An integrated importance-performance analysis and modified analytic hierarchy process approach to sustainable city assessment

Sustainable urban development has been a popular subject in urban studies and related disciplines. Owing to the challenges faced by cities worldwide to accommodate the growing urban populations, it is becoming ever more important for innovative research on sustainable urban development to be performed to help cities achieve sustainability. This study develops and tests an integrated approach to sustainable city assessment, which is a combination of importance-performance analysis (IPA) and modified analytic hierarchy process (AHP). Questionnaires designed following the IPA concept were distributed to residents of three cities. The importance scores from the collected data were factorized and the factors’ relative scores were then calculated using a formula developed in this study to represent pairwise comparisons. The derived criteria weights were applied to the performance scores to evaluate the cities’ relative overall sustainability performance. This approach replaces the AHP’s 1–9 scale with the IPA’s importance rating scale, which is a Likert scale, in the questionnaire. Based on the findings, implications and future research suggestions were provided.


Introduction
In 2015, the United Nations (UN) introduced the 2030 Agenda for Sustainable Development which consists of 17 Sustainable Development Goals (SDGs) and 169 targets to build on the previous Millennium Development Goals and to promote global sustainable development (UN 2015). Sustainable development is defined as "development that meets the needs of the present without compromising the ability of future generations to meet their own needs" (UN 1987). The SDGs guide nations to achieve sustainability through integrated and balanced developments in the economic, social, and environmental dimensions. Goal 11 of the SDGs titled "Make cities and human settlements inclusive, safe, resilient and sustainable," which concerns sustainability issues in human settlements as a result of urbanization, has propelled over 150 countries to develop national policies for sustainable urban development (UN 2018b).
More than half of the global population is living in the cities today and the proportion is expected to increase to 68% by the year 2050 (UN 2018a). As highlighted in the UN's (2017) New Urban Agenda, economic, social, cultural, and environmental impacts are intensifying in cities due to the growing urban populations. Therefore, efforts to ensure sustainable urban growth have to be enhanced to safeguard the wellbeing of urban residents and habitats both in the present and in the future. Urban residents should have access to safe living conditions, affordable housing, sufficient public infrastructure and facilities, clean air and water, and other basic needs to live comfortably in the city (Chan 2020). The UN's slogan for the New Urban Agenda, "Cities 2030, cities for all," signifies that by the year 2030, cities should be environmentally, socially, economically, and culturally sustainable and inclusive with equal benefits and opportunities for all inhabitants, including the present and future generations, to live in sustainable environments and contribute toward the quality of life of all (UN-Habitat 2017).
The differences between cities in terms of landscape, finance, culture, infrastructure, and others would require cities to have customized approaches to sustainable development to suit their needs and conditions. Using multiple criteria decision making (MCDM) techniques, the perceived performances of cities based on various criteria of sustainable development can be measured to facilitate decision making tailored to the cities' development needs. MCDM methods have often been applied in research related to sustainable development (Mardani et al. 2015). The analytic hierarchy process (AHP) by Saaty (1980) is a popular MCDM technique that has been employed in numerous urban sustainability research (e.g., Bunruamkaew and Murayam 2011;Chan and Lee 2019b;Ma et al. 2020;Zhang et al. 2012). Importanceperformance analysis (IPA), a method introduced by Martilla and James (1977) to measure the importance and performance of attributes, has been combined with AHP for analyses in urban development studies (e.g., Akbar et al. 2018;Lirn et al. 2012;López et al. 2019), among other disciplines. However, none has modified the AHP's pairwise comparisons using importance scores from the IPA to measure urban sustainability.
This research develops and tests an integrated IPA and modified AHP approach called the IP-AHP to assess the sustainability of cities from the residents' perspectives in terms of four sustainable development dimensions: economic, social, environmental, and cultural. Using the integrated approach, questionnaires were designed according to the IPA concept and distributed to residents of three cities. From the collected data, the importance scores were first subjected to factor analysis and then the factors' relative scores were calculated using a formula developed in this study to represent pairwise comparisons. The derived criteria weights were applied to the factors' performance scores to gauge the cities' overall sustainability performances compared to each other. Results obtained from the IP-AHP method were discussed along with the IPA's findings to offer suggestions for the cities' sustainable development.

Sustainable urban development assessment
Sustainable urban development has been a popular subject for research in urban studies and related disciplines, especially in recent years (Lee et al. 2021). Wang et al. (2019) performed a bibliometric analysis of over 2,500 articles on sustainable city published between 1992 and 2016 that were retrieved from the Science Citation Index Expanded and Social Science Citation Index databases and found that the number of research on sustainable city increased exponentially between 2004 and 2016. Owing to the double-edged sword of urban growth, e.g., the rising environmental and social issues in the face of economic development, as well as the growing urban populations worldwide, countries and cities are increasingly met with challenges to accommodate the needs of urban residents (UN 2018a;Wang et al. 2019). Therefore, it is becoming ever more imperative for research on sustainable urban development to be performed to help cities gauge their progress toward sustainability and implement sustainable development measures.
Many studies have established measures to assess urban sustainability, using either objective indicators (e.g., Al-Thani et al. 2018;Chan and Lee 2019a;Ma et al. 2020;Peng and Deng 2021), subjective indicators (e.g., Afacan 2015; Lee et al. 2021;Ouria 2019;Senlier et al. 2009), or both (e.g., Chan and Lee 2019b;Mouratidis 2017;Zegras et al. 2004). Objective indicators include measurements using official data, such as population density, employment rate, household income, pollution index, green area ratio, and so on, whereas subjective indicators involve perceptions or evaluations by people, such as the residents' perceived performances of their cities and their satisfaction with the cities or aspects of the cities with regard to sustainable urban development. While the objective indicators are useful for objective sustainability analysis and findings, they may not necessarily represent or project the thoughts of the people.
Urban residents' opinions should be taken into consideration in city development matters. Researchers and policy makers have stressed the importance of public participation in decision making for sustainable urban development (Afacan 2015;Musa et al. 2017;UN 2015). Urban residents have primary experiences with the cities they live in and they are among the most affected by the developments, or the lack thereof, in the cities. Therefore, their perspectives are crucial for the planning and implementation of urban policies. In Malaysia, for instance, the urban government practice a "people first" principle as they believe that in order to build sustainable cities, they must begin with the people (Federal Department of Town and Country Planning [FDTC] 2016). In Europe, the European Commission (EC) examines the quality of life in European cities every 3 years beginning 2004 for the purpose of urban development by surveying the cities' residents (EC 2016). Public perceptions about the cities they live in offer useful insights for the local authorities to identify urban problems, develop sustainability measures, and monitor their performances (Lee et al. 2021). In order for a city to develop sustainably, there should be a balanced growth in the four sustainable development dimensions, i.e., the economic, social, environmental, and cultural dimensions of the city.

Economic sustainability
Economy is a major sustainable development dimension that fuels urban growth. Urban residents' financial health indicators such as cost of living and household income are part of the economic sustainability of cities (Lee et al. 2021;Zenker et al. 2013). Overall cost of living includes the amount of money that the urban residents have to spend on goods and services, housing, food, and other things to live in the city. Housing affordability and adequacy are among the important gauges of urban sustainability mentioned in the SDGs' Goal 11 and often studied in sustainable urban development literature and reports (Chan 2020;Koch and Ahmad 2018). Cities should also have ample of employment opportunities enough to support the needs of the urban populations (EC 2016;Senlier et al. 2009). The residents should have equal access to employment and other opportunities in the city without any form of discrimination, e.g., between different genders and ethnicities (Afacan 2015;Zegras et al. 2004). The diversity of economy, businesses, or industries in a city is also a sign of economic prosperity and sustainability (Andereck and Nyaupane 2011; Ouria 2019)

Social sustainability
Social factors such as public facilities and services that promote the quality of life of urban residents are necessary for the sustainability of cities, for example, public transportation, recreational and sports facilities, educational facilities, and health care facilities (Chan and Lee 2019a;Lee et al. 2021;Musa et al. 2017). Good mobility and accessibility to places in the city, such as with wellconnected roads and transport system, help residents to access such facilities easily (EC 2016; Kakar and Khan 2020). More importantly, residents of all demographic groups, including, in particular, the elderly, women, children, and disabled people, should have the same access to the public facilities and services (Koch and Ahmad 2018;UN 2018b). Health care facilities should be of good quality and sufficient to treat residents in need to ensure public health among the urban community, especially in times of pandemic. There should be adequate retail sites and eateries to accommodate the residents' needs for resources daily (Mouratidis 2017), along with continuous supply of clean water and electricity in the city (Chan and Lee 2019a). The safety and security of residents is another important aspect of social sustainability (Afacan 2015;Ouria 2019;Senlier et al. 2009). Criminal, vandalism, and drug and alcohol abuse cases should be prevented and controlled at all times (Andereck and Nyaupane 2011;Howley et al. 2009). Urban communities should have good relations with each other, preferably with social support networks that keep the neighborhoods safe and healthful (Afacan 2015;Mouratidis 2017).

Environmental sustainability
Environmental quality is another important dimension of sustainable development (Gołaś et al. 2019). Environmental pollutions such as air, water, and noise pollutions are commonly used indicators to gauge environmental sustainability in cities (EC 2016;Senlier et al. 2009). It is necessary for pollution levels to be tolerable by the urban inhabitants without affecting their health and daily activities. Also, there should be efficient waste management systems to ensure that solid wastes are collected regularly and minimized in the city (Chan and Lee 2019a;Hoang et al. 2019). Cities should have overall clean and tidy environments and esthetic qualities (Afacan 2015;Mouratidis 2017). Natural environment, such as green spaces, water bodies, and trees, should be well preserved and sufficiently available in the city (Lee et al. 2021;Yilmaz et al. 2020). There should be adequate open and public spaces, whether natural or artificial, for the use of all residents. Apart from that, built environment like streets and buildings should be well maintained and resilient (EC 2016; Koch and Ahmad 2018). Other important factors affecting the urban environment such as traffic congestions and crowdedness of people in the city should be also reduced and controlled (Chan and Lee 2019a;Lee et al. 2021).

Cultural sustainability
The preservation of cultural and historical sites, buildings, and heritage contributes to the cultural sustainability of cities (Chan and Lee 2019a; Koch and Ahmad 2018; Lee et al. 2021). According to target 11.4 of the SDGs' Goal 11, efforts to conserve and protect cultural and natural heritage should be strengthened (UN 2015). Cultural facilities such as concert halls, theaters, museums, and libraries should be amply provided in the city (EC 2016;Senlier et al. 2009), along with cultural entertainments like arts, fairs, and festivals (Andereck and Nyaupane 2011; Lee and Xue 2020). There should also be cultural harmony and the presence of community sense of belonging among the urban residents (Afacan 2015;Musa et al. 2017).

Modified and integrated AHP
AHP is a popular MCDM technique applied in the literature (Alonso and Lamata 2006;Mardani et al. 2015). The method has been combined with IPA to produce modified IPA grids that replace the importance ratings from the IPA with scores obtained from the AHP's judgment matrix (Lirn et al. 2012;López et al. 2019;Wang et al. 2016). Lirn et al. (2012) also replaced the IPA's performance ratings with weighted performance scores from the AHP. Some researchers that combined both methods further modified the AHP with the fuzzy method (Chang et al. 2015;Hsu et al. 2015;Wang et al. 2016). Wang et al. (2016), for instance, used the fuzzy AHP method to evaluate the extent to which Hongshan Zoo in China is a smart tourist attraction. Besides that, the AHP has also been integrated with other methods in the science field such as the technique for order of preference by similarity to ideal solution (TOPSIS), data envelopment analysis (DEA), quality function deployment (QFD), and more (Emrouznejad and Marra 2017).
While there have been multiple approaches established in the literature to modify or integrate the AHP with other methods, none, except for Pak (2013), has modified pairwise comparisons in the AHP by way of replacing judgments on the 1-9 scale with computations of relative importance scores using ratings obtained from the IPA. Pak (2013) suggested a method to produce pairwise comparison by dividing the average importance score of one attribute by another. This research, however, applies a more robust set of formulas that takes into account the range of the AHP's 1-9 scale in order to achieve the method's consistency ratio, which is important to ensure the reliability of the outcomes from the integrated approach.
The integrated IP-AHP approach developed in this study could potentially be an improved version of the conventional AHP, which has been reported to have its disadvantages (Maleki and Zahir 2012;Nefeslioglu et al. 2013). The Likert scales used in the questionnaire of this approach, instead of the 1-9 scale which requires pairwise judgments, makes it easier for the survey to be implemented on the general public as they could answer the questions without the need to be trained. This can facilitate researchers to acquire larger sample sizes that are more representative of the population compared to the usually small sample sizes used in AHP and fulfill the required sample sizes in case of high variances in expected weights (Melillo and Pecchia 2016). Moreover, the use of a smaller scale coupled with the factorizing of subcriteria reduces the possibility of obtaining inconsistent results, which often increases as the number of measurements increases (Asadabadi et al. 2019;Bafahm and Sun 2018). This also reduces the likelihood of the need to repeat the judgment or survey process occurring when an inconsistent result is found. The use of the factor analysis to remove subcriteria with low loading from the analysis also helps to reduce the rank reversal effect of the AHP (Lirn et al. 2012).
Using the IP-AHP approach, cities' sustainability performance along with the levels of importance of urban sustainability indicators can be gauged through the residents' perspectives. The overall weighted performances scores of cities can be compared with each other to determine the ranking of the cities with regard to their sustainable development progress. In line with the concept of utility/welfare maximization (Varian 2014), this approach could help governments to optimize the use of resources for urban development by allocating them in proportion to cities' relative performance to ensure efficient and balanced developments of the cities in pursuit of sustainability.

Development of survey instrument
To build the questionnaire for this study, measurement items for the four sustainability dimensions described in the previous section were adapted from relevant urban literature as laid out in Table 1. Section 1 of the questionnaire asked the respondents to rate the importance of the items to them on a 5point scale from 1 = not at all important to 5 = extremely important. Section 2 asked the respondents to rate their city of residence's performance in terms of the items on a scale of 1 = poor to 5 = excellent. The final section enquired the respondents' demographics.
A draft questionnaire was developed and sent to a few experts in the field for feedback, following which some adjustments were made. A pilot study was then conducted on 30 residents to examine the face validity and reliability of the measures. Sample size for the pilot study was calculated following Viechtbauer et al.'s (2015) formula to calculate sample size in pilot studies by using a 95% confidence level and a 0.90 probability that problems would not arise in the study. Greatest lower bound to reliability (GLB) was computed on the importance scores collected from the pilot study using R and a resulting value of 0.82 indicated that the measurement set was reliable.

Study areas
Three cities in Malaysia, a developing country in Southeast Asia, were selected for this study, namely Kuala Lumpur, Seberang Perai, and Melaka City. Kuala Lumpur is the capital city and a federal territory of Malaysia, Seberang Perai is the largest city in Penang, a northwestern Malaysian state, while Melaka City is the capital city of Melaka, a state in southwest Malaysia. All three cities/states are situated at the west coast of Peninsular Malaysia where most of the country's urbanization takes place. Kuala Lumpur, Penang, and Melaka are among the most densely populated, urbanized states in the country (Department of Statistics Malaysia [DOSM] 2011). Since the urban population in Southeast Asia is expected to grow in the coming years, it is important for developing countries in the region to ensure urban sustainability (Chan 2020).
Kuala Lumpur covers an area of 243 km 2 with an estimated population of 1.77 million (DOSM 2020). The city, with its 11 subdivisions, is managed by the Kuala Lumpur City Hall. Seberang Perai is one of the two parts of Penang that is  Lee et al. 2021, Musa et al. 2017, Zenker et al. 2013 Reasonable prices of goods and services Andereck and Nyaupane 2011, Lee and Xue 2020, Zenker et al. 2013 Adequate number of affordable houses Lee et al. 2021, Musa et al. 2017, Senlier et al. 2009, Zenker et al. 2013 Diverse economy/businesses in the city Andereck and Nyaupane 2011, Lee et al. 2021, Ouria 2019 Ample of employment opportunities EC 2016, Chan and Lee 2019a, Lee et al. 2021, Senlier et al. 2009 Equal opportunities between residents † Afacan 2015, Koch andAhmad 2018, Zegras et al. 2004 Social dimension Safety and security in the city Afacan 2015, EC 2016, Lee et al. 2021, Senlier et al. 2009 Quality educational facilities and services EC 2016, Lee et al. 2021, Musa et al. 2017, Senlier et al. 2009 Quality health care facilities and services EC 2016, Lee et al. 2021, Musa et al. 2017, Senlier et al. 2009 Continuous access to clean water and electricity Adequate cultural facilities, e.g., museums and theaters EC 2016, Lee et al. 2021, Senlier et al. 2009 Ample of cultural entertainments and activities Andereck and Nyaupane 2011, Lee et al. 2021, Zenker et al. 2013 Cultural diversity and harmony † Andereck and Nyaupane 2011, Musa et al. 2017, Zenker et al. 2013 † Item was removed due to low factor loading situated on the mainland separated from Penang Island. Administered by the Seberang Perai City Council, the city consists of three districts that stretch across 751 km 2 with a total population of about 840,000 recorded in the decennial census (DOSM 2010). Melaka city is a 66-km 2 town under the authority of the Melaka Historic City Council, which also governs other parts of Central Melaka District. Formerly a mining city, Kuala Lumpur has progressed over the years to become one of the nation's most economically developed cities, prompting the city council to implement sustainable development measures to promote the wellbeing of residents, with efforts to provide sufficient green infrastructure in the densely populated city (Danjaji and Ariffin 2017). The country's industrialization policy has also led to growing urbanization in Seberang Perai, which has transformed rural into urban areas, bringing about significant landscape and socioeconomic developments in the large city (Samat et al. 2014). The Melaka City, a developing city with centuries-old history, is recognized as a World Heritage by the United Nations Educational, Scientific and Cultural Organization (UNESCO) for its well-preserved, unique multicultural heritage with influences from the Malay, Portuguese, and Dutch cultures.

Data collection
A cross-sectional, cluster sampling approach was used to administer the questionnaires randomly to clusters of residents in the three cities identified based on geographical regions in the map. The required sample size was calculated using confidence interval at 95% confidence level with 50% estimated variance and 5% margin of error, which gave a minimum total of 385. About 1,000 questionnaires were distributed to residents in the three cities from 22 November to 13 December 2020 by six trained surveyors. Altogether, 623 completed questionnaires were received from the survey, of which 32 were excluded due to multiple/missing responses, resulting in 591 valid for analysis with an effective response rate of 59.1%.

Data analysis
Firstly, exploratory factor analysis (EFA) was performed on the mean importance scores using principal component analysis with promax rotation on SPSS Statistics. Next, confirmatory factor analysis (CFA) was conducted on SPSS Amos to test how well the identified factors from the EFA fit the data. The relative importance scores of the factors were then calculated using the following formulas: on the condition that a j is greater than a k , where a j is the importance score of the jth criterion, a k is the importance score of the kth criterion, n is the number of points in the Likert scale used in the study, x jk is the score derived for the jth criterion relative to the kth criterion, and x kj is the score derived for the kth criterion relative to the jth criterion.
The criteria being evaluated here are the factors. A simple rule of the equations is that when comparing between two criteria, the jth criterion must have a larger importance score than the kth criterion, as the score of the less important criterion, x kj , should be the inverse of the score of the more important criterion, x jk . In the event that a j is equal to a k , then x jk and x kj should each be equal to 1 because in AHP, criteria with the same importance levels are given the value of 1. Also, x jj should be equal to 1 for all j. The reason 1 point has to be added to the difference between the importance scores in Eq. 1 is that the number 1 is the beginning of either side of the spectrums in a pairwise comparison scale with no absolute zero, so the difference should take into account the value of one.
The divisor in Eq. 1 (n − 2) is the range of the IP-AHP's scale. Two points have to be deducted from the Likert scale used in the study because (1) assuming a 5-point scale is used, the difference between a j and a k would be at most 4 points as respondents must select an integer from 1 to 5 for each question, and (2) further laying out the 4 points in 2 spectrums in a pairwise comparison scale with the value 1 in the middle would leave either side of the spectrums with only 3 points. The multiplier value, 8, represents the range of the original AHP's 1-9 scale.
The measures above were applied to each pairwise comparison between the criteria or factors until a complete comparison matrix, m×m, was established, where m was the number of criteria evaluated. The succeeding steps followed the AHP's procedures for normalization, weight calculation, and consistency analysis. Finally, the cities' performance scores from the IPA in terms of the factors were normalized and multiplied by the criteria weights, and the overall sustainability performances of the cities were compared with each other. An IPA grid was plotted to illustrate the importance and performance scores by the respective cities' residents, with the grand means of the scores as crosshairs in the grid. Table 2 shows that about half of the respondents were males while the other half were females. More than a third of the respondents belonged to the 20-29 age group (37%). Nearly half of the respondents had attained a Bachelor's degree (47.9%). Kuala Lumpur residents constituted 40.9% of respondents, 31.3% represented Seberang Perai, and 27.8% were from Melaka City. As much as 43.5% had been in their city of residence for over 20 years.

Results
The reliability of the measurement model, calculated using GLB, was good at 0.84. The resulting Kaiser-Meyer-Olkin value of 0.91 and a significant Bartlett's test (p < 0.001) indicated that the data set was suitable for factor analysis. The EFA produced six factors from the four sustainability dimensions and each item's factor loading was above 0.50 (see Table 3). Six items, one each from the economic and cultural dimensions and two each from the social and environmental dimensions, were removed due to low factor loadings. "Economy" represented the economic dimension as a whole, "wellbeing" and "convenience" were part of the social dimension, "conservation" and "ambience" constituted the environmental dimension, and "culture" stood for the cultural dimension. Analysis from the CFA showed satisfactory fit of the model to the data with chi-square/degrees of freedom (χ2/df) = 2.33, goodness of fit index (GFI) = 0.94, comparative fit index (CFI) = 0.95, root mean squared error of approximation (RMSEA) = 0.05, standardized root mean square residual (SRMR) = 0.05, normed fit index (NFI) = 0.92, and Tucker Lewis index (TLI) = 0.94.
The correlation outcomes with measurements of average variance extracted (AVE) and composite reliability (CR) as exhibited in Table 4 showed that each factor's AVE and CR levels were above 0.50 and 0.70, respectively, thus indicating acceptable convergent validity with the existence of internal consistency (Hair et al. 2014). The larger square root of AVE than the inter-construct correlations of each factor confirmed discriminant validity. Bivariate correlations between the factors that were all under 0.70 signified the absence of redundant factors.
Using the IP-AHP calculation method described above, the importance scores were converted into pairwise comparison scores, from which the normalized weights were derived. The consistency analysis performed on the data set produced a ratio of 0.096 (≤ 0.10) with λmax = 6.60, consistency index = 0.12, and random index = 1.248 (Alonso and Lamata 2006), implying that the matrix was reliable with acceptable consistency. Table 5 exhibits the weighted performance evaluation between the three cities. The city that performed the best in urban sustainability was Melaka City (0.344), followed very closely by Seberang Perai (0.341), while Kuala Lumpur had the lowest score (0.315). The IPA grid further shows the importance and performance ratings of the cities in terms of the factors (see Figure 1).

Discussion and conclusions
Findings from the IP-AHP revealed that Melaka City and Seberang Perai performed better than Kuala Lumpur in sustainable urban development. Both cities are less dense compared to Kuala Lumpur, which is the densest in the nation with 7,299 persons per square kilometer (DOSM 2020). Kuala Lumpur garnered the lowest performance scores for the environmental, economic, and social dimensions between the cities. Items under the environmental "ambience" factor, i.e., pollutions and traffic congestion, were among the worst rated ones by the residents of the city; all of which fell below the average point. In fact, for all cities, the overall environmental scores lied within the low-performance quadrants in the IPA grid, Q1 and Q3. Environmental issues have been some of the biggest challenges faced by cities nowadays (Al-Thani et al. 2018;Koch and Ahmad 2018;Lee et al. 2021), and the Malaysian cities are no exception. Developing countries should give more priority to environmental issues (Chan 2020). With the expected ongoing urbanization in the country, the government should put more effort and investment into curbing pollutions and congestions in the cities, especially in the densely populated cities where the problems are more severe. For examples, they could enforce tighter laws to monitor and control pollutions, and implement smart technology to reduce traffic congestions.  Although the environmental "conservation" factor was found to be a low priority in this study, it is nonetheless important to conserve the natural environments of cities. Environmental impacts such as deforestation could increase carbon emissions in the city (Arshad et al. 2017). Urban green spaces, for instance, are lacking in Kuala Lumpur and Seberang Perai due to the scarcity of land (Maryanti et al. 2017;Samat et al. 2014), which could be even more so in the future as rapid developments continue. The ratio of green spaces to population in Kuala Lumpur is only 12 m 2 per person, which is below the standards set by World Health Organization and Malaysia's National Urbanization Policy (FDTC 2016). Therefore, the conservation of existing urban green spaces along with innovative incorporation of green spaces into built environments is necessary for the cities' environmental sustainability.
Kuala Lumpur also fell short in terms of the economic dimension. In what could be the most economically developed city of the country, the residents of Kuala Lumpur thought that the city was not economically sustainable, at least not for them. Items under the dimension such as cost of living and housing affordability were rated below the average point. The economic performances of Melaka City and Seberang Perai were not so well off either, with each scoring just above average. The citizens' slow income growth rate that does not match the inflation rate and the increase in property prices till unaffordable levels are among the major factors that contribute to high cost of living in the country (Surendran 2019). Therefore, the government has to provide adequate economic support for the lower-income residents in the cities, such as affordable home ownership schemes, subsidized goods and services, and lower barriers for economic activities. The recent  declaration of federal territories in the country as temporary free trade areas is a good example of economic support for the locals (Choong 2020). Other than that, public transportation systems have to be improved to allow people to live in the suburbs, which are more affordable, and be able travel to the city centers to work with convenience. Such effort is important to help develop the suburbs as well (Shao et al. 2019).
With regard to the social dimension, Kuala Lumpur performed the worst in the "wellbeing" factor but the best in the "convenience" factor compared to the other two cities. Regardless, the social performances of all three cities fell within the high-performance quadrants in the IPA grid, Q2 and Q4, with the "wellbeing" factor being the most important of all. These imply that the cities have to keep up the good work in promoting their residents' wellbeing, such as ensuring safety and security in the city and providing quality health care and educational facilities. Although the means of "convenience" belonged to Q4, which signified possible overkill, the factor should also be taken seriously as good mobility and accessibility to places in a city is a crucial element of urban sustainability (Afacan 2015;UN 2015). Overall, the more favorable finding for the social dimension compared to the other dimensions in this study reflects the upmost focus of the Malaysian government in the urban sector, which is "people-oriented approach and community wellbeing," as reported in UN-Habitat (2017 p. iv).
As for the cultural dimension, which was the least important factor in this study, Melaka City performed better than the other two cities. Melaka City is in fact a culturally rich city that is a UNESCO World Heritage site, as described earlier. While this factor might not be that important to the residents, culture is still a necessary element of sustainable urban development (Senlier et al. 2009;UN 2015). The cultural heritage of a city or place should be conserved for the continuation of the community's long-standing identity, pride, and image (Lee and Xue 2020). A city's vibrant culture could be a factor for tourist attraction as well, driving economic growth and cultural heritage development, as in the case of Melaka City (Teo et al. 2014). Therefore, the local authorities have to create awareness on the importance of local culture and heritage among the current generations, so that they can be preserved well for the use of the future generations.

Contributions and implications
Theoretically, this paper contributes to the literature by introducing a new method to modify AHP by integrating it with IPA, called the IP-AHP, which measures pairwise comparisons in AHP using importance scores from the IPA. Compared to a majority of the previous studies that combined the AHP and IPA (e.g., Lirn et al. 2012;López et al. 2019;Wang et al. 2016), this study's integrated approach is different as it replaces judgments on the AHP's 1-9 scale with computations of relative importance scores using ratings obtained from the IPA. Unlike Pak (2013), the formulas developed in this study take into account the range of the 1-9 scale in order to achieve the consistency ratio in AHP, which is necessary to ensure the reliability of the modified approach. Also, in contrast with previous studies that employed the experts' judgments on the importance of criteria using the 1-9 scale (Chang et al. 2015;Hsu et al. 2015;Lirn et al. 2012;López et al. 2019), the use of the IPA's Likert scale in the questionnaire of this approach makes it easier for the survey to be implemented on the general public, who could answer without being trained. This could allow researchers to acquire larger sample sizes that are more representative of the population, which could also help to fulfill the required sample sizes in case of high variances in expected weights (Melillo and Pecchia 2016). Coupled with the reduced scale in the questionnaire and the operation of factor analysis to remove unnecessary subcriteria, the validity and consistency of the findings could be enhanced, thus minimizing the disadvantages of AHP (Asadabadi et al. 2019;Maleki and Zahir 2012;Nefeslioglu et al. 2013).
The application of the IP-AHP method in this research to assess urban sustainability for the purpose of welfare maximization further contributes to sustainable development and urban literature. This research has shown that, other than using the AHP to make decisions by way of selecting the best alternative, the method can be used to compare the relative performance of cities to facilitate decision making for optimized sustainable developments of the cities. Findings from the IP-AHP have also demonstrated the ability of the urban residents to provide meaningful information on the importance level of the sustainable urban development indicators as well as the performance of cities based on the indicators. Beyond that, the indicators compiled in this paper can be adopted and applied in future urban development research.
On the practical side, this paper has established a method for the local authorities to measure and monitor the sustainability of cities, which can facilitate balanced and efficient implementations of sustainable urban development practices. It is recommended that the residents are surveyed every now and then by the authorities to gauge their changing perspectives and needs over time as the cities progress. Results from the IP-AHP method could help the government to decide the proportion of resources to be allocated to the examined cities based on their relative performance. The IPA's findings could also enhance the decision making by providing further insights into the cities' performances in terms of the indicators. Since the survey is easy to be implemented on the general public, the method is suitable not only in academia, but also in organizational practices.
Specifically, the findings of this study apply to the three examined cities in Malaysia. The analysis shows that Melaka City and Seberang Perai performed better overall compared to Kuala Lumpur, which was taught to be more polluted, congested, and costly to live in among the residents. Nevertheless, the former two cities' sustainability performance were not that well off either, as the cities' overall weighted scores and dimension scores were not far apart from each other. Also, even though the IPA ratings were categorized into quadrants of high and low, importance and performance in the grid, the importance ratings were generally within the higher range of above 4.0, whereas the ratings of the cities' performances were all below 4.0. It can be concluded that much more work has to be done by the authorities of the cities to fulfill the needs and expectations of the residents. Further to that, developments in the cities have to be managed strategically in ways that ensure balanced growth in the sustainable development dimensions.

Limitations and future research recommendations
The measurement items used for the survey in this study were adopted from the previous literature and conceptually arranged according to the dimensions. Future research may conduct a preliminary study, such as using the Delphi method, to identify more sustainable urban development indicators relevant to the topic being studied. This research introduces a new, integrated approach to measure the sustainability of cities from the residents' perspectives. Future research could use the same method on the urban experts to gain their perspectives which may add value to the outcomes. The sample of this study includes a total of close to 600 residents from three cities. Future research may interrogate a greater number of respondents from each city to obtain more accurate findings regarding the individual cities. The study locations of this research are three cities in a single country. Future research could assess cities in different countries simultaneously to compare the sustainable development of cities of different nations and provide more substantial recommendations for improvement. Future research in other disciplines could also apply the IP-AHP method to study other entities' relative performance.
Author contribution All authors contributed to the study. Conceptualization, material preparation, data collection, and analysis were performed by Say Wah Lee and supervised by Ke Xue. The draft of the manuscript was written by Say Wah Lee and Ke Xue commented on it. Both authors read and approved the final manuscript.

Declarations
Ethics approval Not applicable Consent to participate Not applicable

Consent for publication Not applicable
Competing interests The authors declare no competing interests.