2018) in a different area than where it was initially identified. The data collection method becomes a major limitation, interviews, for example, could have given more details on how those responsible avoided consequences.
The next question was about whether poor communication was a problem the respondents had experienced and included an open-ended question on what problems this had caused. This resulted in 2.5% answering "Yes, on all"; 25% answering "Yes, on most"; 56% answering "Yes, but only occasionally"; and 13% answering "No, never." This question resulted in a mean of 2.15 with a standard deviation of 0.73.
According to the respondents, the most common side effect was that the initiative failed in some way, which can be seen in the example quotes below. The respondents also believed that it made collaboration more difficult, and that understanding how and why digitalization should be carried out was not understood (see quotes 13-15).
13. "Delays, additional costs, and "worse" solutions as a result." 14. "Does not keep the schedule, does not deliver the desired result, or delivers the project more expensive than expected." 15. "It has happened that when the project has not succeeded all the way, the expected benefits/ effects of a project have not been achieved as intended." At this stage, two assumptions were made. Several respondents mentioned problems with expectations as a consequential problem when communication breaks down, which we assumed to be expectations from the customer's perspective. The assumption is only based on quote 16 because all other respondents who stated the follow-up problem had more general answers (see quote 17). The wording in quote 16 was interpreted as the expectations being from the customer´s side and not internal expectations; for example, expectations about who should do what, which meant that we treated the response as a failed digitalization initiative with support from the definition of "failed digitalization initiatives" (see Table 1, definition 3). Since most respondents who stated the consequential problem did not state anything that indicates whether they mean internal or external, an assumption was required to keep the compilation concrete.
16. "Can lead to delays or disappointment as delivery and expectations are not in balance." 17. "For example, different expectations." Another more comprehensive survey result indicated that being accountable becomes more challenging in the realm of communication. The respondent expressed their belief that the problem lies not only in the lack of accountability but also in the aspiration to assume responsibility coupled with inadequate knowledge. Additionally, the respondent argued that competence is outweighed by seeking assistance from others while making decisions. This becomes problematic when the question of accountability arises, since the person in charge did not make the decision himself and thus does not view themselves as accountable (see quote 18). This finding is noteworthy because it presents an opportunity for further inquiry, emphasizing a need for continued investigations regarding accountability in the context of knowledge about digitalization initiatives despite deviating from the central inquiry pertaining to communication.
18. "I feel that it is above all-around communication that the accountability part becomes difficultthose who do not have the skills, but have the responsibility, would like someone else to decide for them in matters they do not feel they have mastered, which relates to that. It then becomes difficult to pursue the issue of accountability if they then do not feel they were the ones who made the decision." The following quote deviates from Guerin et al. (2018) stance toward accountability. Nonetheless, it is intriguing since it is in line with the finding highlighted by Plesner and Justesen (2022) concerning accountability in the public sector. In addition, the same survey question elicited another response that corresponded with the Plesner and Justesen (2022) findings. This respondent likewise opined that the issue at hand is not the absence of accountability but the absence of knowledge regarding digitalization.
19. "I think the questions asked are a little wrong. Development is usually agile. You test, evaluate, reevaluate, and change and then test again. I do not feel that a lack of accountability is the problem, but a lack of knowledge about digitalization. You simply do not know what you are expected to be accountable for." The last question asked the respondents if they had experienced difficulties in evaluating goal achievement. The question resulted in a response rate of 39.3% and 41% on the answers "Yes, on most" and "Yes, but only occasionally" respectively. This question also received an equal percentage of answers on "Yes, on all" and "No, never" This question resulted in a mean of 2.35 with a standard deviation of 0.77 making this the question the respondents experienced to the highest degree. This was somewhat expected since a lot of literature points toward difficulties with measuring the degree of success in digitalization initiatives (see, e.g., Griffin & Haplin, 2005;Plesner & Justesen, 2022).

CoNCLUSIoN AND CoNTRIBUTIoN
The study aimed to explore mechanisms of accountability in digitalization initiatives by reproducing a study made by Guerin et al. (2018) in a Swedish setting on a municipal level. Based on the results regarding the pattern, it appears all dimensions are something the respondents have experienced, even if it is mostly on a smaller scale because the mid-point value was not reached. Therefore, based on the results and statistical significance, the repetition is insufficient to draw general conclusions about whether any pattern dimensions are repeated in the context of the present study. However, interestingly, the patterns of blaming someone else when digitalization initiatives fail, and the lack of follow-up and consequences, also appeared in this study indicating that digitalization initiatives might be better served by closer investigation throughout. This is an aspect of transparency and trust in public digitalization initiatives. Alongside the large financial investments with tax funds made in digitalization, the demand for responsibility increases, which could increase the importance of a deeper understanding of accountability mechanisms to meet the requirements for transparency.
Thus, the weak tendency to problems based on weak accountability means that the study's main contribution is to claim that a pattern because of weak accountability might exist without any deeper knowledge of what it looks like. Future studies could, therefore, use a more qualitative approach, and examine aspects, stories, and details giving a richer insight into how this is done and maintained. Staying on the topic of problems occurring because of weak accountability: even though the respondents mostly thought it was clear who was accountable, it would also be interesting to further investigate the audit logic in the public sector. The purpose of a municipal audit is to evaluate how resource-effective an initiative is but also look into the higher goal completion concerning official policies stating goals and promises, and digitalization should not be an exception.
Researchers could also investigate the potential gap between digitalization knowledge and accountability within. In line with Plesner and Justesen (2022) and the result gathered from this work (see quote 19), respondents seem to think that accountability might not be the main problem. Rather, the problem might arise when the person responsible does not have the necessary knowledge of digitalization, which forces them to rely on others when making decisions, which obstructs the question of who is accountable.