Exploring a Conceptual Framework of Koreans’ Residential Satisfaction Based on Maslow’s Human Needs: A Qualitative and Quantitative Integrated Study

Abstract

Previous studies on residential satisfaction factors (RSFs) overlooked residents’ psychological needs. To bridge this gap, we linked RSFs to the Modified Maslow’s Hierarchy of Human Needs (MMHN) through a three-step qualitative and quantitative integrated analysis. First, RSFs were derived from the analysis of previous studies. Second, through an analytic hierarchy process and a focus group interview, RSFs and the MMHN were linked. Third, the MMHN-based model was applied to data from the 2020 Korea Housing Survey, and classification and regression tree analysis were performed to derive significant factors, which were then compared to significant factors from the conventional model. The comparative analysis results of the conventional and MMHN-based models were as follows: (1) The MMHN-based model presented significant factors in all five stages of human needs, reflecting various human needs. (2) The MMHN-based model assessed the housing needs of residents in the non-capital region in more detail. (3) The MMHN-based model investigated the differences in residential satisfaction factors between metropolitan and non-metropolitan regions more clearly. (4) Two factors influencing safety needs important in all regions in both models were deemed crucial for residential satisfaction. This study could support the customization of regional housing policies according to unique needs and life circumstances by region.

1. Introduction

As interest in residences has increased since the social disaster of COVID-19, research on the quality of residences is being actively conducted. Factors that determine residence quality include stability and comfort, and many studies have used ‘residential satisfaction’ as an important indicator of residence quality [1]. ‘Residential satisfaction’ has been the subject of long-term research in the field of architecture, and a number of studies have focused on the correlation between individual housing satisfaction and conditions [1]. Mohit, M.A. et al. integrated various conceptual frameworks, such as sociodemographic factors, physical features of the house, housing support, and six facilities, as a multifaceted framework for residential situations [2]. Mohamed I. A. et al. suggested the four components of sociodemographic, housing, neighborhood, and behavioral characteristics in their residential satisfaction framework [3].

Even before COVID-19, the HUD Strategic Plan FY 2014–2018 published by the US Department of Housing and Urban Development suggested that a housing policy should aim to use residence as a platform to improve people’s quality of life [4], confirming that residence is an important element for quality of life. South Korea (Korea hereafter) recorded a low score on life satisfaction, which is one of the topics in the Better Life Index published annually by the Organization for Economic Cooperation and Development (OECD, 2018), on which Korea ranked 33rd among 40 countries [5]. Efforts to improve quality of life are therefore urgently needed to address the issue of Koreans’ subjective quality of life, which is much lower than that of residents in other OECD countries.

Quality of life is a multifaceted concept utilized in various fields. From a theoretical perspective, it comprises the happiness, life satisfaction, well-being, and needs satisfaction approaches [3,6]. Borthwick-Duffy presented three perspectives on quality of life—life condition, satisfaction with life, and a combination of life conditions and satisfaction—and David, F. et al. presented a combination of life conditions and satisfaction weighted by scale of importance [7,8].

In the field of architecture, Campbell et al. attempted to quantify the correlation between residential satisfaction and residents as a useful predictor for their quality of life [9], while Rollero et al. demonstrated that residents’ satisfaction with their residential environment affected their social well-being and quality of life [10]. In particular, as one’s housing condition and interaction with neighbors considerably influence one’s residential satisfaction, empirical studies have confirmed that the more active people are in daily activities with neighbors or the better their housing condition is, the higher their residential satisfaction is. Mousazadeh, H. et al. reported empirical research showing that the sense of place of urban immigrants affects their quality of life [11]. Prior studies have examined the relationship between residential satisfaction and residents’ quality of life and between residential satisfaction and quality of residence. However, prior studies have rarely investigated people’s psychological need for a dwelling in detail by linking factors that influence residential satisfaction directly and specifically. Moreover, to the best of our knowledge, no comprehensive framework has been proposed that integrates the conceptualization of residential satisfaction with subjective quality of life.

To bridge this gap between the psychological field and the architectural field in relation to residential satisfaction, which is essential for improving the quality of life of residents, the aim of this research is to conduct an experimental study linking Abraham H. Maslow’s desire satisfaction approach based on the theory of human motivation [12] in the field of psychology with residential satisfaction factors in the field of architecture. Furthermore, the focus of this study is to provide the theoretical basis of a modified Maslow’s hierarchy of needs to address contemporary situations. Because Maslow’s hierarchy of human needs begins with very basic needs and moves upward from physiological needs, safety needs, love and belonging needs, and esteem needs to self-realization needs, the framework is subject to limitations in demonstrating that needs unconditionally move from lower to higher levels, ignoring that needs can vary depending on individual situations. Recent studies have solved this problem by transforming Maslow’s pyramid into a circle [13].

Therefore, using a qualitative and quantitative integrated analysis in this study, we empirically present a comprehensive framework that connects the influencing factors of residential satisfaction with a modified version of Maslow’s human needs framework, aiming to fill the gap between them.

To this end, we performed the following three tasks (in order): (1) Residential satisfaction variables were identified and organized based on prior research on residential satisfaction factors. (2) To link the identified residential satisfaction factors and each of Maslow’s five human needs, a preliminary analytic hierarchy process (AHP) was carried out with 13 researchers in the housing sector. Based on the outcomes of the preliminary AHP, a focus group interview (FGI) was conducted with six experts from the construction and housing sectors. During the interview, participants had an in-depth discussion on why those residential satisfaction factors were linked to each of Maslow’s human needs. Then, through a second AHP with these experts, the residential satisfaction factors were eventually linked with each of Maslow’s five human needs. (3) A decision tree analysis of the data obtained from the Korea Housing Survey conducted by the Ministry of Land, Infrastructure and Transport (MOLIT) was applied to our conceptual framework in the context of Korea to investigate specific residential needs [14].

In this way, a comprehensive framework of residential satisfaction factors based on Maslow’s human needs can be utilized to concretize the psychological basis of residents’ desires in relation to their residence. As a result, the framework proposed in this study can be applied to regional housing policies and be customized according to unique needs and circumstances of life by region.

2. Related Works

2.1. Maslow’s Hierarchy of Human Needs

According to Maslow (1943), human needs can be arranged in a hierarchy of five levels (or stages), starting with very basic needs, based on relative priority. The first stage of physiological needs represents the most fundamental needs and human instincts, such as food, drink, shelter, clothing, and sleep. Once an individual’s basic physiological needs are satisfied to a certain degree, the needs for physical and psychological safety become salient in the second stage. The third level of human needs involves the pursuit of social belonging and love. Then, one enters the fourth stage of needs, i.e., esteem, which is the desire to be independent and respected by others for status. Finally, the needs for self-actualization are at the top of the hierarchy, representing the desire to realize one’s potential. Maslow proposed that in general, people tend to pursue higher levels of needs when their lower-level needs are fulfilled.

More recently, several researchers approached Maslow’s original hierarchy of needs from a new perspective and either modified it or presented the alternative idea that such needs coexist with one another. Through a cognitive-systemic reconstruction of Maslow’s theory, Heylighen investigated and explained why frustration with lower-level needs does not interfere with self-actualization needs [15]. Kiel examined lifelong learning and self-actualization among students of unconventional ages and older people [16]. The author proposed an ‘open triangle’ model to better reflect the fact that self-actualization, which is at the top of Maslow’s hierarchy of needs, does not end but continues to grow. Kenrick focused on the continuous dynamic interaction between the internal motives of human beings and their functional links with ongoing environmental threats and opportunities [17]. The author proposed a renovated hierarchy of fundamental motives to combine modern characteristics with the classical elements of Maslow’s needs, suggesting that human beings possess higher-level needs even if their lower-level needs are not fulfilled. Andrew, J.H. et al. adapted Maslow’s hierarchy of human needs as a comprehensive framework to conceptualize medical residents’ wellness, with the aim of addressing the issue of medical residents’ burnout negatively affecting both their career satisfaction and patient outcomes [13]. The authors proposed a modified version of Maslow’s hierarchy of needs with the hierarchical characteristics of different needs removed from Maslow’s original work in response to the modern environment (Figure 1). Based on the framework of Andrew et al., in the current study, we investigated how residential satisfaction has changed over time in Korea.

Adopting the framework presented by Andrew et al., we anticipate that the current study will contribute to research and development related to quality of life in the field of architecture, as we investigated residential satisfaction factors identified in prior research in connection with the Modified Maslow’s Hierarchy of Needs (MMHN).

2.2. Residential Satisfaction Factors

2.2.1. Household Characteristics

In prior research on residential satisfaction factors, some studies discovered that individuals’ demographic characteristics influence their residential satisfaction. Contradictory outcomes exist regarding residents’ age, which has been reported as both directly proportional [18,19] and inversely proportional to residential satisfaction [20]. Another study found that household, tenure type, and sociodemographic characteristics were positively linked to overall residential satisfaction [21]. In the current study, we analyzed the effects of factors related to the individual characteristics of Koreans, as reported in previous studies, on housing satisfaction. The first factor of household characteristics that we considered encompassed household demographic characteristics, including age, gender, and marital status.

Income and house ownership also have a directly proportional relationship with residential satisfaction [18]; moreover, the level of education and household size affect the level of residential satisfaction [22,23]. One study examined whether income, residence life, house ownership, and type of household affect residents’ satisfaction positively or negatively [2]. Additionally, some studies included analyses based on tenure type [24,25,26], while others concluded that residential satisfaction is affected by living expenses, including house management expenses [27] and duration of residence [28]. Among the factors considered in prior studies, in the current study, we analyzed secondary factors related to household characteristics, i.e., socioeconomic characteristics of the household. Furthermore, we analyzed detailed factors, such as children, family size, national basic living security benefits, educational background, monthly income, and satisfaction.

2.2.2. House Characteristics

Lane and Kinsey argued that house characteristics are more important than demographic characteristics [29]. This indicates that building features, such as the number of bedrooms, kitchen size, location, and housing quality, are closely related to residential satisfaction [30]. Morris et al. demonstrated that the level of residential satisfaction varies depending on the house type [31], whereas Noriza et al. found that the condition and size of a house have a positive influence on residential satisfaction [32]. Parkes et al. suggested that the structural housing indicator is an important element affecting residential satisfaction [33]. Another study found that residential satisfaction varies depending on residential features, such as the number of bedrooms and bathrooms, kitchen size and location, dining area, and living room, as well as depending on the cross-cultural level [2]. With reference to the abovementioned studies, in the current study, we used dwelling area as the primary factor of house characteristics.

In this study, we included in house characteristics not only the physical characteristics of a house but also the quality of residence provided by the developer, as well as other house-related elements, such as privacy, security, and ventilation. Ibem and Aduwo reported that the most important prediction variables for residential satisfaction are related not only to the security of one’s residence but also to the adequacy of thermal and visual comfort [23]. Based on the studies mentioned above, in the current study, we classified a house’s indoor environment as a secondary factor and considered interior quality, indoor environment level, and interior safety and cleanliness as factors reflecting the interior conditions. Furthermore, reflecting the reality of Korea, the economic value of housing was considered as a third factor. The following detailed factors were included: construction year, house management expenses, housing prices, and tenure type.

2.2.3. Residential Environment Characteristics

Research on how the residential neighborhood environment influences residential satisfaction found that it is closely related to the homogeneity of social class, ethnicity, and race in one’s neighborhood [31], as well as a good school district [34]. In the current study, we focused on the importance of the abovementioned factors and used them as indicators of the social characteristics of the environment, including the availability of neighborhood relationships and school districts.

The level of crime and disorder in the neighborhood [35], safety levels in terms of fire and traffic accidents [36], and crime prevention and accident protection have been found to be positively related to residential satisfaction. Therefore, environmental safety was set as a second factor of residential environment characteristics, consisting of safety from accidents and pedestrian safety.

Mohit and Al-Khanbashi Raja confirmed that people’s satisfaction with their neighborhood is an important predictor variable of residential satisfaction [2], that is, residents living farther away from schools, other houses, shopping areas, and medical services are generally more dissatisfied, indicating that residential satisfaction depends not only on housing itself but also on the neighborhood characteristics [33] and ease of access to social infrastructure, including public transportation [37]. Thus, we regard environmental convenience as the third factor of the residential environment, consisting of accessibility of commercial facilities, medical facilities, public facilities, cultural facilities, public transportation, and parking areas.

Amenities in the environment were used as the fourth factor. Detailed items included in this factor were noise, cleanliness, air pollution around apartment complexes, and the natural environment, which were used to analyze the relationship with residential environmental satisfaction.

3. Proposed Methodology

3.1. Problem Definition

Previous studies on residential satisfaction factors have not considered the satisfaction of residents’ psychological needs. Thus, existing scales need to be reconstructed from a psychological perspective. Therefore, the aim of this study is to derive significant residential satisfaction factors from the conventional model and the MMHN-based model using CART regression and to analyze the distinctive characteristics of the results obtained based on the MMHN-based model from the perspective of human psychology.

In this study, the following three-stage qualitative and quantitative integrated experiment was designed to link factors affecting residential satisfaction and the MMHN. First, residential satisfaction factors were identified based on prior research (see the residential satisfaction factors listed in

Table 1. Residential satisfaction factors.

Classification	Residential Satisfaction Factors		Related Work
Household Characteristics	Household Demographic Characteristics	Age	Lu (1999) [18], Pinquart and Burmedi (2004) [19], Mohit et al. (2010) [20], Mohit and Al-Khanbashi Raja (2014) [21], and CRN (2015) [38]
		Gender	Lee and Kim (2013) [39], Ibem and Amole (2013) [40], and Lu (1999) [18]
		Marital status	Mohit and Al-Khanbashi Raja (2014) [21]
	Household Socioeconomic Characteristics	Children	Lu (1999) [18]
		Family size	Vera-Toscano and Ateca-Amestoy (2008) [22] and Ibem and Aduwo (2013) [23]
		Beneficiaries of the national basic livelihood security	Oh (2010) [41] and Chun et al. (2013) [42]
		Educational background	Vera-Toscano and Ateca-Amestoy (2008) [22], Ibem and Aduwo (2013) [23], and CRN (2015) [38]
		Monthly household income	Lu (1999) [18], Mohit and Al-Khanbashi Raja (2014) [21], Chun et al. (2013) [42], and CRN (2015) [38]
		Debt	Chun et al. (2013) [42]
House Characteristics	Economic Value	House management expenses	Jang and Yoo (2017) [43] and CRN (2015) [38]
		Housing prices	Thomas Davidoff (2006) [44] and Ibem and Aduwo (2013) [23]
		Construction year	Park (2012) [45] and Chun et al. (2013) [42]
		Tenure type (own/rent)	Mohit and Al-Khanbashi Raja (2014) [21], Kaitilla (1993) [24], Kim (2016) [25], Kang et al. (2015) [26], and Thomas Davidoff (2006) [44]
	House Size	Dwelling area	Noriza et al. (2010) [32] and Mohit and Al-Khanbashi Raja (2014) [21]
	Indoor House Environment	Interior quality: flooring, wall finishing material, ceiling material, and lighting	Noriza et al. (2010) [32] and Bae et al. (2014) [46]
		Indoor environment level: noise, ventilation, natural light, heating and insulation, and waterproofing	Mohit and Al-Khanbashi Raja (2014) [21], Ibem and Aduwo (2013) [23], Bae et al. (2014) [46], and NSW (2014) [47]
		Indoor safety and cleanliness: structure, disaster safety, fire safety, security, and sanitation	Parkes et al. (2002) [33], Ibem and Aduwo (2013) [23], and Bae et al. (2014) [46]
Residential Environment Characteristics	Social Characteristics of the Environment	Availability of neighborhood relationships (community relations)	Morris et al. (1976) [31], Mohit and Al-Khanbashi Raja (2014) [21], Jang et al. (2017) [43], NSW (2014) [47], and CRN (2015) [38]
		School district	Jeong et al. (2015) [34] and Mohit and Al-Khanbashi Raja (2014) [21]
	Safety of the Environment	Safety from accidents and incidents and security control	Robinson et al. (2003) [35], Kim et al. (2016) [36], Mohit and Al-Khanbashi Raja (2014) [21], and NSW (2014) [47]
		Pedestrian safety	Kim et al. (2016) [36]
	Amenities in the Environment	Noise around apartment complexes	Jang et al. (2017) [43], NSW (2014) [47], and Ibem and Aduwo (2013) [23]
		Cleanliness around apartment complexes	Jang et al. (2017) [43] and NSW (2014) [47]
		Air pollution around apartment complexes	NSW (2014) [47]
		Natural environment (accessibility of parks and green areas)	Kim et al. (2013) [48]
	Convenience of the Environment	Accessibility of commercial facilities	Mohit and Al-Khanbashi Raja (2014) [21], Ibem and Aduwo (2013) [23], and Kim et al. (2013) [48]
		Accessibility of medical facilities	Mohit and Al-Khanbashi Raja (2014) [21], Clement and Kayode (2012) [49], and Ibem and Aduwo (2013) [23]
		Accessibility of public facilities	Clement and Kayode (2012) [49] and Kim et al. (2013) [48]
		Accessibility of cultural facilities	Ibem and Aduwo (2013) [23] and Kim et al. (2013) [48]
		Accessibility of public transportation	Yoon (2010) [37] and Ibem and Aduwo (2013) [23]
		Accessibility of parking areas	Bae et al. (2014) [46]

). Second, an AHP survey was designed to link the identified residential satisfaction factors with the MMHN. The preliminary AHP was conducted with a group of researchers in the housing sector to identify the connection between residential satisfaction factors and the MMHN. Then, based on the results of the survey, an FGI was conducted with various experts in the housing sector. At this stage, the outcomes of the preliminary AHP were further concretized through an in-depth and realistic discussion of each item to be linked between residential satisfaction factors and the MMHN. Following the FGI with the expert group, a second AHP survey was carried out with the same experts to systematize the final linkage between the MMHN and residential satisfaction factors. The AHP surveys and FGI were conducted with approval from the Korea Advanced Institute of Science and Technology’s institutional review board. Finally, the residential satisfaction factors systematized based on the MMHN, formulated as described above, were applied to the MOLIT’s Korea Housing Survey data for 2020 [14]. The overall process of this study is summarized in Figure 2.

3.2. Identification of Residents’ Satisfaction Factors for Housing

By reviewing prior research related to residential satisfaction, factors that affect residents’ housing satisfaction were reclassified and organized according to type, as shown in Table 1. These factors were categorized into household characteristics, house characteristics, and residential environment characteristics. Household characteristics were then divided into household demographic characteristics and household socioeconomic characteristics; the former included age, gender, and marital status, and the latter included children, family size, beneficiaries of national basic livelihood security, educational background, monthly income, and debt. House characteristics were divided into the following subcategories: economic value, house size, and indoor house environment. Economic value included house management expenses, housing prices, and construction year. House size included dwelling area. Indoor house environment included interior quality, indoor environment level, and indoor safety, cleanliness, and sanitation. Residential environment characteristics were divided into the following subcategories: social characteristics of the environment, safety of the environment, amenities in the environment, and convenience of the environment. These characteristics comprised community relations and school district; safety from accidents and incidents, security control, and pedestrian safety; noise, cleanliness, air pollution, and natural environment; and accessibility of commercial facilities, medical facilities, public facilities, cultural facilities, public transportation, and parking areas, respectively.

3.3. Qualitative and Quantitative Integrated Analysis to Link Residential Satisfaction Factors and Maslow’s Hierarchy of Human Needs

3.3.1. Analytic Hierarchy Process

An objective determination of which stages of Maslow’s hierarchy of human needs are closely linked to which residential satisfaction factors requires a quantitative evaluation of the qualitative elements. To address this matter, in this study, we used decision analysis to make reasonable and logical decisions in uncertain circumstances. The analytic hierarchy process (AHP) was utilized as the decision analysis method. Developed by Thomas L. Saaty, AHP is a decision-making method used to capture an evaluator’s knowledge, experience, and intuition through judgment made via pairwise comparison of different elements [50]. Canco confirmed that AHP is an efficient method for decision-making that involves numerous different criteria [51].

Prior research in the housing, urban development, and other sectors utilized AHP as a multi-criteria decision-making method. In the housing sector, Jung et al. utilized AHP to analyze important determinants of consumer preference regarding housing in Dubai [52], and Özsubaşı et al. determined the order of importance of eight real estate evaluation criteria, regarding the location of housing, using AHP [53]. In the urban development sector, Aldossary et al. used AHP to identify the best future urban development project for the Al-Baha region [54], while Canesi defined the degrees of sustainability and efficiency of different urban projects in Italian marginal areas using AHP [55]. Additionally, AHP was utilized to develop a methodology for real estate risk assessment regarding public–private partnership projects [56].

The reason this method is appropriate for this study is that it can be used to reflect experts’ perceptions of a phenomenon during evaluation. AHP is also suitable because it asks experts to rank the relative importance of relevant elements and utilizes surveys with a ratio scale for decision-making; thus, it can convert qualitative opinions into quantitative values for analysis [57] through the ratio scale [58].

The theoretical logic is described in this section. Pairwise comparison matrix A may contain inverse numbers centering on the diagonal matrix, as shown below (1). When the relative intensity of the importance of n elements for comparison within one level is W_i (i = 1, …, n), the a_ij of the pairwise comparison matrix can be estimated as W_i/W_j. (i,j = 1, …, n):

$$A=\left[\begin{array}{ccc}{a}_{11}& \cdots & a_{1n}\\ ⋮& \ddots & ⋮\\ a_{n1}& \cdots & a_{nn}\end{array}\right]\cong \left[\begin{array}{ccc}\raisebox{1ex}{$W_1$}\!\left/ \!\raisebox{-1ex}{$W_1$}\right.& \cdots & \raisebox{1ex}{$W_1$}\!\left/ \!\raisebox{-1ex}{$W_n$}\right.\\ ⋮& \ddots & ⋮\\ \raisebox{1ex}{$W_n$}\!\left/ \!\raisebox{-1ex}{$W_1$}\right.& \cdots & \raisebox{1ex}{$W_n$}\!\left/ \!\raisebox{-1ex}{$W_n$}\right.\end{array}\right]$$

(1)

Here, ∀ i, a_ij = 1/a_ji and a_ii = 1.

W can be obtained using the eigenvalue method according to Equation (2):

$$A·W=n·W$$

(2)

where W = (W₁, W₂, W₃, …, W_n) is an eigenvector, and n is an eigenvalue of matrix A.

When the evaluator does not know W or the weighted intensity of importance for each element of the pairwise comparison matrix A, this matrix is called A′, the estimated intensity of importance of which (W′) is obtained using Equation (3), where λ_max is the greatest intensity of importance of matrix A′:

$$A^{\prime }·W^{\prime }={\lambda }_{max}·W$$

(3)

Regarding the validation of the consistency ratio of decision outcomes, if λ_max is a value closely approaching n, the numbers in the pairwise comparison matrix A can be said to be consistent. Such a level of consistency is utilized to determine whether judgments made by decision-makers maintain logical consistency. When the consistency ratio is below 0.1, the pairwise comparison is determined to be consistent. The level of consistency can be measured using the consistency index, the random index, and the consistency ratio as follows:

$$\begin{gathered} CI=\frac{{\lambda }_{max}-n}{n-1} \\ CR=\frac{CI}{RI}\times 100\% \end{gathered}$$

(4)

The resulting overall process of the AHP method is summarized in Figure 3.

Regarding the structural design of the AHP (Figure 4), a decision-making system of the experts’ judgments was organized for the AHP analysis according to the hierarchy based on the residential satisfaction factors (Table 1). The purpose of decision-making is to ‘determine residential satisfaction factors of their housing that can match each of the five stages of the MMHN for residents’. With this goal, household characteristics, house characteristics, and residential environment characteristics were considered as the first evaluation criteria, under which the second and third sub-criteria for evaluation were specified. At the bottom of the hierarchy, 30 specific items of residential satisfaction factors were included to be linked to the five levels of the MMHN. As such, the hierarchy shown in Figure 4 was established, and a 1:1 pairwise comparison was performed on factors at all levels of the hierarchy from criteria 1 to criteria 3 (Appendix A), to which a nine-point scale was applied for numerical evaluation to assign the degree of suitability (

Table 2. Pairwise comparison scale for the analytic hierarchy process. Source: Saaty [50].

Intensity of Importance	Definition
1	Equal importance of both elements
3	Weak importance of one element over another
5	Essential or strong importance of one element over another
7	Demonstrated importance of one element over another
9	Absolute importance of one element over another
2, 4, 6, 8	Intermediate values between two adjacent judgments

).

3.3.2. Focus Group Interview

A focus group interview (FGI) is a qualitative research method that supports an in-depth approach to a research topic through the process of discussion. An FGI offers the advantage of collective intelligence by allowing interview participants to converse directly on the research topic, thereby providing specific information [59]. Such a method can also be used to obtain multidimensional outcomes by reflecting the opinions of research participants that cannot be easily captured in quantitative research using survey sheets. Another advantage is the practical approach to research problems, which can be used to recognize the in-depth perceptions of participants regarding problems through structured and unstructured questions. The FGI method offers high utility as it can overcome the limitations of analytical results obtained by deducing priorities from the intensity of importance, as in the case of AHP. If the resulting values are determined on the basis of a very small difference, the given problem is reviewed again through another in-depth discussion before reaching a conclusion. Therefore, a discussion group should consist of at least five participants who are experts from various fields. However, effectively bringing together experts for problem-solving may prove difficult. Moreover, participation in such research requires considerable time commitment, which can be burdensome for participants.

3.3.3. Classification and Regression Tree (CART)

A decision tree is a machine-learning algorithm used for classification, regression, or prediction [60]. The main attribute of decision trees is that they recursively split sets of data or nodes based on questions regarding the value of independent variables. The recursive process of splitting nodes forms a tree-shaped structure (Figure 5).

A node is a branching point in a decision tree that corresponds to a set of data. The root node is the uppermost node in a tree that contains every observation in the dataset. The parent node is the node being divided, whereas the nodes derived from the division are child nodes. The terminal node is a node that is no longer divided. The advantages of decision trees include easy interpretation and explanation, easy handling of qualitative predictors, and broad applicability to both classification and regression problems [61]. Various algorithms can be used to construct decision trees, including CART, ID3, C4.5, C5.0, CHAID, and MARS [62].

The CART algorithm is an algorithm that constructs a tree-structured classifier through binary recursive partitioning [63]. The CART algorithm repeatedly splits the parent node into two child nodes using a set of yes/no questions regarding the independent variables [64]. The decision of optimal questions to split parent nodes, the decision of whether to continue splitting a node or to declare it a terminal node, and the optimization of the tree must be considered during the development of the tree structure using the CART algorithm.

The process of constructing a tree using the CART algorithm consists of three steps: splitting the nodes to construct a tree, choosing the appropriate tree size, and classifying new data using the tree. The first step is to construct a tree by splitting nodes. This step involves the ‘splitting rule’, which decides the optimal question to split a node in two. In a classification problem, the Gini index serves as a parameter for the optimal split. The Gini index measures node impurity, and its formula is as follows:

$$\mathrm{Gini}=1-{\sum }_{i=1}^j{P\left(i\right)}^2$$

(5)

where j represents the number of different values of the dependent variable, and P(i) represents the ratio of the number of observations with the ith value of the dependent variable relative to the total number of observations in the node. The lower the Gini index, the purer the node; thus, the optimal split occurs when the sum of the Gini index in the child nodes is the smallest. In a regression problem, reduction in variance serves as a split criterion. The formula for variation reduction is as follows:

$$\mathrm{Reduction}=\mathsf{\phi }\left(\mathrm{X}\right)-\{\phi \left(X_1\right)+\phi \left(X_2\right)\}$$

(6)

where X, X₁, and X₂ represent the set of data in the parent node and the first and second child nodes, respectively, and φ(X) represents variance. The best split is the split with the maximum value of reduction. Next, the problem of whether to declare a node terminal must be addressed. In most cases, a node is declared terminal when no further split of the node can induce a decrease in the Gini index or variance. When all child nodes are declared terminal, the tree stops growing, and the grown tree is denoted as T₀. However, the constructed tree may overfit the data, which means that the maximal tree attempts to follow every peculiarity of the data and loses the ability to predict general future datasets. Therefore, a second step is required.

The second step is to prune the tree and determine its optimal size. Pruning can be conducted in two ways: pre-pruning and post-pruning. Pre-pruning is conducted through early cessation of decision tree growth using preconditions. For example, a node can be declared terminal when the number of observations in the node falls below a certain number, for example, 10, or the depth of the tree exceeds a certain extent. Post-pruning is conducted through the assessment of cost complexity in the case of regression. Cost complexity assessment uses the following function:

$$f\left(\alpha ,T\right)=\sum _{m=1}^{\left|T\right|}\sum _{x_i\in R_m}{\left(y_i-\widehat{y_{R_m}}\right)}^2+\alpha |T|$$

(7)

For each value of α, there exists a subtree ($T\subset T_0$) in which f(α, T) is minimized, where |T| is the number of terminal nodes in T, and $\widehat{y_{R_m}}$ is the mean value of the dependent variable in the mth terminal node. The left part of the equation represents variances or errors in the regression tree, whereas the right part represents the size of the tree, which means that the value of α regulates the balance between compactness and accuracy [61]. Determining the optimal value of α is a problem to be solved in the third step.

The third step is to test the tree with new datasets. Two major methods are used in this step: K-fold cross-validation and validation with a separate test dataset. K-fold cross-validation divides observations into K subsets, repeats tree formation for all but the kth subset for each k = 1, …, K, measures the mean squared error of prediction for the kth subset data, and selects a value of α that minimizes the error [61]. Validation with a separate test dataset divides the data into a training dataset and a testing dataset, constructs a decision tree based on the training dataset, and sets the value of α to minimize the mean squared prediction error on the testing dataset. The former method is used on datasets with 5000 or fewer observations, whereas the latter is used on datasets with more than 5000 observations.

The CART algorithm also has the feature of measuring the relative importance of each independent variable. The importance of each independent variable is measured based on the reduction in model accuracy when the variable is removed from the model. Relative variable importance is the ratio of the importance of a variable to the importance of the most important variable in the model, which is often measured in percentage.

The CART algorithm can be broadly utilized as independent variables of all formats—categorical, ordinal, or continuous—can be analyzed together. Examples of decision tree use include the usage of AHP and a decision tree for the model decision-making process that occurs during the implementation of construction projects [65], and the utilization of the CART algorithm to develop accident classification systems [66].

Several data mining packages that automate decision tree formation have been commercialized to enable data analysis using decision trees. In this study, we utilized the CART program embedded in Minitab Statistical Software version 27, developed by Minitab LLC. in Pennsylvania, U.S. Minitab Statistical Software automates the entire process of decision tree construction, allowing for adjustments such as the selection of variables and pre-pruning. Pre-pruning was conducted by setting the minimum number of observations in the terminal node to 10 and the maximum depth of the tree to 5.

4. Implementation and Results

4.1. Result of Analytic Hierarchy Process

The preliminary AHP survey was conducted with researchers at the Korea Research Institute for Human Settlements (KRIHS) and the Korea Land & Housing Institute (see

Table 3. Information about the experts involved in the preliminary analytic hierarchy process.

Classification	Number	Gender	Years of Work Experience	Number	Percent
Korea Research Institute for Human Settlements	7	Male: 3	More than 3 years and less than 5 years	2	53.8%
		Female: 4	More than 5 years	5
Korea Land & Housing Institute	6	Male: 2	More than 3 years and less than 5 years	1	46.2%
		Female: 4	More than 5 years	5

). These researchers had experience with either research or tasks involving residential satisfaction factors. The survey was prepared to allow the collection of careful and in-depth opinions on its contents and results. The survey participants included seven researchers from KRIHS who had been members of departments that are either responsible for or related to the Korea Housing Survey for three years or longer, as well as six researchers from the Korea Land & Housing Institute who had been working at the institute for at least three years. All the participants had either a master’s or doctoral degree; five were men, and eight were women. The survey was conducted online from June 28 to July 16, 2021. ‘I Make It’ software was used for the AHP questionnaire, which was emailed to the participants. The responses were also received via email. The survey, which consisted of a total of 46 questions, was conducted in the following order of pairwise comparison (see Appendix A). For each item corresponding to one factor influencing residential satisfaction and for two of Maslow’s five levels of human needs, the residential satisfaction factor was considered more strongly related to the level of needs in the form of pairwise comparison. Each problem used a two-sided, nine-point scale. When the respondents finished responding to each question, the level of needs most strongly connected to the factors influencing residential satisfaction was calculated based on the response data.

The survey designed as described above yielded the following results based on the highest score: The influencing factors of physiological needs were age and beneficiaries of the national basic livelihood security. The factors influencing safety needs were gender; children; family size; construction year; indoor environment level; safety from accidents; pedestrian safety; noise; cleanliness; air pollution around the house; and accessibility of medical facilities, parking areas, and public facilities. The factors influencing love and belonging needs were marital status, house management expenses, availability of neighborhood relations, and accessibility of public transportation. The factors affecting esteem needs were interior quality and natural environment. The factors affecting self-actualization needs were educational background, monthly income, debt, housing prices, tenure type, dwelling area, indoor safety and cleanliness, school district, accessibility of commercial facilities, and accessibility of cultural facilities (see Appendix B).

The results of the preliminary AHP survey, which matched the residential satisfaction factors shown here with the MMHN, were limited because, in each qualitative assessment, several needs scored 20% and 40% lower than the highest scores, making it difficult to make a definitive selection using the results with the highest scores as the basis. To address this limitation, we decided that more objective results needed to be obtained by conducting a new focus group interview with another expert group, as well as a second AHP survey.

4.2. Result of Focus Group Interview

The FGI was designed to bring together experts from various fields related to housing to have an in-depth discussion regarding the system of RSFs based on the MMHN from various viewpoints. Our aim was to allow for a convergence of opinions through a careful and in-depth review of the preliminary AHP survey results. The focus group consisted of a total of six experts: two architects with more than 10 years of experience as registered architects, an interior architect with more than 10 years of experience, a research professor of housing studies, a residential researcher from the Korea Institute of Regional Development, and a public official in charge of architecture (see

Table 4. Information about the experts who participated in the focus group interview.

Classification	Number	Gender	Years of Work Experience	Education Degree
Architect	2	Male: 1	More than 10 years	ABD
		Female: 1		MD
Interior Architect	1	Female: 4	More than 10 years	MD
Research Professor, Department of Interior Architecture & Built Environment	1	Male: 1	More than 10 years	Ph.D.
Research Fellow, Ulsan Research Institute	1	Female: 1	More than 10 years	Ph.D.
Public Officer of Architecture, City Hall	1	Male: 1	More than 10 years	MD

). These experts had either a master’s or doctoral degree; three were men, and three were women. The in-depth interview took place on 28 July 2021, from 5:30 p.m. to 9 p.m. Due to the COVID-19 pandemic, the discussion proceeded online using Zoom and was recorded with the consent of the attendees.

In the FGI, a total of 36 questions were posed to the respondents. Each question was about which stage of the five stages of needs in Maslow’s hierarchy a given residential satisfaction factor would best correspond to. The responses of the respondents as to why each factor corresponded to a specific needs stage were collected and summarized (see Appendix C).

4.3. Result of the Second Analytic Hierarchy Process

Following the FGI, a second AHP survey was requested of the six participating experts to obtain the final results. The second AHP survey was conducted from 28 July to 4 August, and as with the preliminary survey, the ‘I Make It’ online survey software was used. The final results obtained from the second survey are discussed below. The inconsistency ratio was below 0.1, indicating that the survey results are significant. Moreover, for each item, the stage of the MMHN that had the highest value in terms of intensity of importance was determined.

The survey led to the results reported in Appendix D. The results based on the highest score were as follows: The factors influencing physiological needs were age and beneficiaries of the national basic livelihood security. The factors influencing safety needs were gender; children; family size; construction year; indoor environment level; safety from accidents; pedestrian safety; noise; cleanness; air pollution around the house; and accessibility of medical facilities, parking areas, and public facilities. The factors influencing love and belonging needs were marital status, house management expenses, availability of neighborhood relations, and accessibility of public transportation. The factors affecting esteem needs were interior quality and natural environment. The factors affecting self-actualization needs were educational background, monthly income, debt, housing prices, tenure type, dwelling area, indoor safety and cleanliness, school district, accessibility of commercial facilities, and accessibility of cultural facilities. The results were summarized in a diagram shown in Figure 6. These results were clearer and more reliable than those from the preliminary AHP survey, presumably because the FGI allowed the survey respondents to analyze the study’s topic and purpose from various perspectives to better understand its subject and topic, resulting in clearer final AHP results.

4.4. CART Analysis Results of the Nationwide Data

The data for 48,908 households from the 2020 Korea Housing Survey were used for this study. All observed data were utilized, as CART analysis was capable of handling missing values. A separate dataset was used to validate the model. To this end, 70% of the data were randomly selected as the training dataset, and the remainder became the testing dataset. The depth limit of the trees was set to five as a pre-pruning measure. The units of all variables used in the CART analysis are summarized in a table (see Appendix E).

In the conventional model, CART analysis was conducted using residential satisfaction and residential environment satisfaction as dependent variables. The resulting CART trees are presented in Appendix F.

In the MMHN-based model, residential satisfaction factors were classified into five groups based on the MMHN, and CART analysis was conducted using residential satisfaction and residential environment satisfaction as dependent variables for each of the five independent variable sets. The CART analysis results on the effect of factors influencing self-actualization needs on residential satisfaction and residential environment satisfaction are presented in Appendix G.

Variables that appeared in the CART structure were deemed important. These variables and their relative importance were analyzed (see Appendix H).

4.4.1. CART Analysis Results on Residential Satisfaction in the Conventional Model

The mean residential satisfaction score of the total dataset was 2.98573, and the standard deviation was 0.537402.

The most important variable was indoor safety and cleanness. Other significant variables included (in decreasing order of importance) indoor environment level, cleanness around the house, safety around the house, air pollution around the house, accessibility to parking areas, school district, and construction year.

4.4.2. CART Analysis Results on Residential Environment Satisfaction in the Conventional Model

The mean residential environment satisfaction score was 2.96042, and the standard deviation was 0.513808.

The most important variable was safety around the house. Other significant variables included (in decreasing order of importance) accessibility of commercial facilities, accessibility of medical facilities, accessibility of public facilities, pedestrian safety, school district, accessibility of cultural facilities, neighborhood relationships, indoor safety and cleanness, noise around the house, and accessibility of parking areas.

4.4.3. CART Analysis Results on Residential Satisfaction in the MMHN-Based Model

CART analysis result on the effect of factors in the safety needs stage on residential satisfaction is shown in Figure 7.

In the stage of physiological needs, beneficiaries of the national basic livelihood security was found to be an important variable.

In the stage of safety needs, indoor safety and cleanness was found to be the most important variable, followed by the indoor environment level, cleanness around the house, air pollution around the house, accessibility to parking areas, and construction year.

In the stage of love and belonging needs, availability of neighborhood relationships was the most important variable. Accessibility to public transportation, marital status, children, and house management expenses were the next important factors in order.

In the stage of esteem needs, natural environment was found to be the most important variable. Housing prices, interior quality, and age were the next important variables in descending order.

In the stage of self-actualization needs, school district was the most important variable. The next important variables in order of decreasing importance were accessibility to cultural facilities, accessibility to commercial facilities, dwelling area, monthly income, and educational background.

4.4.4. CART Analysis Results on Residential Environment Satisfaction in the MMHN-Based Model

CART analysis result on the effect of factors in the safety needs stage on residential environment satisfaction is shown in Figure 8.

In the stage of physiological needs, beneficiaries of the national basic livelihood security was found to be an important variable.

In the stage of safety needs, safety around the house was found to be the most important variable. Accessibility to public facilities, cleanness around the house, noise around the house, accessibility to parking areas, air pollution around the house, accessibility to medical facilities, indoor environment level, and indoor safety and cleanness were the next most important variables in descending order.

In the stage of love and belonging needs, accessibility to public transportation was the most important variable. Availability of neighborhood relationships was the next most important variable.

In the stage of esteem needs, natural environment was the most important variable. Housing prices and interior quality were the next most important in order of decreasing importance.

In the stage of self-actualization, accessibility to commercial facilities was found to be the most important variable. School district, accessibility to cultural facilities, and dwelling area were the next important variables in order.

4.5. Regional CART Analysis Results

Human needs in relation to housing are closely linked to the location of the house as location reflects the need of its residents to live in areas of certain conditions, such as metropolitan cities or rural areas. However, the analysis conducted using the nationwide data could not address the effect of regional characteristics on human needs and residential satisfaction. Thus, to properly analyze factors affecting residential satisfaction from the perspective of human needs, the data must be divided by region and reanalyzed.

To account for regional differences in residential satisfaction, the 2020 Korea Housing Survey data were classified according to the 17 administrative divisions of the Republic of Korea (

Table 5. Administrative divisions of the Republic of Korea and the number of households analyzed.

Region	Administrative Division	Households	Regional Category	Administrative Division	Households
Capitol Region (A)	Seoul Metropolitan City (A1)	2877	Gyeong-sang Region (C)	Daegu Metropolitan City (C1)	1800
	Incheon Metropolitan City (A2)	1505		Ulsan Metropolitan City (C2)	1148
	Gyeonggi Province (A3)	3669		Busan Metropolitan City (C3)	2056
Chungcheong Region (B)	Daejeon Metropolitan City (B1)	1165		North Gyeong-sang Province (C4)	1801
	Sejong Self-governing City (B2)	499		South Gyeong-sang Province (C5)	1877
	North Chungcheong Province (B3)	1360	Jeol-la Region (D)	Gwangju Metropolitan City (D1)	1396
	South Chungcheong Province (B4)	1549		North Jeol-la Province (D2)	1565
Gangwon Region (E)	Gangwon Province (E1)	1366		South Jeol-la Province (D3)	1698
			Jeju Region (F)	Jeju Island (F1)	666

). A total of 27,997 households from 17 divisions were analyzed, and observations with missing values were excluded.

In the analysis of capital and non-capital regions, the survey data were divided into two groups: one group consisting of data from the capital region, and the other group consisting of data from all other regions, representing non-capital regions.

In the analysis of metropolitan and non-metropolitan regions, the survey data were divided into two groups: one group representing metropolitan cities consisting of data from the seven metropolitan cities shown in Table 5, and the other group consisting of data from all other regions, representing non-metropolitan cities.

4.5.1. Residential Satisfaction Analysis Results of Capital and Non-Capital Regions in the Conventional Model

The results of the capital region were as follows: The most important variable was indoor safety and cleanness. Other significant variables in order of decreasing importance included indoor environment level, availability of neighborhood relationships, and accessibility of public facilities.

The results of the non-capital regions were as follows: The most important variable was indoor safety and cleanness. Other significant variables in order of decreasing importance included indoor environment level, availability of neighborhood relationships, and accessibility of public facilities.

4.5.2. Residential Environment Satisfaction Analysis Results of Capital and Non-Capital Regions in the Conventional Model

The results of the capital region were as follows: The most important variable was accessibility of parking area. Other significant variables in order of decreasing importance included school district, pedestrian safety, safety around the house, accessibility of cultural facilities, availability of neighborhood relationships, accessibility of public facilities, indoor environment level, indoor safety and cleanness, and accessibility of public transportation.

The results of the non-capital regions were as follows: The most important variable was school district. Other significant variables in order of decreasing importance included accessibility of commercial facilities, accessibility of public transportation, safety around the house, air pollution around the house, noise around the house, cleanness around the house, natural environment, availability of neighborhood relationships, accessibility of public facilities, and indoor safety and cleanness.

4.5.3. Residential Satisfaction Analysis Results of Capital and Non-Capital Regions in the MMHN-Based Model

The results of the capital region were as follows:

ⓐ

Regarding safety needs, indoor safety and cleanness, indoor environment level, cleanness around the house, pedestrian safety, accessibility of public facilities, and construction year were significant in order of decreasing importance.

ⓑ

Regarding love and belonging needs, availability of neighborhood relationships and accessibility of public transportation were significant in order of decreasing importance.

ⓒ

Regarding esteem needs, natural environment and housing prices were significant in order of decreasing importance.

ⓓ

Regarding self-actualization needs, accessibility of cultural facilities, accessibility of commercial facilities, school district, dwelling area, and monthly income were significant in order of decreasing importance.

The results of the non-capital regions were as follows:

ⓐ

Regarding physiological needs, beneficiaries of the national basic livelihood security was a significant factor.

ⓑ

Regarding safety needs, indoor environment level, indoor safety and cleanness, cleanness around the house, pedestrian safety, safety around the house, air pollution around the house, and construction year were significant in order of decreasing importance.

ⓒ

Regarding love and belonging needs, accessibility of public transportation, availability of neighborhood relationships, and marital status were significant in order of decreasing importance.

ⓓ

Regarding esteem needs, housing prices, natural environment, interior quality, and age were significant in order of decreasing importance.

ⓔ

Regarding self-actualization needs, school district, accessibility of commercial facilities, accessibility of cultural facilities, dwelling area, and monthly income were significant in order of decreasing importance.

4.5.4. Residential Environment Satisfaction Analysis Results of Capital and Non-Capital Regions in the MMHN-Based Model

The results of the capital region were as follows:

ⓐ

Regarding safety needs, accessibility of parking areas, safety around the house, accessibility of public facilities, pedestrian safety, air pollution around the house, noise around the house, indoor safety and cleanness, and indoor environment level were significant in order of decreasing importance.

ⓑ

Regarding love and belonging needs, availability of neighborhood relationships and accessibility of public transportation were significant in order of decreasing importance.

ⓒ

Regarding esteem needs, natural environment and housing prices were significant in order of decreasing importance.

ⓓ

Regarding self-actualization needs, accessibility of commercial facilities, accessibility of cultural facilities, and school district were significant in order of decreasing importance.

The results of the non-capital regions were as follows:

ⓐ

Regarding physiological needs, beneficiaries of the national basic livelihood security was significant.

ⓑ

Regarding safety needs, accessibility of parking areas, accessibility of medical facilities, safety around the house, cleanness around the house, noise around the house, accessibility of public facilities, air pollution around the house, indoor environment level, and indoor safety and cleanness were significant in order of decreasing importance.

ⓒ

Regarding love and belonging needs, accessibility of public transportation and availability of neighborhood relationships were significant in order of decreasing importance.

ⓓ

Regarding esteem needs, natural environment and housing prices were significant in order of decreasing importance.

ⓔ

Regarding self-actualization needs, school district, accessibility of commercial facilities, accessibility of cultural facilities, and dwelling area were significant in order of decreasing importance.

4.5.5. Residential Satisfaction Analysis Results of Metropolitan and Non-Metropolitan Regions in the Conventional Model

The results of the metropolitan cities were as follows: The most important variable was indoor environment level. Other significant variables in order of decreasing importance included indoor safety and cleanness, cleanness around the house, noise around the house, accessibility of parking area, natural environment, and construction year.

The results of the non-metropolitan cities were as follows: The most important variable was indoor environment level. Other significant variables in order of decreasing importance included indoor safety and cleanness, cleanness around the house, safety around the house, construction year, air pollution around the house, and natural environment.

4.5.6. Residential Environment Satisfaction Analysis Results of Metropolitan and Non-Metropolitan Regions in the Conventional Model

The results of the metropolitan cities were as follows: The most important variable was accessibility of cultural facilities. Other significant variables in order of decreasing importance included school district, noise around the house, availability of neighborhood relationships, accessibility of public facilities, air pollution around the house, accessibility of parking area, and indoor environment level.

The results of the non-metropolitan cities were as follows: The most important variable was accessibility of public transportation. Other significant variables in order of decreasing importance included safety around the house, pedestrian safety, accessibility of commercial facilities, natural environment, indoor environment level, and air pollution around the house.

4.5.7. Residential Satisfaction Analysis Results of Metropolitan and Non-Metropolitan Regions in the MMHN-Based Model

The results of the metropolitan cities were as follows:

ⓐ

Regarding physiological needs, the factor related to beneficiaries of the national basic livelihood security was significant.

ⓑ

Regarding safety needs, indoor environment level, indoor safety and cleanness, cleanness around the house, safety around the house, construction year, and air pollution around the house were significant in order of decreasing importance.

ⓒ

Regarding love and belonging needs, availability of neighborhood relationships and accessibility of public transportation were significant in order of decreasing importance.

ⓓ

Regarding esteem needs, natural environment and housing prices were significant in order of decreasing importance.

ⓔ

Regarding self-actualization needs, school district, accessibility of cultural facilities, accessibility of commercial facilities, dwelling area, and monthly income were significant in order of decreasing importance.

The results of the non-metropolitan cities were as follows:

ⓐ

Regarding physiological needs, beneficiaries of the national basic livelihood security was significant.

ⓑ

Regarding safety needs, indoor environment level, indoor safety and cleanness, cleanness around the house, safety around the house, air pollution around the house, and construction year were significant in order of decreasing importance.

ⓒ

Regarding love and belonging needs, accessibility of public transportation, availability of neighborhood relationships, and marital status were significant in order of decreasing importance.

ⓓ

Regarding esteem needs, natural environment and housing prices were significant in order of decreasing importance.

ⓔ

Regarding self-actualization needs, school district, accessibility of commercial facilities, dwelling area, and monthly income were significant in order of decreasing importance.

4.5.8. Residential Environment Satisfaction Analysis Results of Metropolitan and Non-Metropolitan Regions in the MMHN-Based Model

The results of the metropolitan cities were as follows:

ⓐ

Regarding safety needs, accessibility of parking area, accessibility of medical facilities, accessibility of parking area, cleanness around the house, pedestrian safety, indoor environment level, indoor safety and cleanness, and noise around the house were significant in order of decreasing importance.

ⓑ

Regarding love and belonging needs, availability of neighborhood relationships and accessibility of public transportation were significant in order of decreasing importance.

ⓒ

Regarding esteem needs, natural environment was significant.

ⓓ

Regarding self-actualization needs, accessibility of commercial facilities, school district, and accessibility of cultural facilities were significant in order of decreasing importance.

The results of the non-metropolitan cities were as follows:

ⓐ

Regarding safety needs, safety around the house, accessibility of medical facilities, accessibility of parking area, cleanness around the house, pedestrian safety, indoor environment level, indoor safety and cleanness, and noise around the house were significant in order of decreasing importance.

ⓑ

Regarding love and belonging needs, availability of neighborhood relationships and accessibility of public transportation were significant in order of decreasing importance.

ⓒ

Regarding esteem needs, natural environment and housing prices were significant in order of decreasing importance.

ⓓ

Regarding self-actualization needs, accessibility of commercial facilities, school district, and dwelling area were significant in order of decreasing importance.

4.6. Commonalities and Differences in the Results of CART Analysis

Commonalities between the conventional and MMHN-based models and common significant variables across different regions are summarized in a table (see Appendix I). Likewise, differences between the conventional and MMHN-based models and differences with regard to significant variables across different regions are summarized in

Table 6. Differences in terms of significant factors between the conventional and MMHN-based models.

	Dependent Variable	Significant Factors
		Conventional Model		MMHN-Based Model
Nationwide	Residential Satisfaction	safety around the house	HC	beneficiaries of the national basic livelihood security	PhN
				marital status	SN
				children	LBN
				house management expenses	LBN
				availability of neighborhood relationships	LBN
				accessibility of public transportation	LBN
				age	EN
				housing prices	EN
				interior quality	EN
				accessibility of natural environment	EN
				educational background	SAN
				monthly income	SAN
				dwelling area	SAN
				accessibility of commercial facilities	SAN
				accessibility of cultural facilities	SAN
	Residential Environment Satisfaction	pedestrian safety	REC	beneficiaries of the national basic livelihood security	PhN
				indoor environment level	SN
				cleanness around the house	SN
				accessibility of public transportation	LBN
				housing prices	EN
				interior quality	EN
				accessibility of natural environment	EN
				dwelling area	SAN
Capital region	Residential Satisfaction	accessibility of public facilities	REC	accessibility of public facilities	SN
		availability of neighborhood relationships	REC
	Residential Environment Satisfaction	indoor environment level	HC
		pedestrian safety	REC	pedestrian safety	SN
		accessibility to cultural facilities	REC
		accessibility to parking area	REC
Non-capital regions	Residential Satisfaction	construction year	HC	beneficiaries of national basic livelihood security	PhN
		natural environment	REC	air pollution around the house	SN
		safety around the house	REC	safety around the house	SN
		noise around the house	REC	marital status	LBN
		cleanness around the house	REC	age	EN
				interior quality	EN
	Residential Environment Satisfaction	accessibility to commercial facilities	REC	beneficiaries of national basic livelihood security	PhN
		natural environment	REC	cleanness around the house	SN
		noise around the house	REC	accessibility of medical facilities	SN
		cleanness around the house	REC	dwelling area	SAN
		air pollution around the house	REC
Metropolitan cities	Residential Satisfaction	noise around the house	REC	noise around the house	SN
		accessibility to parking area	REC	accessibility of parking areas	SN
				accessibility of public facilities	SN
				accessibility of cultural facilities	SAN
	Residential Environment Satisfaction	school district	REC
		accessibility of parking areas	REC
		availability of neighborhood relationships	REC
		noise around the house	REC
		accessibility of public facilities	REC	accessibility of public facilities	SN
		accessibility of cultural facilities	REC	accessibility of cultural facilities	SAN
Non-metropolitan cities	Residential Satisfaction	safety around the house	REC	safety around the house	SN
		air pollution around the house	REC	air pollution around the house	SN
				marital status	LBN
	Residential Environment Satisfaction	pedestrian safety	REC	pedestrian safety	SN
		natural environment	REC	cleanness around the house	SN
		accessibility of public transportation	REC	housing prices	EN
		safety around the house	REC	dwelling area	SAN
		accessibility of commercial facilities	REC

HC and REC refer to house characteristics and residential environment characteristics, respectively. PhN, SN, LBN, EN, and SAN refer to physiological needs, safety needs, love and belonging needs, esteem needs, and self-actualization needs, respectively.

.

4.6.1. Commonalities of the Conventional and MMHN-Based Model in the Nationwide Residential Satisfaction Analysis Results

House characteristics, such as construction year, indoor environment level, and indoor safety and cleanness, and residential environment characteristics, such as cleanness around the house, air pollution around the house, accessibility to parking area, and school district, had a significant effect on residential satisfaction in both models.

4.6.2. Commonalities of the Conventional and MMHN-Based Models in the Nationwide Residential Environment Satisfaction Analysis Results

House characteristics, such as indoor safety and cleanness, and residential environment characteristics, such as safety around the house, noise around the house, air pollution around the house, accessibility to medical facilities, accessibility to public facilities, accessibility to parking area, school district, accessibility to commercial facilities, and accessibility to cultural facilities, significantly affected residential environment satisfaction in both models.

4.6.3. Commonalities across the Capital and Non-Capital Regions in the Residential Satisfaction Analysis Results

In the case of the conventional model, house characteristics, such as indoor environment level and indoor safety and cleanness, were common significant factors.

In the MMHN model, factors influencing safety needs including construction year, indoor environment level, indoor safety and cleanness, pedestrian safety, and cleanness around the house; factors influencing love and belonging needs including neighborhood relationship and accessibility to public transportation; factors influencing esteem needs including housing prices and natural environment; and factors influencing self-actualization needs including monthly income, dwelling area, school district, accessibility to commercial facilities, and accessibility to cultural facilities were significant factors in both types of regions.

4.6.4. Commonalities across the Capital and Non-Capital Regions in the Residential Environment Satisfaction Analysis Results

In the conventional model, house characteristics, including indoor safety and cleanness, and residential environment characteristics, including safety around the house, accessibility to public facilities, neighborhood relation, accessibility to public transportation, and school district, were common significant factors.

In the MMHN model, factors influencing safety needs including construction year, indoor environment level, indoor safety and cleanness, noise around the house, air pollution around the house, accessibility to public facilities, and accessibility to parking area; factors influencing love and belonging needs including neighborhood relationship and accessibility to public transportation; factors influencing esteem needs including house price and natural environment; and factors influencing self-actualization needs including school district and accessibility to commercial facilities and cultural facilities were significant in both types of regions.

4.6.5. Commonalities across the Metropolitan and Non-Metropolitan Regions in the Residential Satisfaction Analysis Results

In the conventional model, house characteristics, such as construction year, indoor environment level, and indoor safety and cleanness, and environmental characteristics, such as natural environment and cleanness around the house, were common significant factors.

In the MMHN model, factors influencing physiological needs including beneficiaries of the national basic livelihood security; safety needs including construction year, indoor environment level, indoor safety and cleanness, and cleanness around the house; love and belonging needs including neighborhood relationships and accessibility to public transportation; esteem needs including housing prices and natural environment; and self-actualization needs including monthly income, dwelling area, school district, and accessibility to commercial facilities were common significant factors.

4.6.6. Commonalities across the Metropolitan and Non-Metropolitan Regions in the Residential Environment Satisfaction Analysis Results

In the conventional model, house characteristics, including indoor environment level, and residential environment characteristics, including air pollution around the house, were common significant factors.

In the MMHN model, factors influencing safety needs such as construction year, indoor environment level, indoor safety and cleanness, noise around the house, accessibility to medical facilities, and accessibility to parking area; love and belonging needs including availability of neighborhood relationships, accessibility to public transportation; esteem needs including natural environment; and self-actualization needs including school district and availability of commercial facilities accessibility were common significant factors.

4.6.7. Differences across the Conventional and MMHN-Based Models in the Nationwide Housing Satisfaction Analysis Results

Safety around the house, which is an environmental characteristic, was significant only in the conventional model.

Factors influencing physiological needs such as beneficiaries of the national basic livelihood security; love and belonging needs including marital status, children, house management expenses, availability of neighborhood relationships, and accessibility to public transportation; esteem needs including age, housing prices, interior quality, and accessibility to natural environment; and self-actualization needs including educational background, monthly income, dwelling area, accessibility to commercial facilities, and accessibility to cultural facilities were significant only in the MMHN model.

4.6.8. Differences across the Conventional and MMHN-Based Models in the Nationwide Residential Environment Satisfaction Analysis Results

Pedestrian safety, an environmental characteristic, was significant only in the conventional model.

Factors influencing physiological needs such as beneficiaries of the national basic livelihood security; safety needs including indoor environment level and cleanness around the house; love and belonging needs including accessibility to public transportation; esteem needs including housing prices, interior quality, and natural environment; and self-actualization needs including dwelling area were significant only in the MMHN-based model.

4.6.9. Differences across the Capital and Non-Capital Regions in the Housing Satisfaction Analysis Results

In the conventional model, residential environment characteristics including availability of neighborhood relationships and accessibility to public facilities were significant only in the capital region. On the other hand, house characteristics, such as construction year, and residential environment characteristics, such as natural environment, safety around the house, noise around the house, and cleanliness around the house, were significant only in the non-capital regions.

In the MMHN-based model, accessibility to public facilities, a factor influencing safety needs, was significant only in the capital region. On the other hand, factors influencing physiological needs including beneficiaries of the national basic livelihood security; safety needs including safety around the house and air pollution around the house; love and belonging needs including marital status; and esteem needs including age and interior quality were significant only in the non-capital regions.

4.6.10. Differences across the Capital and Non-Capital Regions in the Residential Environment Satisfaction Analysis Results

In the conventional model, the indoor environment level, which is a housing characteristic, and accessibility to cultural facilities, pedestrian safety, and accessibility to parking area, which are residential environment characteristics, were significant only in the capital region. On the other hand, residential environment characteristics, such as accessibility to commercial facilities, natural environment, noise around the house, cleanness around the house, and air pollution around the house, were significant only in the non-capital regions.

In the MMHN-based model, pedestrian safety, a factor influencing safety needs, was significant only in the capital region. On the other hand, factors influencing physiological needs including beneficiaries of the national basic livelihood safety; safety needs including cleanness around the house and accessibility to medical facilities; and self-actualization needs including dwelling area were significant only in the non-capital regions.

4.6.11. Differences across the Metropolitan and Non-Metropolitan Regions in the Housing Satisfaction Analysis Results

In the conventional model, residential environment characteristics including noise around the house and accessibility to parking area were significant only in metropolitan cities. On the other hand, residential environment characteristics such as safety around the house and air pollution around the house were significant only in non-metropolitan regions.

In the MMHN-based model, factors influencing safety needs, such as noise around the house, accessibility to public facilities, accessibility to parking areas, and factors influencing self-actualization needs, including accessibility to cultural facilities, were significant only in metropolitan cities. On the other hand, factors influencing safety needs, including safety around the house and air pollution around the house, and factors influencing love and belonging needs, including marital status, were significant only in non-metropolitan regions.

4.6.12. Differences across the Metropolitan and Non-Metropolitan Regions in the Residential Environment Satisfaction Analysis Results

In the conventional model, residential environment characteristics, such as school district, accessibility to cultural facilities, availability of neighborhood relationships, noise around the house, accessibility to public facilities, and accessibility to parking area, were significant only in metropolitan cities. On the other hand, residential environment characteristics, including accessibility to commercial facilities, natural environment, accessibility to public transportation, safety around the housing, and pedestrian safety, were significant only in non-metropolitan regions.

In the MMHN-based model, accessibility to public facilities, a factor influencing safety needs, and accessibility to cultural facilities, a factor influencing self-actualization needs, were significant only in metropolitan cities. On the other hand, factors influencing safety needs including pedestrian safety and cleanliness around the house; esteem needs including housing prices; and self-actualization needs including dwelling area were significant only in non-metropolitan regions.

5. Discussion

In this study, we explored the differences between the MMHN-based model and the conventional model using CART analysis. To this end, the differences in residential satisfaction factors derived from the CART analysis results based on the conventional model and the MMHN-based model were analyzed.

Three major differences were observed in the CART results based on the MMHN-based model. First, the influence of factors corresponding to love, respect, and self-actualization needs on residential satisfaction can be more clearly identified. When comparing the conventional model and the MMHN-based model, the results show that the factors corresponding to love and belonging needs, such as availability of neighborhood relationships; respect needs, such as natural environment; and self-actualization needs, such as accessibility of cultural facilities, are significant only in the MMHN-based model. Therefore, the MMHN-based model can better reflect the influence of various human psychological needs on residential satisfaction.

Secondly, the MMHN-based model can identify the influencing factors related to residential satisfaction in non-metropolitan areas in more detail. According to the MMHN-based model, beneficiaries of the national basic livelihood security, safety around the house, air pollution around the house, marital status, age, and interior quality are all factors that have a significant effect only in non-metropolitan areas, which is a result that was not obtained in the analysis based on the conventional model, demonstrating that the MMHN-based model can broaden the scope of understanding of residential satisfaction in non-metropolitan areas.

Third, the MMHN-based model can more clearly identify the differences in residential satisfaction factors between metropolitan and non-metropolitan areas. Compared to the conventional model, we also found that accessibility of parking areas and accessibility of cultural facilities are significant only in metropolitan cities when using the MMHN-based model, while marital status is significant only in non-metropolitan cities. Therefore, the MMHN-based model can shed additional light on differences in housing needs according to regional characteristics.

However, we also identified commonalities between the two models. For example, indoor environment level and indoor safety and cleanness have an important influence on residential satisfaction, regardless of the model used, in the whole of Korea, as well as in metropolitan and non-metropolitan cities, which means that these factors have considerable influence on residents’ perception of housing regardless of the regional context. In addition, school district and neighborhood relationships have an important influence on satisfaction with the residential environment.

6. Conclusions

In this study, we reconstructed residential satisfaction factors based on the MMHN concept and analyzed the differences between the CART analysis results based on the conventional model and those based on the MMHN-based model in terms of human psychology. Furthermore, regional characteristics were taken into consideration to further examine the differentiating factors of the MMHN-based model. The results show that the MMHN-based model not only can reflect various human needs better than the conventional model but it can also be used to assess the differences in housing needs of residents living in different regions. Thus, this study can contribute to housing research by integrating the psychological concept of Maslow’s hierarchy of human needs with the concept of residential satisfaction and presenting a new framework for the examination of housing from the perspective of residents’ psychological needs. Additionally, the results of the regional analysis suggest that the MMHN-based model can also be utilized to support region-specific housing policies by assessing the housing needs of residents in specific regions.

However, this study is based on a 2020 residential survey dataset, and due to the nature of the survey sample, the conclusions of this study cannot be generalized to the characteristics of Korea over time. Therefore, in future research, the proposed framework should be applied to analyze the Korea Housing Survey data in a time series, derive research results from the psychological aspect of changes in factors influencing residential satisfaction among Koreans over time, and analyze the change patterns of psychological factors and general characteristics of Koreans in relation to housing. In addition, our analysis of the Korea Housing Survey data divided Korea into capital and non-capital regions or metropolitan and non-metropolitan regions. However, each regional category might contain both urban and rural regions, making it difficult to clearly characterize each region. In the future, more in-depth research using data division may further elaborate the findings of this research by more clearly classifying regions according to their characteristics, e.g., metropolitan cities, small cities, rural regions, and mountainside areas.