The potential of modular product design on repair behavior and user experience – Evidence from the smartphone industry

Innovation through modular product design is a promising strategy for product lifetime extension and material recyclability. Yet, sustainability benefits of the design do not automatically come into effect, but require additional process innovation through service activities that have to be attractive and made use of by users. This study investigates to what extent sustainability-focused modular product design in the smartphone sector promotes (1) self-repair versus use of repair service and (2) a positive user experience with repair instructions and services. It further analyzes how circular economy attitudes and perceived self-repairability moderate these effects. The quantitative analysis finds that, strengthened by higher perceived self-repairability, users of modular smartphones are more likely to use repair instructions. Also, modular smartphone design increases positive experience with repair instructions. Consequently, successful implementation and management of complementary product and service innovation are key to promoting product lifetime extension in the smartphone industry.


Introduction
Manufacturing and supply chains, including the use of energy and precious metals, contribute disproportionally to the total environmental impact of smartphones (Schischke et al., 2019b;Suckling and Lee, 2015).Product innovation through modular redesign supports the repair and exchange of modules, which makes it a promising approach for product lifetime extension and material recyclability (Schischke et al., 2019a).Yet, product lifetime is not only affected by absolute obsolescence due to technological failure, but also by relative obsolescence due to the user's decision to discard a product (e.g. for reasons of fashion) although it is still functional (Cooper, 2004).This suggests that the sustainability benefits of modular smartphone design only materialize if this product innovation is complemented by process innovation offering and incentivizing the use of circular service offers (e.g.repair, reuse, fashionable upgrading) by customers.Extant literature proposes establishing circular loops in sequence from repair to recycling to retain the maximum value of products, components, and materials (Stahel, 2010).In contrast to the focus of leading smartphone manufacturers on the recycling of discarded devices, this paper explores the potential of repair activities to foster sustainability, and thus contributes to the right to repair legislation (European Commission, 2019).Identifying obstacles to repair resonates with the sufficiency strategy, which emphasizes longer product lifetimes and reduced consumption (Bocken and Short, 2016;Freudenreich and Schaltegger, 2020;Iyer and Muncy, 2009).The benefits of repairing products requires easy disassembly and reassembly to support not only professionals but also users in efficiently repairing a product (Agrawal and Ülkü, 2013).However, the possibility of repairing a product and whether this option is actually utilized in practice may diverge.
Sustainable modular product design (SMPD) can foster sustainability-oriented innovations (SOI) and help to unfold their benefits.Modularity enables both product (i.e., the technical design) and process (i.e., circular services) innovations along the entire life cycle, including use and recovery, circularity, and reduced overall negative sustainability impact (Bustamante, 2020;Den Hollander et al., 2017;Hansen et al., 2009).However, previous research has not explored SOI in depth from a user perspective (Piscicelli and Ludden, 2016;Wieser and Tröger, 2017) particularly regarding the investigation of different repair activities and user experience with repair services.Lofthouse and Prendeville (2017) emphasize that the adoption of a user-centric perspective to innovation is crucial to the design of successful business models and to realizing the potential to create transformative change.After all, the user determines whether an innovation's sustainability potential is realized through repair and other circular measures that affect the product's lifetime (Nazlı, 2021;Sonego et al., 2018;van der Laan and Aurisicchio, 2019).Despite the importance of considering user behavior in determining modularity's sustainability potential, a literature review on modularity and sustainability by Sonego et al. (2018) indicates that past research has neglected this life cycle phase.The authors recommend that future research should develop a deeper understanding of the consumption of modular products to determine which modularization strategy creates the highest sustainability impact (Sonego et al., 2018).
Against this background, this study investigates the relationship between technological and complementary circular service innovation from a user perspective.In order to improve understanding of circular user behavior and experience with modular products and the crucial role of process innovation in complementing product innovation, we attempt to answer the following research question: Does sustainable modular product design (SMPD) influence the repair behavior of users, and if so to what extent?In principle, there are two repair options: self-repair using instructions or professional repair services (Wieser and Tröger, 2017).Consequently, this study analyses user experience with these two repair options.
This study is based on survey data of users of an original equipment manufacturer (OEM), here pseudonymized as SmartMod, and applies multiple logistic and linear regression models.SmartMod is a sustainability pioneer in manufacturing modular smartphones.This study analyses survey data from a pool of 1,720 observations, of which 588, 306, and 478 are used for the regression analyses of Hypotheses 1, 2, and 3 respectively.The findings show that, strengthened by higher perceived self-repairability, SMPD successfully increases the likelihood of selfrepair.While a SMPD approach promotes a positive experience with repair instructions, the data does not show that it promotes a positive experience with repair services.In addition, for users with strong circular economy (CE) attitudes modular design is less important for them to have positive experiences with repair instructions.Consequently, management should consider complementarities between product and service innovation as well as supporting communication to leverage the technical design's sustainability potential.
The remainder of this paper is structured as follows.First, the relevant literature on modularity and SOI is presented, followed by the development of the hypotheses.After an explanation of the quantitative method used in the study, the results of modularity's impact on repair behavior and user experience with repair instructions and services are discussed, and finally conclusions are drawn for research and practice.

Sustainable modular product design
Modularity is a promising approach to promoting the circularity of products and materials.Starr (1965) originally introduced the modular concept as a way to support increasing product variety at relatively low cost by allowing for a maximum combination of parts.Later, the perspective developed from a production focus to a more user-centric focus, and accelerated the mass customization of products (Sanchez, 1995).Today, modularity is being further developed as a design principle to facilitate (corporate) sustainability improvements throughout a product's life cycle (Sonego et al., 2018).This empirical study builds on the last perspective, called sustainable modular product design (SMPD).
Following Gershenson et al. (1999), modularity describes the bundling of interdependent components into independent modules.Modules in turn are connected by interfaces to form a reversible technical system.This creates benefits throughout the entire life cycle by allowing circular service operations for replacements, upgrades, and recovery (Das and Posinasetti, 2015;Hankammer et al., 2018).This differentiates SMPD from conventional modularity, as product circularity reduces negative environmental impacts of product and material obsolescence, and helps to extend product lifetime with the ultimate goal of advancing sustainability (Agrawal and Ülkü, 2013;Kim and Moon, 2019).
In particular, SMPD simplifies the recovery of products, modules, and components through easy disassembly and reassembly as well as upgradability, which saves process costs for repair and creates an environmental advantage compared to conventional products (Agrawal and Ülkü, 2013;Mesa et al., 2018).First, products can be maintained and repaired at lower cost, as replacing defective modules becomes easier (Agrawal and Ülkü, 2013).Second, functioning modules of a damaged product can be reused in another user's product (Gu et al., 1997).Third, products can be refurbished or remanufactured by reusing well-functioning modules and/or modules can be upgraded, for example, by updating software (Gu et al., 1997;Pialot et al., 2012).Fourth, if modules can no longer be reused or repaired, easier material separation increases the component's recyclability and recycling efficiency (Gu et al., 1997;Kim and Moon, 2015).However, SMPD alone does not suffice to promote product sustainability (Den Hollander et al., 2017).If, for example, SMPD were combined with a conventional "sell more, sell faster" business model, the product would still be unsustainable.Whether SMPD unfolds its sustainability potential depends on its embedding in a whole product-service system, including ease of repair, module exchange, upgrading, etc. Complementary services are therefore crucial to ensure modular products actually contribute to sustainability.

SMPD as enabler for sustainability-oriented innovation
SMPD can foster sustainability-oriented innovation (SOI) through the introduction of redesigned products and complementary services.SOI describes innovative products and/or services that create a sustainability advantage by considering three underlying dimensions (Hansen et al., 2009).These underlying dimensions of SOI, proposed by Hansen et al. (2009) and Hansen and Große-Dunker (2013), encompass a target dimension, a life cycle dimension, and an innovation type dimension.

SOI's target dimension: corporate sustainability and circularity
Early research on SOI suggested corporate sustainability as its target dimension (Hansen et al., 2009).In light of recent research, this target should include circularity as a strategy supporting sustainability at the economic and societal level (Weissbrod and Bocken, 2017).Our study examines how SMPD (i.e. the product's technical composition facilitating circularity) can support interlinking environmental and economic aspects at the corporate level and beyond.As humankind has already exceeded several planetary boundaries (Rockström et al., 2009;Steffen et al., 2015), CE literature emphasizes a strategy of keeping materials and durable consumer goods or components in use as long as possible (Van Nes and Cramer, 2005).The CE concept supports environmental and economic objectives for product lifetime extension through four product circularity strategies, that is, maintain/repair, reuse/redistribute, refurbish/remanufacture, and recycle (Bocken et al., 2016, Ellen MacArthur Foundation, 2013;Geissdoerfer et al., 2017).Because product repair has the potential to extend maximum use, this study investigates user behavior related to slowing resource loops (Bocken et al., 2016;Geissdoerfer et al., 2017;Stahel, 2010).This resonates with the sufficiency debate that aims to reduce consumption in order to diminish negative sustainability impacts (Bocken and Short, 2016;Freudenreich and Schaltegger, 2020;Iyer and Muncy, 2009).Businesses can promote sufficiency by offering repair services to extend product lifetimes (Freudenreich and Schaltegger, 2020).

SOI's life cycle dimension: emphasis on extended use and recovery to prevent obsolescence
The life cycle dimension of SOIs spans an innovation's entire value chain from supply to recovery (Hansen et al., 2009).This study focuses on the closely linked use and recovery phases, which have the most direct effects on extending the lifetime of durable consumer products.One option to increase product lifetime is durable design.However, consumables such as smartphones with fast technological change are often not discarded because of material deterioration but due to other reasons, such as decreasing speed due to software updates creating insufficient storage and processing capacity.Product obsolescence must therefore be prevented in use and recovery phases.In the use phase sustainability effects can be created by repairs performed by users themselves or by professionals, while in the recovery phase service providers can refurbish the smartphone (Hansen and Große-Dunker, 2013).The use phase is especially difficult for companies to control as users are themselves responsible for the repair of their product (Hansen and Große-Dunker, 2013).SOI allows companies to extend product lifetime with design decisions of complementary products and servicesespecially for products whose lifetime should be maximized to unleash their sustainability potential (Bakker et al., 2014;Den Hollander et al., 2017).Studying use and recovery phases implies considering the options and impact of the design and manufacturing phases too.
Products are discarded for various types of obsolescence (Cooper, 2004).Cooper (2004) notes that product design not only influences absolute obsolescence, which describes technical failure, but also relative obsolescence, which describes the user's decision to discard a still functional product.The concept of relative obsolescence explains why users refrain from product repair.There are three types of relative obsolescence: technological, economic, and psychological obsolescence (Cooper, 2004).Technological obsolescence occurs if users are drawn to new functions in more recent models because they fear future lock-ins of their current technology and are pressured by technological change (Cooper, 2004).Economic obsolescence occurs if the product's perceived monetary value decreases from the user's point of view (Cooper, 2004;Lee and Lee, 1998).This type of obsolescence also arises if replacement parts or repair services are too expensive or time-consuming compared to product replacement (Amankwah-Amoah, 2017).Psychological obsolescence occurs if the product's emotional value for the user decreases, that is, the user is no longer attracted to or satisfied by the product (Cooper, 2004;Packard, 1960).For example, the user feels pressured by social norms to buy a new product.

SOI's innovation type dimension: bundling products and services
Product, process, and organizational innovation is required to postpone perceived or relative obsolescence (Hansen et al., 2009).This study focuses on product and process innovation and their complementarities, since both can be leveraged through SMPD.Recent research by Bustamante (2020) emphasizes the importance of both product and process innovation to contribute to sustainable production and consumption.Product redesign towards modularity can leverage product innovation with a focus on technological features, while SMPD supports repair services which leverage process innovation.In turn, when both types of innovation are complementary, they enable the success of the four product circularity strategies (i.e.maintain/repair, reuse/redistribute, refurbish/remanufacture, and recycle) (Bocken et al., 2016).Thus, the bundling of products and services into so-called product-service systems (PSS) can enhance SOI by enhancing the product innovation level (Fain et al., 2018).This study investigates PSS, which in the SmartMod business model still focusses on the sale of products complemented by additional services (Aas et al., 2020;Hansen and Große-Dunker, 2013;Tukker, 2004).These additional services are typically either product-related services or consultancy and advisory services, and are provided during the product's use phase (Tukker, 2004).Examples of product-related services that foster circularity and sustainability are repair services and product take-back systems.Examples for consultancy and advisory services are repair instructions and information provided by the manufacturer on circular product use.The provision of after-sales services is especially recommended in marketing as they create an economic and environmental advantage compared to providing replacement goods (Vezzoli and Manzini, 2008).
In summary, SMPD can enable SOI through complementary product and service provision to support product circularity and create a sustainability advantage especially during the use and recovery phases.This study contributes to knowledge about SOI in general and the success of circularity in modular product innovation in particular.Hypotheses are developed to empirically test the effect of SMPD on product circularity, in particular, of SMPD's effect on users use of and experience with different repair services and on factors that influence user repair behavior.

Development of hypotheses
Based on this review of the relevant literature, we generate three hypotheses concerning the relationship between SMPD in smartphones and consumers engaging in self-repair as moderated by circular economy attitudes and perceived self-repairability.The following sections outline how these hypotheses have been derived.

Use of repair instructions compared to professional repair service
To decrease the environmental burden created largely in their manufacturing phase, smartphones need to be repaired to extend their lifetime (Schischke et al., 2019b;Suckling and Lee, 2015).Repairs can either be performed by users themselves or by professional service providers.However, unsuitable product designs, lack of repair information, and high costs have been assessed as key barriers for self-repair in many product categories, including smartphones (Bigerna et al., 2021;Dao et al., 2020;Ellen MacArthur Foundation, 2016;Wieser and Tröger, 2017).Unsuitable product designs increase the difficulty for professional repair too.However, SMPD may have a greater potential to advance self-repair over professional services.Repair instructions for modular products are easier to understand and follow compared to those for conventional products.Barriers to self-repair modular devices are also lower because less effort and technical skills are required (Agrawal and Ülkü, 2013;Kim and Moon, 2019).Decreased efforts go hand in hand with lower repair costs, and self-repair costs are lower than using a professional service.In addition, do-it-yourself (DIY) repair modularity (Schischke et al., 2019a), with proprietary designed product architecture based on repair statistics, directly targets self-repair of modules that are most frequently defective.Therefore, we hypothesize that: H1.Modular smartphone design increases the likelihood of users using repair instructions compared to professional repair services.
Actual repair behavior depends on user attitudes (Mugge et al., 2005).Studies show that users with pro-environmental attitudes are more prone to adopt repair practices (Bigerna et al., 2021;Cerulli-Harms et al., 2018;Nazlı, 2021).This study investigates the influence of CE attitudes.The construct to measure CE attitudes also includes items to measure environmental as well as repair attitudes (Cerulli-Harms et al., 2018).Higher CE attitudes increase user choice to self-repair a damaged modular device, and SMPD would therefore increase the motivation for self-repair.Hence, the likelihood that users choose self-repair might depend on their attitudes towards CE.Therefore, we hypothesize that:

H1a. Positive CE attitudes support (positively moderate) the positive effect of modular smartphone design on the likelihood of users to use repair instructions for self-repair compared to professional repair services.
Furthermore, user perceptions may influence their decision to repair (Nazlı, 2021).Users who have doubts whether a product can be repaired are more likely to refrain from self-repair and choose repair service instead (Ackermann et al., 2018;Wieser and Tröger, 2017).This is because they could fear they are incapable of repairing the device, they could do more harm than good in the process or they could need too much time.This in turn could result in overall higher repair costs than using repair service from the start.Perceived self-repairability could therefore motivate users of modular products to conduct self-repair.Therefore, we hypothesize that: H1b. High perceived self-repairability supports (positively moderates) the positive effect of modular smartphone design on the likelihood of users to use repair instructions for self-repair compared to professional repair services.

User experience with repair instructions
Repair instructions and repair services have both been proposed as approaches to create positive user experience, which would translate into higher satisfaction and loyalty (Zomerdijk and Voss, 2010).User satisfaction resulting from a positive experience would promote product lifetime extension (Ackermann et al., 2018;Sabbaghi et al., 2016).An appropriate technological design may benefit both user experience with repair instructions as well as with repair services.For example, repair instructions for modular products may be simpler and thus easier to understand and carry out.Also, the use of repair instructions for modular products may require less time and effort as well as increase the success of self-repair, which would also impact user experience.Therefore, we hypothesize that:

H2. Modular smartphone design increases positive user experience with repair instructions.
While attitudes generally influence repair behavior (Fogg, 2009;Mugge et al., 2005;Nazlı, 2021), they may also influence user experience (Michalco et al., 2015).Modular smartphone users who possess strong CE attitudes may have more previous self-repair experience with household products, clothing, etc. and thus have more positive experiences with repair instructions.Therefore, we hypothesize that:

H2a. Positive CE attitudes support (positively moderate) the positive effect of modular smartphone design on positive user experience with repair instructions.
More information about repairability incentivizes users to self-repair damaged products (Bigerna et al., 2021).Information about repairability can also be indirectly transferred with product design, that is, modular phones are intuitively perceived as more repairable.Therefore, users of a modular smartphone who perceive their device as easily self-repairable may have a more positive attitude towards repair.High perceived self-repairability may influence users to make an even better experience with repair instructions of modular smartphones.Therefore, we hypothesize that: H2b. High perceived self-repairability supports (positively moderates) the positive effect of modular smartphone design on positive user experience with repair instructions.

User experience with repair services
Modular design allows for a quicker exchange of certain modules, which increases the efficiency of professional repair services, decreases costs, and allows the device to be returned more quickly to its owner.These advantages could overcome barriers to users seeking professional repair service (Bigerna et al., 2021) and increase their positive user experience.Therefore, we hypothesize that:

H3. Modular smartphone design increases positive user experience with professional repair services.
As with Hypothesis 2a, environmental attitudes influence repair experience.This may also be the case for user experience with professional repair services.Users with positive CE attitudes are more likely to have higher previous experience with repair services.However, because repair service of modular products is more convenient and of better quality, user experience may be higher.Therefore, we hypothesize that:

H3a. Positive CE attitudes support (positively moderate) the positive effect of modular smartphone design on positive user experience with professional repair services.
User experience of repair services for modular products may be influenced by perceived self-repairability.Users who know their smartphone is easily repairable, but who lack the skills or willingness to repair the device themselves, may choose a professional repair service instead because they have greater trust that the repair will meet their expectations.Therefore, we hypothesize that:

H3b. Positively perceived self-repairability supports (positively moderates) the positive effect of modular smartphone design on positive user experience with professional repair services.
Fig. 1 illustrates the research design.

Methodology
This section outlines the data collection, describes the case and how the data is analyzed, outlines the measures included in data analysis, and summarizes the descriptive statistics.

Data collection
To test the hypotheses, this study adopts a quantitative approach using data from users of a sustainability pioneer in the smartphone industry.This study focuses on the smartphone market, as smartphones contribute most to the carbon footprint of information and communication technology (ICT) (Belkhir and Elmeligi, 2018).ICT products are at the core of the EU's CE Action Plan (European Commission, 2020).Among ICT products, smartphones provide a high potential for repair (Cerulli-Harms et al., 2018), which can be leveraged through modular design (Schischke et al., 2019a).
In the CE literature, modularization of smartphones is often discussed as an approach that can help improve the product's sustainability performance in terms of lifetime extension through improved repair and recycling (Hankammer et al., 2018;Schischke et al., 2019a).To better understand and increase innovation success, an analysis of user repair behavior after a change of the technical design is crucial (Lofthouse and Prendeville, 2017).
Smartphones were chosen for this study because pioneers such as SmartMod have started to adopt modularity as an innovative product design.SmartMod is a vertically-integrated OEM (Hansen and Revellio, 2020) based in Germany and specialized in the sale of semi-modular and modular smartphones while providing an in-house repair service and repair instructions.SmartMod launched its first semi-modular device in 2015, and the first modular device in 2018.Its semi-modular smartphones are based on existing product architectures, but include repairability upfront.The modular smartphone design adopts DIY repair modularity (Schischke et al., 2019a) with a proprietary designed product architecture based on repair statistics.iFixit (2020) rated Smart-Mod's semi-modular devices with a repairability score of 6/10 and the modular devices with 9/10, reflecting different degrees of modularity (Schischke et al., 2019a).Comparing these two technical designs allows to investigate the difference in modularity on user repair behavior.Furthermore, SmartMod has a well-maintained relationship to their users, who also take advantage of their in-house services to care for their smartphones.Their user attitudes toward circular behavior is thus relatively positive, which differentiates them from the mass-market consumers.SmartMod also aims to engage users in product innovation, creating a potentially stronger relationship with their devices and enhanced circular behavior, such as repair (Mugge et al., 2005;Page, 2014).
Primary data was obtained with an online survey (see supplementary material) from a population of more than 50,000 current and former users of SmartMod smartphones.The questionnaire was distributed through SmartMod's communication channels: their newsletter, Facebook page, as well as their forum, and a user-organized Facebook group.Users were able to participate from June until November 2020.In total, 3,801 subscribers and followers responded to the survey, 2,068 of whom owned a SmartMod device, with 2,005 owning a new SmartMod device.Those indicating an unrealistic age or no age (i.e. one respondent) were excluded.As a further quality assurance procedure, respondents who finished the questionnaire too quickly were excluded to avoid speeder bias.In general, answer times varied since respondents were not required nor able to answer all questions (e.g. a respondent who only used repair instructions cannot answer the questions on repair services).Therefore, answer times were separately controlled for in each regression model.The sample sizes differed depending on the variables considered by the respective regression model, as not all respondents provided answers and data for all questions.From a pool of 1,720 observations, 588 observations were included to test Hypothesis 1, 306 to test Hypothesis 2, and 478 to test Hypothesis 3.

Measures
Table 1 outlines the variables included in the data analysis according to the hypotheses.

Data analysis and descriptive statistics
The survey data was analyzed using descriptive statistics (Table 2), binary logistic regression models for Hypothesis 1, and multiple linear regression models for Hypotheses 2 and 3. SPSS, including the PROCESS macro by Hayes (2017), was used for data analysis.
Table 2 outlines the number of observations (N), the mean (M), and standard deviation (SD), of each variable included in the regression analysis as well as correlations between the variables.
The correlations between the dependent variables and primary independent variables show preliminary support for the hypotheses.Modular smartphone design is positively correlated with self-repair compared to using repair service (ρ = 0.361**), as well as a positive experience with both repair instructions (ρ = 0.264**) and repair service (ρ = 0.098*).Also, the means of both experience with repair instructions (M = 3.996, max.= 5) and repair service (M = 3.883, max.= 5) are relatively high, suggesting a successful service design by SmartMod.
Table 3 displays the socio-demographic composition for each sample used for the regression analysis of the hypotheses.Male respondents predominate in the observations in all samples, particularly in the second group of regression models.The age distribution, presented in categories, reveals a slightly higher number of younger respondents.The majority of users has completed university studies and has a monthly income between 2,251 and 4,500 euros.
Regarding the study's overall research aim -How does modular design impact repair behaviorpreliminary support can be drawn from the relative repair behavior concerning modular and semi-modular devices (see Fig. 2).
The descriptive statistics show that in case of defects both modular (dark shaded columns in Fig. 2) and semi-modular devices (light shaded columns) are often repaired (see first three column pairs in Fig. 2).A slightly higher percentage of defective semi-modular devices was not repaired (15%, left column).Considering the two repair options, modular devices are three times more often repaired by users using repair instructions (39%) than semi-modular devices (13%) (comparison of the first two columns on the left in Fig. 2).As a consequence, twice as many semi-modular smartphone users chose SmartMod's repair service instead of self-repair.Furthermore, by comparing cases that only use repair instructions or repair service, slightly more than half of the modular devices were self-repaired using instructions while three quarters of the semi-modular devices were repaired by the service.This data, which is similar when considering only display repairs, corresponds with SmartMod's sales data for display service repairs versus display orders as an indicator for self-repair.
While slightly more than 10% used both repair instructions and service, this result does not provide information whether self-repairs failed and were then given to a repair service.The analysis of experiences of all users compared to those with experience with repair instructions who also used a repair service does not indicate self-repair failure.Instead, different services could have been used for different defects or even different devices.
Few users did not repair their damaged smartphone (14% in total), primarily because they were able to continue to use it without repair or because they perceived the effort to repair the device as too high.Because agreement with the other items was rather low, we can derive that users did not choose not to repair their smartphone because they found the repair too expensive or the device out of fashion, had an irreparable defect, did not know how or where to repair it, or could not get the needed spare parts.Also, only a minority of those who did not C. Amend et al. repair chose a replacement or device upgrade instead.

Results
Fig. 1 shows our research design.For each hypothesis, a regression is performed.To test Hypothesis 1 -the effect of modular smartphone design on the likelihood to use repair instructions compared to a professional repair servicebinary logistic regression models were applied due to the dependent variable's binary character.The second and third hypothesesthe effects of modular smartphone design on user experience with repair instructions and repair servicesare tested applying multiple linear regression models.All logistic and multiple linear regression models have a maximum variance inflation factor (VIF) below 10, suggesting that they are not affected by multicollinearity (Kennedy, 1992).

Effect of modular smartphone design on the likelihood to use repair instructions over a professional repair service
Table 4 outlines the results of the binary logistic regression models showing the exponentiations of the β-coefficients, which are odds ratios.Odds ratios are displayed instead of β-coefficients because their interpretation is more straightforward, that is, they show the relative change in the dependent variable by a one-unit increase of the independent variable (Urban and Mayerl, 2008).Further, the reciprocal ( 1 exp(β) ) is calculated for odds ratios below 1 to balance the unequally scaled value areas (from 0 to 1, and from 1 to +∞) (Urban and Mayerl, 2008).
Model 1 tests the influence of modular smartphone design on the likelihood of using repair instructions compared to using repair service.The results reveal that users of a modular smartphone are more than three times more likely (exp(β) = 3.538; p < 0.01) to choose to repair the defect themselves than are users of a semi-modular smartphone.Therefore, we can confirm Hypothesis 1 that modular smartphone design enabling easy and fast repairs is effective in promoting self-repair with instructions.The control variable "CE attitudes" does not influence the likelihood to use repair instructions, while, as expected, perceived self-repairability exerts a significant influence (exp(β) = 2.282; p < 0.01), that is, users are more than twice as likely to repair their device themselves if they perceive their device more repairable by one point.Female users are 59% less likely to choose repair instructions ( 1 exp(β) = 1 0.613 = 1.585; p < 0.05), while age, education, and income do not exert a statistically significant influence on the likelihood to use repair instructions.
Model 1a tests the moderating effect of positive CE attitudes on the positive influence of modular smartphone design on the likelihood to use repair instructions over repair service.The results do not show a statistically significant influence (exp(β) = 1.294; p ≥ 0.1) and thus cannot support Hypothesis 1a.Perceived self-repairability has a positive and being female a negative influence, while age, education, and income do not influence the likelihood of self-repair.
Model 1b tests the moderating effect of high perceived selfrepairability on the positive influence that modular smartphone design has on the likelihood to use repair instructions compared to repair service.Due to the positive, statistically significant interaction term of modular smartphone design and perceived self-repairability (exp(β) = 2.204; p < 0.05), the results support Hypothesis 1b.Thus, when perceived self-repairability is high, the tendency to self-repair a smart-

Table 1
Variables included in data analysis.

Use of repair instructions vs. service
Binary variable; 1 = only repair instructions used, 0 = only SmartMod's repair service used.Repair instructions include manuals and support (e.g. via email) by SmartMod as well as third-party manuals (e. g. by iFixit).All users who self-repaired used some kind of repair instructions or support.

Positive experience with repair instructions
Average of 6 items (requirement, helpfulness, experience, availability, accessibility, and fun of repair instructions) on a 5-point Likert scale adapted from Cerulli-Harms et al. (2018).While the Cerulli-Harms et al. ( 2018) study asked for experience with professional repair services, this study adapted items to fit prior experience with repair instructions.Cronbach's alpha = 0.658 (Loewenthal, 2004).

Positive experience with repair service
Average of 5 items (accessibility, convenience, speed, friendliness, and quality of repair services) asked on a 5-point Likert scale based on consumer research conducted by Cerulli-Harms et al. (2018).Only repair service offered by SmartMod is considered because it was adopted by the majority of users (only 8 respondents used a third-party service).This also reduces the potential influence of service providers.Cronbach's alpha = 0.811 (Loewenthal, 2004).2020), on a 5-point Likert scale.Cronbach's alpha = 0.672 (Loewenthal, 2004).Centralized values are used for the interaction with modular smartphone design.

Times repair instructions used
Count variable expected to have an influence on positive experience with repair instructions.

Times repair service used
Count variable expected to have an influence on positive experience with repair service.

Age
Coded in years; respondents indicating an unrealistic age were excluded from analysis.Prior research found that older consumers are more likely to repair an appliance (McCollough, 2010).

Education
Categorical variable: 1 = no school-leaving qualification (or still in school education), 2 = completion of 8th or 9th grade, 3 = middle school degree with completion of 10th grade, 4 = general or subject-specific university entrance qualification, 5 = completed studies (bachelor's, master's, or equivalent), 6 = doctoral degree or post-doc.Higher education levels may correlate with a decreasing lack of knowledge to repair products (Dao et al., 2020;Laitala et al., 2021).Income
Higher income is associated with a lower likelihood to repair products in general (McCollough, 2010).However, it might increase the likelihood to choose the more expensive repair service option.
C. Amend et al. phone using instructions rather than hand it in for professional repair is increased by more than 120% for those with a modular smartphone compared to those with a semi-modular smartphone.For those with a low self-repairability perception, there is no difference.Regarding the socio-demographic variables, female and age exert a negative, while education and income exert no influence on the likelihood to choose repair instructions.

Effect of modular smartphone design on positive experience with repair instructions
Table 5 displays the β-coefficients of the multiple linear regression models testing the effects of modular smartphone design on positive experience with repair instructions (H2) as well as the moderating effects of CE attitudes (H2a) and perceived self-repairability (H2b).
Model 2 tests the influence of modular smartphone design on positive experience with repair instructions.According to the results, modularity increases positive experience with repair instructions (b = 0.156; p < 0.05), supporting Hypothesis 2. CE attitudes do not influence positive experience with repair instructions, while perceived selfrepairability exerts a strongly positive influence.For those cases where repair instructions are used, all socio-demographic variables do not influence positive experience with repair instructions.
Model 2a tests the moderating effect of positive CE attitudes on the influence of modular smartphone design on positive experience with repair instructions.The direct effects of modular smartphone design (b = 0.188; p < 0.05) and CE attitudes (b = 0.266; p < 0.1) positively influence the experience made with repair instructions, meaning that users with a strong orientation towards circular behavior are more likely to make a positive experience.However, the interaction term of modular smartphone design and CE attitudes has a tendentially significant negative coefficient (b = − 0.319; p < 0.1).Although statistically significant, when a significance level of 10% is accepted, the moderation's direction is negative, which contradicts Hypothesis 2a.Thus, the stronger the user's CE attitudes, the lower the impact of modular

Table 2
Descriptive statistics and correlations of variables.smartphone design on positive experience with repair instructions.This means that users with a strong orientation towards circular behavior are not more or even less incentivized by the modular design to make a positive experience with repair instructions.For those with weak CE attitudes, there is no difference.Perceived self-repairability also positively influences the outcome variable, and when repair instructions are used, the socio-demographic variables again have no influence.Model 2b tests the moderating effect of high perceived selfrepairability on the positive influence of modular smartphone design on positive experience with repair instructions.Although perceived selfrepairability wields influence on positive experience with repair instructions (b = 0.659; p < 0.01), it does not have a moderating effect (b = − 0.092; p ≥ 0.1).Therefore, Hypothesis 2b cannot be confirmed.The remaining control variables are also not found to influence the experience.

Effect of modular smartphone design on positive experience with repair services
Table 6 displays the β-coefficients of the multiple linear regression models testing the effects of modular smartphone design on positive experience with repair services (H3) as well as the moderating effects of CE attitudes (H3a) and perceived self-repairability (H3b).
Model 3 tests the influence of modular smartphone design on positive experience with repair service.The analysis does not find support for Hypothesis 3. Therefore, modular smartphone design does not increase positive experience with repair services (b = − 0.084; p ≥ 0.1).Only the control variable times repair service used is found to have a slight negative impact while education is found to have a slight positive impact on experience with repair services.
Model 3a tests the moderating effect of CE attitudes on the influence of modular smartphone design on positive experience with repair service.The moderation of CE attitudes is statistically insignificant (b = − 0.344; p ≥ 0.1), meaning Hypothesis 3a cannot be confirmed.While perceived self-repairability and level of education increase positive experience with repair service, the frequency of repair service use reduces positive experience.
Model 3b tests the moderating effect of perceived self-repairability on the influence of modular smartphone design on positive experience with repair service.Hypothesis 3b also cannot be confirmed due to the statistically insignificant moderation of perceived self-repairability (b = 0.092; p ≥ 0.1).The control variables exert similar effects on positive experience with repair service as in the previous model.
In sum, this analysis shows that Hypotheses 1, 1b, and 2 can be supported, while Model 2a shows the opposite effect than the one hypothesized (see Fig. 3).Hypotheses 1a, 2b, 3, 3a, and 3b cannot be supported.
Overall, these results have two implications.On the one hand, modular smartphones are, first, more likely to be self-repaired using   C. Amend et al. instructions than sent to SmartMod's repair service.This is positively moderated by perceived self-repairability (see Fig. 4).Higher perceived self-repairability increases the likelihood of using repair instructions compared to a professional repair service more significantly for modular models than for semi-modular models.This is visualized by the different slopes for modular and semi-modular models in Fig. 4. Second, modular smartphone design increases the user's positive experience with repair instructions, while strong CE attitudes reduce this positive effect.As shown in Fig. 5, the stronger the user's CE attitudes, the slightly lower is the positive experience made with repair instructions for modular models.For semi-modular models the slope is positive, meaning there is no difference.Thus, stronger CE attitudes increase the positive experience made with repair instructions for semimodular models.Apart from that, modular smartphone design does not influence positive experience with repair service.Consequently, modular smartphone design is particularly effective in promoting selfrepair.

Discussion
The following section discusses the results with regard to repair behavior, compares the choice for using repair instructions over repair service, and discusses the relevance of positive experience with both repair instructions and repair services.

General repair behavior
Comparing the use of repair options by the study's users with those of average EU citizens, SmartMod's users are more inclined to engage in circular behavior (Cerulli-Harms et al., 2018).Overall, this empirical analysis found that among smartphone users of the sustainability pioneer SmartMod, who tends to attract CE-oriented users, the readiness to repair (that is to self-repair or have the device repaired by a professional service) is very high, particularly when they own a modular smartphone.The general readiness to repair is comparatively high.Only 13% of modular smartphones and 15% of semi-modular smartphones were not repaired, while for conventional smartphones the figure was 55% (Laitala et al., 2021).This indicates that modularity facilitates successful repair behavior through easier repair.In line with the users' positive repair attitudes, those who did not self-repair had understandable reasons for not doing so (e.g., they could continue using their smartphone without a repair), which do not lessen the relevance or effect of the repair options.This shows that circular services and technical design complement each other well, enabling positive repair behavior.Additionally, the higher CE orientation of SmartMod users compared to average EU users (Cerulli-Harms et al., 2018) could also explain the very positive repair behavior.
Repair barriers of conventional smartphones found by Sabbaghi et al. (2016) were the unavailability of spare parts, repair tools, repair manuals, the expensiveness of spare parts, time needed, or complexity of repair.These barriers prevented only a minority of this study's respondents from repair behavior.Key reasons are that SmartMod applies a SMPD approach and offers self-repair support (instructions, screwdriver, etc.) and complementary services.Such complementary services include support that increases repair ease, such as availability of affordable spare parts and professional repair services.Taken together this provision of support promotes an extended product lifetime (Cooper, 2005).

Self-repair complemented by repair services
This study reveals that SMPD incentivizes self-repair using instructions compared to using SmartMod's repair service.However, it should be considered that defects under guarantee are probably more frequently repaired by SmartMod, which could slightly skew our results.Nonetheless, this finding extends previous research in that modular devices are more often self-repaired compared to conventional devices.For example, Laitala et al. (2021) found that slightly more than a third of all damaged conventional devices were self-repaired.Although the semi-modular devices in our sample were only self-repaired by 13%, 39% of modular devices were self-repaired, and an additional 13% was both self-repaired and sent to repair service.This study therefore reveals that SMPD can successfully enable easy and fast exchange of defective modules in practice (Bonvoisin et al., 2016;Schischke et al., 2017).Furthermore, Dewberry et al. (2017) find that a repair choice also depends on available product knowledge and technical information as well as on availability of spare parts.Because our research reveals that only few users refrained from repair because of these reasons, information asymmetries between this study's case OEM and their users are relatively small.In particular, we can extend previous research in showing that offering repair instructions and services reduces information asymmetries, which increases the number of repairs.Self-repair is especially incentivized through the provision of complementary repair instructions and support.Next to SMPD, these offers are crucially important since all users chose some kind of repair help to perform self-repair.
Previous research identified lack of confidence as a repair barrier (Dao et al., 2020).Thus, both modular design and the provision of repair instructions could create higher repair confidence.In addition, repair confidence could be leveraged through higher self-repairability perception of modular devices as well as positive previous repair experience, which is reflected in our sample's high CE attitudes.This study reveals that higher perceived self-repairability can strengthen the likelihood to choose to self-repair modular products using repair instructions.
With regard to future research, this analysis could inform studies of product-person relationships (Mugge et al., 2005;Page, 2014).If users have an emotional bond to a product, they are expected to take better care of the product, which would likely postpone product replacement and thus extend product lifetime (Schifferstein and Zwartkruis-Pelgrim, 2008).Self-repair may strengthen this product-person relationship because the user actively engages with the product's physical design.However, our findings show that more complex modules, such as the mainboard, are more often sent to SmartMod's repair service.A repair service does not therefore become obsolete, as some users do not have the skills, willingness, and time for self-repair, and some modules are less suitable for self-repair.As higher repair rates merely delay the C. Amend et al. replacement of products (Van Nes and Cramer, 2005), OEMs should consider additional circular activities.These will be outlined in the implications section.
Contrary to our expectations for this study and previous findings (Bigerna et al., 2021;Cerulli-Harms et al., 2018;Nazlı, 2021), CE attitudes are not found to influence the choice for self-repair.Possible reasons could be the strong positive CE attitudes in our sample of smartphone users and their positive general repair behavior, both resulting in low variance.However, OEMs can influence circular behavior not only through SMPD, but also through user engagement, active communication and by offering complementary services.A high perceived self-repairability can successfully incentivize self-repair behavior, which could be leveraged through communication methods such as nudges.Also, service offers, such as repair instructions (e.g.videos) can lower self-repair barriers.

Positive experience with repair instructions
This study complements earlier research by Gregson et al. (2009), Laitala et al. (2021), andSabbaghi et al. (2016), who identified the lack of repair instructions as a barrier to repair.For example, 28% of self-repairs were unsuccessful according to a consumer survey conducted by Laitala et al. (2021).First, this study confirms the importance of OEM-offered circular services, such as the provision of repair instructions.Second, it reveals that a SMPD approach increases positive experience with repair instructions.Positive experiences, in turn, help users understand and follow repair instructions for other modular products and future repairs.This emphasizes the importance of complementary product and process innovation to achieve SOI (Hansen et al., 2009).It highlights that vertically-integrated OEMs can successfully promote circularity and sustainability (Hansen and Revellio, 2020).In addition, an appropriate repair service can absorb some degree of technological incompatibility.Thus, it is critical to align the objectives of both product and service designs to unfold as much sustainability potential as possible.
Positive experience with repair instructions for modular products can be further increased by other factors, such as time, appearance, social norms, or previous care experiences (Ackermann et al., 2018).While CE attitudes reflect previous care experiences, this study found they reduce the positive effect of SMPD on experience with repair instructions.Although this finding contradicts previous findings (Ackermann et al., 2018;Cerulli-Harms et al., 2018;Nazlı, 2021), their research focused on environmental attitudes, and only implicitly on CE attitudes.Research by Cerulli-Harms et al. (2018), which has informed this study's CE attitude scale, found that although many users have positive attitudes toward CE, they actually use repair options relatively rarely.As Cerulli-Harms et al.'s (2018) research did not explicitly investigate the impact of CE attitudes on repair behavior, our study extends previous research as it additionally investigated the impact of a change in technical design on experience with repair instructions.Since SmartMod's users have especially strong repair attitudes compared to average EU citizens (Cerulli-Harms et al., 2018), their expectations about repair instructions may be rather high and consequently not met (Michalco et al., 2015).This may result in a less positive experience with repair instructions.Hence, OEMs must ensure that higher expectations of users with strong CE attitudes are met in order to promote satisfaction and successful future repairs.Furthermore, modular design may not be as important for users with already strong attitudes toward circular behavior, as those users make positive experiences with repair instructions, even with conventional smartphones.This mirrors the importance of coordinating product and service design, and aligning user experiences with both.
While CE attitudes are found to have a moderating effect, perceived self-repairability is not found to moderate the effect of a SMPD approach on experience with repair instructions.However, high perceived selfrepairability has a direct positive effect on experience with repair instructions.Therefore, to create a positive experience, it may be sufficient to have either a modular design or high self-repairability perception.The provision of convenient services to facilitate self-repair (i.e.easily understandable, well-designed repair instructions) to create high self-repairability perception could even lower repair barriers of conventionally designed products and support a positive experience made with repair instructions.

Positive experience with repair service
For the smartphone users of SmartMod, no impact of modular design could be found on their positive experience with repair services.As the majority makes use of SmartMod's in-house repair service and as thirdparty services were excluded from the analysis to reduce bias, the quality of repair services for both semi-modular and modular devices are comparable, with the result that the analysis revealed only small differences between the two groups.Hence, from a user perspective, technical design has no significant effect on experience with repair services.However, experience with repair as such is rather high, which may promote subsequent repair behavior and extend product lifetime (Ackermann et al., 2018;Cooper, 2004).In contrast to previous research that has identified frequent repair barriers, for example, high repair charges (King et al., 2006), such barriers are not relevant for the investigated sample of users.One reason may be that SmartMod's repair services are less expensive compared to their competitors.Also, no moderating effects of CE attitudes or perceived self-repairability could be identified.Further research could analyze the effect of SMPD on professional repair processes with regard to cost, duration, and quality of repairs.

Conclusions and implications
This study investigated the relationship between technological innovation and complementary circular service innovation from a user perspective to deepen understanding of SMPD's potential in promoting sustainability.This analysis specifically investigated (1) to what extent SMPD enables the use of instructions for self-repair of smartphones compared to using a professional repair service, ( 2) to what extent SMPD promotes a positive user experience with repair instructions and repair services, and (3) how CE attitudes and perceived self-repairability moderate these effects.The analysis applied logistic and multiple regression models based on survey data from a sample of 1,720 smartphone users of a sustainability-pioneer among smartphone OEMs.The empirical findings reveal that SMPD increases perceived selfrepairability of the device, which supports the use of instructions for self-repair compared to using a repair service.The findings also confirm that SMPD promotes a positive experience with repair instructions.Since CE attitudes were strong in the sample, they did not further increase positive experience with repair instructions.These findings have theoretical and practical implications.
A first theoretical implication for SOI and management research is the high relevance of complementary product and process innovation.SMPD encourages users of modular products to make more frequent use of repair instructions because of its repair-friendly product design and well-designed repair supportand it leads to more positive experiences with repair.Our results confirm earlier research findings that companies should not discourage self-repair by its users (Sabbaghi et al., 2016).Repair instructions should be complemented by individual repair support service (e.g.text, chatbot, hotline).Consequently, the sustainability potential of product innovation in technical design can only be leveraged through complementary circular business models and services (e. g., Bakker et al., 2014;Hansen et al., 2009).Our analysis shows that while modularization of products does not automatically contribute to extended product lifetimes, SMPD does however enable easier and more efficient repair and the provision of more convenient repair instructions.This supports DIY repair modularity, a type of SMPD (see Schischke et al., 2019a ande.g. Bonvoisin et al., 2016;Tseng et al., 2008).Well-designed repair instructions and accessible repair services can help to extend product lifetime as they incentivize users to repair instead of replace a damaged product.
Second, the combination of SMPD and high perceived selfrepairability enhances self-repair behavior.Product, services, and communication strategies must therefore be designed and aligned accordingly, and perceived repairability must be actively managed to promote repair activity.Besides well-designed instructions and products, choice architectures, such as nudges, present another opportunity to influence repairability perception by "organizing the context in which people make decisions" (Thaler and Sunstein, 2008, 428).For example, repairability scores or icons (Bovea et al., 2018) could function as perception nudges to influence repair behavior.
Third, CE attitudes affect user experience with circular services, such as repair instructions.This confirms earlier research on the impact of attitudes on behavior (Fogg, 2009;Mugge et al., 2005).However, expectations of repair services may be unmet for users with high CE attitudes (Michalco et al., 2015).Therefore, circular services must be well aligned to user CE attitudes to prevent dissatisfaction, and services may be as important as modular design.Different levels of CE attitudes should be considered in the design of repair instructions and products.For example, circular services could be customized based on the user characteristics to promote circular behavior (e.g. to undertake repair and make positive experiences with repair).The provision of two levels of repair instructionswith basic information provided for all users and additional advanced information for users with prior repair experiencecould prevent less positive repair experiences and potential disappointment.
Fourth, not only repair but also product replacement needs to be actively managed to leverage circularity and sustainability potential, since although repair extends product lifetime it does not prevent but only delays product replacement (Van Nes and Cramer, 2005).Hence, value chain actors should implement additional circular services, such as deposit systems for product take-back (Hansen and Große-Dunker, 2013), or managed secondary markets for used and refurbished devices.Overall, the implementation of a product-service system can alleviate circularity barriers.Companies could offer smartphones or other consumer electronic devices as a service for a monthly fee.In such a business model, the product is returned to the manufacturer at the end of the contract period and can then be re-used by others (Zufall et al., 2020).Since users are found to be more willing to engage in repair behavior if costs are covered by the manufacturer (Mugge et al., 2005), repairs and other circular measures could be included in the service contract.

Limitations and further research
This study does not come without limitations.First, the survey is based on smartphone users of a sustainability pioneer in the industry.Hence, respondents are likely to have stronger sustainability attitudes and be more prone to repair than conventional users.As the company already offers well-designed circular services, even their semi-modular smartphone models are easier to repair than conventional products (iFixit, 2020).Therefore, the technical differentiation (i.e., semi-modular versus modular smartphones) might be too small in the case of a sustainability pioneer.Future research could survey users of more conventional smartphone manufacturers to compare user repair behavior with larger design differences (i.e., between non-modular and modular devices).Second, although previous repair experiences may affect future repair behavior (Hielscher and Jaeger-Erben, 2019;Sabbaghi et al., 2016), this study did not explicitly ask users for this information.Although repair attitudes were also measured by items in the CE attitude scale, future research should further investigate the impact of previous repair experience on the effect of SMPD on repair behavior and experience.Also, the CE attitude construct could be adapted to achieve a higher Cronbach's alpha.Additional variables such as social norms could be tested (Ackermann et al., 2018).Third, future studies could adapt the survey to investigate whether repair services are used in case self-repair fails, as this would allow conclusions to be drawn for the improvement of repair instructions and services.Fourth, actual product lifetime was not analyzed in this study due to concerns about response bias and different smartphone release times.Comparing non-modular and modular products released at a similar time could therefore be worthwhile in future research.Fifth, after this study was conducted, the Joint Research Centre (JRC) of the European Commission published a study proposing a repairability score for smartphones (Spiliotopoulos et al., 2021), which differs slightly from the scoring system used in this study (iFixit, 2020).The JRC study identifies repair criteria for product information (e.g.comprehensiveness, availability, duration, and access price) (Spiliotopoulos et al., 2021), which overlap with the ones used in this study for experience with repair instructions.A revised survey based on JRC recommendations would be interesting for future research.Sixth, future research could also investigate the effect of other types of process innovation complementing modular products, for example, environmental labeling (Bakker et al., 2014) and product-service systems including guarantee and repair services during contract lifetime (Bakker et al., 2014;Mugge et al., 2005).Finally, user customization of products and/or complementary services could be studied to test its impact on the circular behavior of users with different CE attitudes.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.Research designeffect of modular smartphone design on repair behavior and experience as moderated by CE attitudes and perceived self-repairability.
because at least one of the variables is constant.

Table 3
Socio-demographic composition for each hypothesis model.
a Centralized values are used.b Based on F-test.
a Centralized values are used.bBasedon F-test.C.Amend et al.
Unstandardized regression coefficients are displayed.Standard errors are provided in brackets.Levels of significance: ** p < 0.01; * p < 0.05; † p < 0.1.a Centralized values are used.b Based on F-test.