Don’t let perfect be the enemy of good: how perfectionism influences human-centred designing engagement and communal design production in civil engineering

The products and processes designed by civil engineers impact society in a variety of direct (e.g., provision of services) and indirect (e.g., impact from pollution, connectivity of communities, and impact on overall health) ways. In recent years, there have been an increasing number of calls for engineers and engineering students to gain a greater understanding and awareness of the social impact and social value of engineering design (AHEP 4 2020; ICE Community blog 2021; ICE usefulprojects 2020; Lawlor 2016; UK-SPEC 4 2020).

The consequences of not valuing social concerns could lead to not valuing and taking into account the opinions and needs of stakeholders in the community, which could negatively impact the design and the community designed for. Therefore, it is of concern that a contemporary body of evidence indicates that engineers devalue social concerns over the course of their studies and careers (Bielefeldt 2018; Bielefeldt and Canney 2016; Cech 2014). It is thus of interest to the field of Civil Engineering and broader society to understand what drives this devaluing of social concerns that lie in opposition to the stated desire of the sector to improve engagement with social responsibility.

An example of this can be found in Fitton and Moncaste’s (2019) study, where three detailed case studies of flood alleviation projects in England, UK, were examined. Commenting on the data they obtained from speaking with engineers and community members involved in these case studies, they highlighted a seemingly authoritarian engineering manner of decision-making. They point out an engineer’s apparent reluctance to decentralise the decision-making power over to the local community members in one of the case studies, and as a result of that, the community was adversely affected.

Zamojska and Próchniak (2017) stated that social costs and benefits from impact remain challenging to predict and quantify, and may be perceived differently by authorities, decision-makers and project developers. They, therefore, emphasise the differential role engineers’ individualistic characteristics and attitudes may play in the design judgement and decision-making process.

Gajanayake et al. (2021) designed a decision-making aiding tool that assesses the social, economic, and environmental expenses for a disaster-induced bridge failure reconstruction project. They found that 25–30% of the total impacts of the bridge failure account for socio-economic and environmental impact. They suggest including a comprehensive range of sustainability impacts in the decision-making phase during reconstruction, as they warn that only considering economic ones would lead to suboptimal social outcomes.

Reasons for the tension between the stated aims of the sector to meet societal needs and the evidence that that does not always happen, appear to have multiple contributing factors. One factor could be the lack of training engineers and engineering students receive in working and engaging with such public welfare considerate initiatives (as criticised by Cech (2014) and others). Another factor, we argue, can be due to the engineers’ innate drive to work and engage with such initiatives.

This paper argues that perfectionism (i.e., “the tendency to believe there is a perfect solution to every problem, that doing something perfectly (i.e., mistake-free) is not only possible, but also necessary, and that even minor mistakes will have serious consequences” (OCCWG 1997)) and positivism (i.e., the tendency to “take a realist position and assume that a single, objective reality exists independently of what individuals perceive” (Hudson and Ozanne 1988, p. 509; see also Bagozzi 1980; Burrell and Morgan 1979; Morgan and Smiricich 1980), resonate in the manner of problem-solving—they both pursue the single, unflawed, correct solution.

Given that engineering paradigms are predominantly positivistic in nature (Downey and Lucena 2003; Erden 2003), and the alignment with perfectionism, this paper hypothesises that engineering students may be perfectionists, as opposed to non-perfectionists, and proceeds to test for it. This paper also tests for the association of perfectionism to empathy and prosocialness (i.e., characteristics that are thought to be positively associated with human-centred designing ethos), and subsequent production of what we term, Communal Designs.

Communal Designs are considered a specific form of human-centred, human-need-based design, characterised by the particular attention to needs that involve and encourage end-users’ communal engagement and interaction, sense of ‘togetherness’, and social identity. It is, therefore, the result and manifestation of the empathy, social consciousness, and communal values present in engineers or engineering students.

In the current study, Communal Designs are characterised by the consideration of selected interaction-orientated human needs from Max-Neef et al.’s Matrix of Human Needs and Satisfiers (1991, pp. 32–33; see Table 1). If a design considered and addressed peoples’ needs but did not address the interaction needs in Table 1, it was discounted as a Communal Design. Communal Designs are thus inclusive of both metaphysical human needs as well as their physical ones (e.g., shelter and sanitation); see (Al Kakoun et al. 2021a, b, p. 13) for more information. The concept of Communal Designs aligns with the calls proposed by the Institute of Civil Engineers (ICE Community blog 2021) and the UK Government (HM Government, Department for Digital, Culture, Media and Sport 2018, pp. 36–45) for implementing strategies to ‘design out loneliness’ and achieve ‘a connected society’, respectively. Declaration of Communal Designs will be further addressed in the Methodology section.

Social desirability, i.e., “the need of Ss [subjects] to obtain approval by responding in a culturally appropriate and acceptable manner” (Crowne and Marlowe 1960, p. 353), will also be tested in civil engineering students. This is to observe the engineering students’ intrinsic, as opposed to extrinsic, motives for the responses and designs they present; and also, to reason if an intention-behaviour gap arises, shall it arise. Al Kakoun et al. (2021a, b) addressed an apparent dissonance, or intention-behaviour gap, surfaced in civil engineering students who perhaps intended to, but then failed to produce Communal Designs as part of a human-centred designing assignment.

Adaptive perfectionists were found to have better career decision-making self-efficacy compared to maladaptive perfectionists and non-perfectionists (Ganske and Ashby 2011), and socially prescribed perfectionism was found to be associated with inferior problem-solving orientation (Flett et al. 1996). Maladaptive perfectionism positively associated with cognitive test anxiety, and thus negatively with academic performances in university students (Eum and Rice 2011); whilst in STEM disciplines, “low-stressed adaptive perfectionists followed by moderately stressed maladaptive perfectionists” were found to have relatively higher GPA scores than their peers (Rice et al. 2015).

With regard to how perfectionism manifests in the social context, perfectionists (both adaptive and maladpative) were found to have higher social perspective-taking scores (i.e., “the ability to judge a situation from the perspective of another person”) compared to non-perfectionists (Gilman et al. 2014). Adaptive perfectionists, followed by maladaptive perfectionists and non-perfectionists, were also found to have higher positive interpersonal relationships (Gilman et al. 2010). Gilman et al. (2010) further state that “peers rated both perfectionism groups as more prosocial and less disruptive than nonperfectionists”, and that “adaptive perfectionists were more liked than maladaptive perfectionists”.

On the other hand, Hewitt et al. (2006; 2017) address a positive association between perfectionism and interpersonal hostility and social disconnection; although this was critiqued by Stoeber et al. (2017), when found that only other-oriented and socially prescribed perfectionism showed associations with signs of hostility and social disconnection, and that self-oriented perfectionism contrastingly showed associations with signs of social connection and low hostility (particularly regarding physical aggression and spitefulness). Moreover, higher levels of aggression behavior (i.e., “anger, hostility, physical aggression, and verbal aggression”) were found reported by adolescent maladaptive perfectionists, compared to non-perfectionists and adaptive perfectionists, respectively (Ruiz-Esteban et al. 2021).

On another note, Louis and Kumar (2016) found that there exists a “significant number of maladaptive perfectionists” in engineering, and that “they [maladaptive perfectionists] experienced higher levels of personal and societal demands leading to a negative emotional well-being in comparison to the adaptive perfectionists”. Additionally, it has been regarded that perfectionism was rather encouraged in engineering education, as “50% good enough will not be acceptable in industry” (Guzzomi et al. 2015).

The present methodology has been adopted from (Al Kakoun et al. 2021a, 2021b), with slight modifications on the medium of delivery and duration of the human-centred designing assignment workshop. Note that this paper does not cover the priming methodologies used in (Al Kakoun et al. 2021a, 2021b).

A total of 145 survey responses were collected in Phase I, and a total of 184 designs followed by a total of 179 survey responses were collected at the end of Phase II of the workshop (see Fig. 1). Note that as responding to the questionnaires was made voluntary, the response rates could not have been controlled nor paired (with other questionnaire responses, or the number of designs produced); therefore, matching the responses collected in Phase I to those collected in Phase II and the designs produced, has reduced the number of counts involved in the statistical tests computed.

Cut-off points used to classify perfectionists (maladaptive and adaptive) from non-perfectionist in Phase I (see Fig. 1), were adopted from (Gilman et al. 2010). The cut-off points were: Standard Scale score ≥ 37 to be declared a perfectionists (as opposed to a non-perfectionists), along with (if declared perfectionist) a Discrepancy Scale score ≥ 42 to be declared a maladaptive perfectionist (as opposed to an adaptive one).

Pearson chi-square tests were computed to test for the statistical significance of the relationships between the categories addressed (Fisher 1922; Pearson 1900); similarly, Fisher exact tests will be used to assess for the significance of the relationships between categories whenever there is a categorical count less than 5 (Fisher 1922). Statistical analyses resulting in p-values \(<\)0.05 are considered statistically significant, whilst those resulting in p-values \(<\)0.1 can be argued to be ‘tending-to-be significant’ (see Andrade 2019; Thiese et al. 2016; Benjamin et al. 2018) for more information). Moreover, note that in this study, results addressing ‘higher-than-average’ and ‘lower-than-average’ scores imply that the scores in the discussion are higher than the average score of the group involved, or lower than it, respectively.

A factor analysis followed by a binary logistic regression was computed to assess how the personal characteristics addressed (like perfectionism, empathy, prosocialness, and social desirability) simultaneously contribute to, or associate with the probability of Communal Design production (as opposed to how they individually associate with Communal Design production—like what will be observed as a result of the chi-squared tests). Results are displayed in the same order as the research questions proposed. Result tables referred to in this section can be found in the appendix.

RQ1. How common is perfectionism amongst civil engineering students?

74.5% of civil engineering students were found to be categorised as perfectionists—with 68.5% of the perfectionists being maladaptive, and only 31.5% being adaptive (see Table 2). These results indicate that the majority of civil engineering students indeed tend to hold perfectionistic traits, thus supporting our first hypothesis proposed (H1).

For further efficiency and simplicity of result display and analysis, results thereafter will be addressed with regard to a ‘Non-Perfectionist’ as opposed to a ‘Perfectionist’ category—thus, maladaptive and adaptive perfectionists' results will be combined into one ‘Perfectionists’ Category.

RQ2. How does perfectionism associated with Communal Design Production?

Via computing a Pearson chi-square test, we tested the relationship between perfectionism and the production of Communal Designs and found it to be statistically significant (X² (1, N = 145) = 7.767, p = 0.005); perfectionists were found less likely than non-perfectionists to produce Communal Designs (see Table 3). These results thus support the latter part of the second hypothesis (H2-ii).

RQ3. How does perfectionism associate with other characteristics (like prosocialness and empathy) that are known to be positively associated with human-centred designing engagement and thus Communal Design production?

Via computing a Fisher exact test, we tested the relationship between perfectionism and prosocialness scores during the students’ production of Communal Designs, and found it to be statistically significant(p < 0.05); perfectionists were found more likely than non-perfectionists to produce Communal Designs whilst having higher-than-average ‘Prosocialness’ scores (see Table 4).

Similarly, via computing Fisher exact tests, we tested the relationship between perfectionism and the different facets of empathy; it was found that perfectionists were more likely than non-perfectionists to produce Communal Designs whilst having higher-than-average ‘Empathy: Empathic Concern’ scores (p < 0.05), and ‘Empathy: Fantasy’ scores (p < 0.05) (see Tables 5 and 6, respectively). No significant associations were found between the subjects’ perfectionism, Communal Design production, and Empathy: Perspective Taking scores (see Table 7) nor Empathy: Personal Distress scores (see Table 8).

These results are interesting, as they defy the first part of the second hypothesis (i.e., H2-i); contrary to what was hypothesised, it was found that perfectionists were more likely than non-perfectionists to show signs of engagement (i.e. showed higher empathy and prosocialness scores) with the human-centred designing assignment.

RQ4. How does Communal Design production and perfectionism associate with Social Desirability scores—thus, with the intrinsic or extrinsic motives of the engineering students for the design?

Via computing Pearson chi-square tests, we tested the relationship between perfectionism and social desirability scores during the students’ production of Communal Designs and found it to be not statistically significant (X² (1, N = 49) = 2.119, p = 0.145) (see Table 9). However, when tested for the relationship between engineers’ perfectionism and social desirability in general, it was found tending-to-be statistically significant (X² (1, N = 142) = 3.497, p = 0.061), perfectionists were tending-to-be more likely than non-perfectionists to have higher-than-average ‘Social Desirability’ scores in general (see Table 10); i.e., perfectionists were more likely to have responded in a manner that was thought to be socially appropriate, indicating their extrinsic drive for the responses and designs they proposed, as opposed to their true, intrinsic motives.

Finally, to observe how personal characteristics collectively associate with Communal Design production, a factor analysis followed by a binary logistic regression was computed. A factor analysis has been computed to test for the suitability of the data collected and to thereafter perform a logistical regression model that predicts students’ production of Communal Design based upon their personal characteristics’ scores (i.e., predictors’) collective and simultaneous influence on the output.

A factor analysis was conducted on the 12 items, or characteristics, (shown in Table 11) with no rotation. The Kasier-Meyer-Olkin measure verified the sampling adequacy for the analysis, KMO = 0.730 (i.e., ‘middling’ according to Kaiser and Rice (1974)); see Table 12. Moreover, Bartlett’s Test of Sphericity shows significance (i.e., p < 0.001), indicating that the variables (characteristics addressed) are related and are therefore suitable for further structure detection (Bartlett 1954), or in other words, are suitable to be entered into a binary logistic regression model; see Table 12.

A binary logistic model was developed using the aforementioned 12 characteristics as predictors of Communal Design production. Table 13 shows the number of cases included in the regression is 142 out of 184 (i.e., 42 are missing cases due to the present limitation on matching the responses of Phase I to those of Phase II). The first classification table (i.e., Block 0, null table) shows that before entering the predictors (i.e., the personal characteristics’ scores) into the model to predict the probability of producing Communal Designs, the model was able to predict 65.5% of the outputs (see Table 14). However, after entering the predictors, the model was shown to have then been able to predict 71.8% of the outcomes (see Table 15), indicating that the predictive ability was enhanced after entering these variables, or personal characteristics’ scores. This was also seen in the Contingency Table for Hosmer and Lemeshow Test (Table 17), where after entering the variables, out of the 12 observed Communal Designs produced, the model successfully predicted 9 of them—the Hosmer and Lemeshow Test showed statistical non-significance (p = 0.626) indicating acceptable predictability (see Table 16).

Block 1 Variables in the Equation table (see Table 19) show the different characteristics’ ‘slopes’, i.e.., magnitudes (via observing the coefficients’ value, B) and direction (either positive or negative), contributing to the regression line predictive of the probability of Communal Design production. These coefficients can be assembled in an equation that helps predict the probability of Communal Design production (coded 1), or not (coded 0), based on the characteristic scores entered.

Equation 2, assembled based on the generic Eq. 1:

Equation 1—Generic equation predictive of probability of (y) output

(see Field 2018; Sreejesh et al. 2014; Sweet and Grace-Martin 2003) for more information), and the current case study data, is the following:

where \({\beta }_{0}\) is the constant; and the \({\beta }_{x}{\prime}s\) are the values displayed in the ‘B’ column in Table 19; i.e., \({X}_{1}\)=APS-R Standard Scale score, \({X}_{2}\)= APS-R Order Scale score, \({X}_{3}\)= APS-R Discrepancy Scale score, \({X}_{4}\)= Prosocialness Scale score, \({X}_{5}\)= Empathy: Empathic Concern score, \({X}_{6}\)= Empathy: Fantasy score, \({X}_{7}\)= Empathy: Perspective Taking score, \({X}_{8}\)= Empathy: Personal Distress score, \({X}_{9}\)= Self Oriented Perfectionism score, \({X}_{10}\)= Other-Oriented Perfectionism score, \({X}_{11}\)= Socially Prescribed Perfectionism score, and \({X}_{12}\)= Social Desirability score. It is seen that Empathy: Empathic Concern (\({X}_{5})\), followed by Prosocialness (\({X}_{4}\)), were the top two influential characteristics on the probability of Communal Design production in the present study—Empathy: Emapthic Concern contributing with a negative influence, whereas Prosocialness contributing with a positive influence (see Eq. 2).

However, it was found that none of the characteristics had a significant p-value, indicating that the null hypothesis could not have been rejected, and that the characteristics do not have a significant effect on the probability of Communal Design production. This could also be observed under the ‘95% C.I.for EXP(B)’ columns in Table 19, where all confidence intervals cross the value of 1, indicating that the data values entered into the model are equally suggestive of improving or decreasing the probability of Communal Design production (see (Field 2018, p.904) for more information). Moreover, the Nagelkerke R² value indicated that only 13.1% of the variance in the production of Communal Design probability can be explained by variances in the 12 predictive characteristics entered into the model (see Table 18), indicating a low fitness of the model.

Present results revealed that the majority (74.5%) of civil engineering students classify as perfectionist, with 68.5% of these perfectionists being maladaptive, i.e., the ‘unhealthy’ form of perfectionists (see Table 2). The first hypothesis (H1: engineering students are likely to be perfectionists, rather than non-perfectionists) is therefore fulfilled. This finding supports Louis and Kumar’s (2016) indication of the presence of a “significant number of maladaptive perfectionists” in engineering.

Results also indicated that engineering students with higher perfectionistic scores, or being declared as perfectionists as opposed to non-perfectionist, were less likely to produce Communal Designs (see Table 3 These findings therefore support the second part of the second hypothesis (H2-ii: perfectionists are less likely to produce Communal Designs). These results were expected, as the literature indicated that perfectionism hinders creativity (Goulet-Pelletier et al. 2021), which is a prime aspect of design thinking and human-centred designing. It was therefore expected that perfectionistic engineers were less likely to ‘fully’ engage with the human-centred designing task at hand, and as a consequence, less likely to produce the Communal Designs.

However, what was not expected, was the first part of the second hypothesis (H2-i: perfectionists to be less likely than non-perfectionists to ‘fully’ engage with the human-centred designing assignment) to be overruled, especially when the resulting product, or the second part of the hypothesis (H2-ii), was fulfilled.

Present findings reveal that although the perfectionist were found less likely to produce Communal Design (fulfilling H2-ii), they were found more likely to have higher scores of prosocialness, empathy, empathy: empathic concern and empathy: fantasy when producing the Communal Designs (compared to the non-perfectionists) (see Table 4, Table 5 and Table 6, respectively). This not only defies the first part of the second hypothesis (H2-i) but also indicates that perfectionists were more likely to show signs of more engagement (via engaging more empathically and prosocially) with the human-centred design initiative. Supporting these surprising findings, are the findings of Gilman et al. (2014) and Gilman et al. (2010) on perfectionists having higher social perspective taking (which is a form of empathy (Davis 1983)) and are perceived as more prosocial amongst their peers, respectively. However, contradicting these findings, are of Hewitt et al. (2006; 2017) and Ruiz-Esteban et al. (2021) on perfectionists’ positive associations with interpersonal hostility and social disconnection, and aggression behavior, respectively.

This study interprets this dissonance in perfectionists being less likely to produce Communal Designs, but simultaneously more likely to ‘fully’ engage with the design initiative, hinting towards the findings of Al Kakoun et al. (2021a). Al Kakoun et al. (2021b) addressed an apparent dissonance, or intention-behaviour gap, in which civil engineers may intend to, but fail to produce Communal Designs. To contextualise, when referring to intention in this study, it is interpreted as the intention to produce Communal Design (i.e., by showing higher, intuitive engagement with the human-centred designing initiative), and behavior as the actual/eventual production of the Communal Design. In explaining the Intention-Behavior Gap, Sheeran and Webb (2016) indicated that people may refrain from acting upon an intention not because they lack motivation or value for it, but rather because they may lack the technical competencies required for acting upon these intentions, motivations and values. This resonates with the discussion of Cech and Sherick (2015) on how the depoliticised nature of engineering education and culture may be hindering engineers’ interest and capacity for work and engagement with human-centred, public-welfare-considerate initiatives in engineering—but are still expected to adequately perform their duties consisting them, nonetheless!

Moreover, this dissonance can also be interpreted via the social desirability scores of the engineering students. Social desirability is associated with the “the need of Ss [subjects] to obtain approval by responding in a culturally appropriate and acceptable manner” (Crowne and Marlowe 1960, p. 353). Present results show that perfectionists were tending to have higher Social Desirability scores than non-perfectionists in general (see Table 10). This, therefore, indicates that perfectionists may have likely provided responses reflecting higher prosocialness and empathy due to their general desire to deliver more socially desirable and socially acceptable responses.

On another note, Fermandel (2015) found that perfectionists are likely to hold self-protective and anxiety-avoidant motives and drivers, for persevering “self and existing status quo” and “serving to cope with anxiety”. Fermandel’s (2015) findings, combined with Fitton and Moncaster’s (2019) finding on engineers’ undesirability of decentralising the power of decision-making in the design process, reassures this paper’s allegation that such conservative act of centralising decision-making power may indeed be defensive, and is predicted to be protective of the engineers’ image and ‘peace of mind’. This may tap into reasons as to why engineers have, and continue to struggle with the integration of the more complex human-centred consideration into practice; as such wicked problem solving not only demands new, challenging, creative and more complex approaches to problem-solving, it may challenge the current natural positivistic state of problem-solving in engineering, and with that, the engineers’ ‘peace of mind’ and status-quo, possibly triggering their anxiety. This could also be an answer to Elsbach and Stigliani’s (2018) questioning of why perfectionistic cultures are likely to hinder design thinking.

Additionally, with the present findings show that the majority of civil engineering students are more likely to hold perfectionistic traits (– with a high percentage of them being maladaptive) combined with the extensive research on the negative associations of perfectionism to mental or physical wellbeing (Blatt et al. 1995; DiBartolo et al. 2008; Geranmayepour and Besharat 2010; Molnar et al. 2006), this paper sheds light and recommends engineering cultures and curriculums to actively implement strategies that tend to their subjects’ perfectionism and consequential wellbeing. This recommendation also applies to other cultures and paradigms where positivism and perfectionism are known to be predominant—in both the academic and practicing fields. Moreover, this paper recommends interventions to be set for mitigating perfectionism to also facilitate design thinking, engagement with human-centred and public-welfare-considerate initiatives, and Communal Design production in engineering.

Finally, the present findings shed light on the importance of further exploring how influential engineering mindsets and characteristics can be on engineers’ decision-making processes, design judgements, and their engagement with human-centred, public-welfare-related initiatives.

In seeking to support perfectionist students, learning and assessment strategies likely require the incorporation of less convergent and more divergent learning methods, together with cognitive tools to support students to recognise and challenge any tendency to perfectionism. The regular use of open-ended and contextualised teaching and learning activities can normalise the consideration of complex, integrated socio-technical design problems. This exposure could make scenarios less daunting for perfectionist students who may enter higher education initially preferring the comfort of a defined answer (as usually experienced in secondary education in STEM). Examples of strategies that appear well suited include the ‘Grassroots/ Popular Engineering’ related by Cordeiro Cruz (2021) with its commitment to foregrounding equity and liberation in the design process. CDIO (Conceive, Design, Implement, Operate) related by Crawley et al. (2014) is a learning framework that encourages student learning through real design. CDIO advises the building of opportunities to learn iteratively from failure, which could help perfectionist students to reframe failure as a normal part of the learning process rather than interpreting it as a personal deficit. Critical reflection is a pedagogic tool suited to increasing awareness of decision-making through the design process, and offers a chance for students to reconsider their actions, seeking the root causes driving their decision-making (Orbaek White et al. 2020). A useful exercise suited to improving understanding of inclusivity may be to conduct critical reflection on a completed engineering design process together with information about perfectionism and other traits. This could enhance students’ reflexive capability to identify how different minds approach the same situation, giving them insight into how their own cognitive processes may differ to others (Table 17).

Lastly, a model predictive of the probability of producing Communal designs was developed using the 12 characteristics addressed in this study, which had showed an acceptable sampling adequacy (see Table 12) after conducting a factor analysis. The model, however, showed low fitness and none of the characteristics showed significance in influencing the outcome (see Tables 18 and 19, respectively). However, the model produced an equation, tailored to the present particular case, that showed that Empathy: Empathic Concern followed by Prosocialness scores were the most influential on the probable outcome of Communal Design production—where Empathy: Empathic Concern contributed negatively to the probability of producing Communal Designs, and Prosocialness contributed positively to the probability of Communal Design production (see Eq. 2).

This study assessed the prevalence of perfectionism in an undergraduate cohort of civil engineers, and its potential association with their engagement with a communal, human-centred, social-welfare considerate initiative (Communal Designs). This study found that 74.5% of civil engineering students categorise as perfectionist, with 68.5% of those perfectionists being maladaptive. We found that although perfectionists were less likely to produce Communal Designs, they were still found more likely to hold higher prosocialness and empathy scores during the Communal Design production process (i.e., showing more engagement with the human-centred designing initiative). This hinted towards an intention-behaviour gap in which engineering students perhaps intended to but then failed to produce Communal Designs that met the metaphysical criteria addressed in Table 1 (extracted from Max-Neef et al.’s (1991) Matrix of Human Needs and Satisfiers). This could plausibly be due to engineering culture’s dissonance in training and expectations of performance on societal and public welfare considerate initiatives. This dissonance was also discussed in terms of social desirability, as this study found that perfectionists were tending to hold higher social desirability scores, i.e., were tending to have provided responses reflecting higher prosocialness and empathy due to their desire to deliver what they may consider as more socially acceptable responses to those assessing their work.

We conclude that in the face of ambiguity (often inherent in complex socio-technical design), the power-centralised and conservative engineering design decision-making process may become defensive due to the perfectionist tendencies of self-protection and anxiety-avoidance. This could inform why some engineers struggle to move from rigid and positivist learning and practice to more creative, empathic, human-centred, and socially considerate initiatives. It is recommended to implement strategies in both learning and assessment design to mitigate perfectionism in engineering. Strategies discussed include more regular exposure to contextualised design to normalise complexity and creativity, together with strategies for reflection on how different types of minds may approach a situation in different ways.

Limitations of the study include the small dataset on which these conclusions are based, and as the assessment was held during COVID, the assessment was completed online rather than in person. It would be worthwhile to replicate the study at a greater scale and in more typical conditions to confirm if the association seen in this study remains persistent. Moreover, as responses collected in Phase I of the intervention could not have been matched to those collected in Phase II, the response numbers were reduced further.

Further in future work, to avoid considering factors that are not statistically significant, we will use a more advanced statistical technique, such as LASSO (least absolute shrinkage and selection operator), to reduce the contributing factors to the most significant ones. LASSO will also eliminate any potential factor selection bias, and will therefore produce more concise and objective results. Moreover, we will conduct PCA (principal component analysis) with rotations to eliminate the risk of potential covariations in the dataset.

There is further work required to determine how perfectionist (and other) traits of civil engineers and engineering students compare to those of other disciplines in engineering, or other fields. There is also further work required on what strategies can be used to help student wellbeing and support students to overcome the limitations that perfectionism and maladaptive perfectionism may be placing on achieving fully rounded engineering designs for an equitable society.