Dairy purchase behaviors: Increasing understanding of Chinese consumers using a consumer involvement segmentation approach

Understanding consumers’ purchase behaviors is fundamental to the success of the dairy industry. With its economic importance, the Chinese market is critical to dairy producers in most countries around the world. However, understanding consumers in this market is particularly challenging, as these consumers often have a different relationship with dairy products than consumers elsewhere in the world, given the country’s historical dairy-related scandals. This special relationship can be characterized by what consumer behavior researchers call “high involvement,” indicating that Chinese dairy consumers often attempt to reduce the level of risk associated with buying dairy products. Surpris-ingly, although this relationship affects several important aspects of purchase behavior, examination of the concept of involvement in the dairy sector has not been widely considered. Of note, there is no understanding of how Chinese consumers vary in their involvement levels and their implications on their dairy purchase behaviors. Nor have there been involvement-based insights provided on how dairy companies can position their marketing strategy to suit the needs of these consumers better. Thus, this study proposes a new approach to understanding Chinese consumer dairy decisions by introducing “consumer involvement” as a segmentation tool through which individuals’ behaviors can be predicated according to their involvement profile. Based on an online survey administered in Shanghai using 1,073 dairy consumers, principal component analysis confirmed involvement with dairy is a multidimensional construct with the following 4 factors: pleasure value, symbolic value, risk importance, and risk probability. A 2-step cluster analysis identified 4 consumer clusters based on their involvement profile: face-concerned dairy lover, carefree dairy consumer, cautious dairy lover, and confused dairy consumer. According to a one-way ANOVA test and cross-tabulation with χ 2 test, these consumer segments behave differently in relation to the extensiveness of decision making, cue utilization, trust of information sources, and consumption behavior. The outcomes in this paper further explain why efforts to restore consumer trust for dairy products do not work among some consumers, as individuals may exhibit diverse attitudes toward such information due to their heterogeneous involvement levels. The study also provides suggestions for market practitioners and organizations to develop effective target market strategies and policies according to different consumer clusters


INTRODUCTION
The dairy sector in China is a big deal.Dairy is now considered to be an essential part of a healthy diet for Chinese consumers, and the Chinese government has supported the growth of this industry (Wu et al., 2018;Maitiniyazi and Canavari, 2020).The growth of the industry has been impressive.According to the China Dairy Quality Report (Ming, 2022), the demand for dairy products in China increased significantly from 2019 to 2021, with the per capita consumption of dairy products in China being 35.8 kg in 2019, 38.3 kg in 2020, and 42.3 kg in 2021, an increase of 4.4%, 7%, and 10.4%, respectively.This increase exceeds the average growth rate of 3.6% in the past decade.In addition, in 2019, the national dairy product output was 27.2 million tonnes, an increase of 5.6% over the previous year.The output of liquid milk was 25.4 million tonnes, an increase of 5.8% over the previous year, the output of cheese dairy products was 1.8 million tonnes, an increase of 2.5% over the previous year, and the output of milk powder was 1.1 million tonnes, an increase of 2.4% over the previous year (Wang, 2020).Given that the dairy sector aims to maximize opportunities in this complex market, it is important to gain a detailed understanding of how domestic consumers are involved in the purchase and consumption of dairy products.
Consumers in this market have a complex relationship with the dairy sector, with the industry often criticized by the public because of the frequent dairy scandals, such as the melamine-contaminated baby formula event in 2008, and illegal additions to dairy products of hydrolyzed proteins in 2009, mercury in 2012, and detergent in 2012 (Wu et al., 2018).These incidents have dramatically enhanced Chinese consumers' perceived risk related to consuming dairy products, because they are highly concerned that making a wrong choice could have a severe health effect.
The risk-sensitive nature of these domestic consumers fundamentally changes how they purchase dairy products (Li et al., 2019).To reduce the level of risk associated with buying dairy products, some consumers are now motivated to employ a range of prepurchase search strategies, such as seeking reliable and trustworthy information about the companies, brands, and products to conduct a comprehensive evaluation and make reasoned decisions when buying such products (Liem et al., 2016;Maitiniyazi and Canavari, 2020).Given this context, we cannot automatically assume that Chinese consumers' dairy purchases are similar to most other consumer food and beverage purchases, which are usually considered to be low involvement.
Given that the Chinese government and enterprises have found limited effects of attempts to restore consumers' confidence and trust in dairy products through tracking production systems, improving standards, and providing third-party certification (Bai et al., 2013;Li et al., 2019), it makes sense to start to think differently about why consumers are behaving the way they are in relation to their dairy purchases.Research has found that some groups of consumers have a low level of interest in safety policies (e.g., policies relating to certification), and consumers still do not trust domestic products and view imported dairy as superior (Liem et al., 2016;Yang et al., 2018).Such results might depend on and be explained by individuals' involvement levels.Thus, whether government policies attract consumers' attention and are reflected in purchase behaviors should be assessed in relation to the dimension of consumer involvement.Given that previous research has limited contributions focusing on the motivations and driving forces for dairy purchase in China, this paper adopts a novel perspective to explore domestic dairy purchase decisions by applying consumer involvement as a segmentation tool.
Overall, this study aims to investigate consumer involvement as a segmentation variable to identify and characterize distinct consumer clusters in relation to their dairy-related behaviors.To the best of our knowledge, this is the first time research has linked the concept of consumer involvement with dairy consumption; thus, the study provides a fresh lens through which to understand consumer behavior.More specifically, the research objectives of this paper are as follows: (1) to explore the antecedents of consumer involvement with dairy products; (2) to segment consumers based on their involvement profile; and (3) to investigate the effect of involvement on its possible behavioral consequences (i.e., the effect of the extensiveness of decision making, preference for cue utilization on quality evaluation, trust of information sources, and consumption habits).The outcome of this research allows segmentation to be firmly embedded in an organization, inform all marketing activities from product development, inform above and below-the-line communication activity, and other portfolio management activities.It will not only provide valuable knowledge to dairy enterprises in designing and targeting business strategies to support further growth but will also assist governments to form policies targeted to different clusters of consumers.
Consumer involvement refers to the perceived relevance, interest, and importance attached to the specific product category (Zaichkowsky, 1985;Havitz and Mannell, 2005).Owing to its abstract nature, involvement has been regarded as a latent construct measured by 5 antecedent variables: product importance, hedonic value, symbolic value, risk importance, and risk probability (Laurent and Kapferer, 1985).Consumers are likely to be involved in products when these products are perceived as being important in catering to individuals' needs, providing hedonic value, reflecting on self-image, and being associated with a high perceived risk (e.g., the perceived probability of making wrong choices and perceived importance of the negative consequences of a mispurchase; Verbeke and Vackier, 2004;Charters and Pettigrew, 2006;Rahman and Reynolds, 2015).
As consumer involvement is recognized as an essential predictive variable in purchase behavior, numerous studies have summarized the considerable influence of consumer involvement in relation to the following factors: (1) extensiveness of decision making (e.g., spending much time on the purchase decision, comparing many product alternatives, actively searching for product-relevant information, and consulting and exchanging opinions about the products with others; Verbeke and Vackier, 2004;Laurent and Kapferer, 1985); (2) cue utilization (i.e., the utilization for product attributes; Josiassen et al., 2008;Bruwer et al., 2017a;Koksal, 2021); (3) trust in different information sources (both social media information and personal source information; Bruwer et al., 2014;Koksal, 2021); and (4) different consumption habits (e.g., purchasing frequency and preferred shopping locations; Mittal and Lee, 1989;Beharrell and Dennison, 1995;Thach and Olsen, 2015).The overall theoretical framework of consumer involvement is shown in Figure 1.In sum, given that significant variations are found among consumers toward specific product categories, research must translate different levels of involvement into different behavioral consequences.
The involvement studies in the food sector have generally been overlooked, as food is mostly low cost, low risk, and has a low potential to reflect self-image, which leads to a lack of stimulation motivating consumers for deeper engagement in the purchase decision (Beharrell and Dennison, 1995;Verbeke and Vackier, 2004;Espejel et al., 2009).Nevertheless, such a statement may need to be revised to analyze some food products associated with high perceived risk, such as fresh meat consumption (Verbeke and Vackier, 2004).The study of dairy products in China further supports the argument that consumers do not always have low involvement in food purchase decisions, due to the increased perceived risk after a series of dairy scandals.In addition to risk perception potentially enhancing Chinese consumers' involvement, the role of symbolic value in determining consumer involvement is notable in Chinese culture.Given the symbolic value that food has acquired (Robinson and Getz, 2016), consumers increasingly feel their food choices can identify them, a shift that can explained by the saying, "You are what you eat" (Wang et al., 2019).Chinese consumers whose purchase behavior is heavily influenced by face culture (e.g., miànzi, which refers to a cultural understanding of respect, honor, and social standing) are actively engaged in purchase decisions with the kind of products that can carry symbolic benefits (Hu, 1944;Wang et al., 2019;Xu et al., 2020).Extensive papers prove that price and brand significantly positively affect consumers' face perception and symbolic value (Wang and Griskevicius, 2014;Zhang and Wang, 2019;Xie and Shi, 2022).Dairy products in China tend to be considered high end; for example, Li and Cui (2021) find that dairy products, such as organic milk, Jersey milk, cheese, and yogurt, are perceived as high-end protein supplement products that are associated with social status signals, given their high quality and relatively high price.In sum, Chinese consumers tend to be highly motivated to become involved in dairy purchase decisions when these products could bring perceived symbolic value and imply an individual's social status.Thus, it is appropriate and timely to study consumer involvement with dairy products in China.

Data Collection
An online survey was conducted in Shanghai through Wenjuanxing (an online panel data service similar to  (Laurent and Kapferer, 1985;Verbeke and Vackier, 2004).
Qualtrix), China's largest online survey platform, in June 2021.Respondents received reward points directly from the platform, and these points in their Wenjuanxing account can be exchanged for currency.
Shanghai is China's economic and financial hub, due to the concentration of commercial resources, accessibility of transit network, and variety of opportunities for its urban residents.Hence, it is home to a large consumer base and is commonly used as a research target in many similar dairy studies (Li et al., 2019(Li et al., , 2021;;Zhu et al., 2019).

Sociodemographic Characteristics of the Respondents
A total of 1,073 valid questionnaires were gathered (see Table 1), and 117 incomplete questionnaires were removed.In the sample, 53.45% of participants were female, indicating that women are responsible for buying dairy products in a majority of Chinese families.Notably, most participants drank milk either daily (52.7%) or weekly (42.5%).They mainly purchase dairy for themselves or their children.Although dairy is relatively new as a daily consumption product in China, 28.4% of the consumers responded that they are knowledgeable about dairy products, and 69.2% think they understand dairy information on a basic level.

Questionnaire Development
The questionnaire contains measurement scales on 5 subdimensional variables of consumer involvement, 4 behavioral consequences for dairy consumption, and several sociodemographic questions.All the measurement scales in this study are based on the nature of the data, and all measurement items were adapted to the context of dairy consumption in China.
Given that most studies have found that consumer involvement should be examined as a multidimensional construct (e.g., Laurent and Kapferer, 1985;Verbeke and Vackier, 2004;Kim, 2005), rather than capturing its characteristics by the single factor of product importance, this study measured consumer involvement using consumer involvement profile (CIP), which was proposed by Laurent and Kapferer (1985).The CIP is composed of 5 factors (i.e., subdimensional variables): (1) product importance: consumers' perceived ego-importance in relation to the product; (2) hedonic value: consumers' perceived pleasure or reward value associated with the products; (3) symbolic value: consumers' perceived self-expression and self-identity attached to the product; (4) risk importance: consumers' perception of the importance of the negative consequences of a mispurchase related to the product; and (5) risk probability: consumers' perceived probability of making poor choices related to the product.The 5 subdimensional variables for consumer involvement with 13 items were measured on 7-point Likert scales (1 for "fully disagree" to 7 for "fully agree").
The 4 behavioral consequences are (1) the extensiveness of decision making, (2) trust of information sources, (3) cue utilization, and (4) consumption habits.The extensiveness of decision making was measured using 4 items on 7-point Likert scales (ranging from 1 = "super few" to 7 = "a lot of") that measured time spent, alternatives compared, utilization of information, and opinion consultation (Verbeke and Vackier, 2004).Similarly, trust of information sources measured social media and personal information on 7-point Likert scales.Given the efficiency and importance of communication, it is essential to investigate the effect of consumer trust in relation to different information sources (Li et al., 2021).The sources examined in this study were social media (advertisements, food bloggers, salesperson recommendations, experts' suggestions, and celebrity recommendations) and personal sources (friends and family).Cue utilization was conceptualized as a series of cues the consumer would utilize during a dairy purchase; this was also measured on 7-point Likert scales (ranging from 1 = "not important" to 7 = "extremely important").The dairy cues used were texture, taste, protein content, fat content, calcium content, country of origin (CoO), organic, price, packaging, retailer information, ownership labeling, process information, and third-party information (Wibowo et al., 2019;Mirosa et al., 2020).
Finally, the effects of consumer involvement on dairy consumption habits have been measured by both categorical variables and 7-point Likert scales; specifically, categorical variables measured consumption frequency and favorite purchasing locations, and 7-point Likert scales measured recent shift in dairy consumption, shift in dairy consumption because of the melamine contamination controversy, and intention for future dairy consumption.

Involvement as Segmentation Tool
Identifying market segmentation has been a topic of interest in recent consumer research studies, as it enables practitioners to discover distinct market opportunities that can be utilized to ensure effective marketing strategies and resource allocation (Bruwer et al., 2017b;Koksal, 2021;Nassivera et al., 2020).Market segmentation is defined as the identification of homogeneous groups or customer segments in the market, which will respond to changes in marketing mix in a consistent and predictable manner (Koksal, 2021).As the consumers become fragmented and the market becomes more complex, practitioners take segmentation as a meaningful tool for reducing such complexity to a manageable level (Aurifeille et al., 2002;Morris and Schmolze, 2006;Nassivera et al., 2020).
Early studies conduct segmentation based on sociodemographic factors, such as gender, income, or social class (Tassiopoulos et al., 2004;Taylor, 2004;Cortiñas et al., 2019).However, segments driven by sociodemographic variables may exhibit little difference in terms of psychographic measures such as attitudes and values, because demographic variables are known to be poor predictors of consumer behavior (Brown et al., 2007;Nassivera et al., 2014).Morris and Schmolze (2006), for instance, find that segmentation based on sociodemographic information exhibit little difference in terms of attitude and purchase habits for products, therefore, failing to provide a good framework for new product development.More recently, studies such as Koksal (2021) and Molina et al. (2015) found that psychological variables, such as motivation and personal traits, can better predict customers' purchase decisions and explain the purchase process (Molina et al., 2015;Koksal, 2021).
In this paper, consumer involvement is seen as a pivotal psychological construct to capture consumer heterogeneity, because it has a high predictive value for consumer consumption and purchase behavior (Brown et al., 2007;Broderick et al., 2007).Consumer involvement represents the motivational force leading to different behavioral consequences (Verbeke and Vackier, 2004;Kim, 2005), such as extensiveness of decision making, preference for product attributes, trust in different information sources, and different consumption habits.Individuals normally show dramatic differentiation among involvement profiles even within the same product.

Data Analysis
This study conducts a factor-cluster analysis to segment Chinese dairy consumers using 3 statistical procedures which corresponded to the 3 research questions (see Figure 2).Data were analyzed through IBM Statistical Product and Service Solutions Statistics (SPSS) 25.As there are no previous studies focusing on consumer involvement with dairy products, the first step applied exploratory factor analysis using the principal component analysis (PCA) with varimax rotation for initial CIP to identify the real involvement dimension for dairy consumption.The rotation of factors makes the extracted factors easier to interpret with clearer social meanings (VanPool and Leonard, 2011).The varimax method is effectively used in most consumer research literature (McHugh, 1999;Bai, 2022).Only components with eigenvalues larger than one and items with factor loadings higher than 0.5 were kept per standard criteria proposed by Kaiser (1974).Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett's test were performed before factor analysis to confirm the possible intercorrelation between the studied variables and ensure the data were adequate for applying factor analysis (Broen et al., 2015).Then, the reliability analysis for internal consistency in each factor was tested using Cronbach's α with a 0.7 threshold value (Jarvis et al., 2003).The study's second objective is to identify the distinct dairy consumer clusters based on individual involvement profiles.In this context, the second stage of this study employed a 2-step cluster analysis method to identify distinct dairy consumer clusters.Compared with other traditional cluster methods, 2-step cluster method has several distinct advantages (Kent et al., 2014;Norusis, 2010), such as providing an exploratory identification of numbers of clusters based on statistical measures of fit (e.g., Schwarz's Bayesian criterion) rather than being chosen subjectively by scholars, having the capability of handling large data sets (n > 1,000) and dealing with atypical values (e.g., outliers).As its name indicates, 2-step cluster analysis realized the segmentation through 2 independent steps.In the first step, the cluster feature tree is generated according to distance (e.g., log-likelihood distance) to create the preclusters among the original data sets (Bacher et al., 2004;Schiopu, 2010).Then, the preclusters identified in the previous step were statistically merged into the optimal number of final clusters based on the Bayesian information criterion (BIC) with the lowest BIC value (Bacher et al., 2004;Schiopu, 2010).The cluster quality was evaluated by average silhouette value and the ratio of sizes (the ratio of the largest cluster to the smallest cluster; Kent et al., 2014;Supandi et al., 2021).The average silhouette value explains how the cluster is far from other clusters ranging from −1 to 1, with a cutoff larger than 0.2 (Supandi et al., 2021).In addition, the threshold value of the size ratio should be smaller than 2 (Kent et al., 2014).Once the clusters were confirmed, one-way ANOVA tests of mean scores with Tukey post hoc multiple comparisons were conducted to explore whether the clusters showed statistical differences in terms of each involvement factor, and cross-tabulation with χ 2 tests was performed to detect the sociodemographic variations between different clusters.
Finally, to investigate the effect of involvement on possible behavioral consequences, one-way ANOVA Ftests of mean scores (with Tukey post hoc multiple comparison tests) were used to analyze differences across involvement-based clusters based on the extensiveness of decision making, trust of information sources, cue utilization, and purchase habit.Cross-tabulation with χ 2 association tests was conducted to access the differences between consumer segments in terms of consumption habits, consumption frequency, and favorite purchasing locations.

Involvement Profile in Dairy Products
Principal component analysis using varimax rotation is performed on 13 involvement items.The KMO index indicating sampling adequacy is 0.788, and a significant Bartlett's test with χ 2 = 5,040.641(df = 105; P = 0.000) suggests that the sample is suitable for factor analysis.Based on the criteria conducted by Kaiser (1974) that an item should be deleted if its factor loading is less than 0.5, one item from symbolic value and one item from risk importance are excluded due to their low factor loadings (below 0.5).
Four dimensions of product involvement using 11 items are confirmed (see Table 2), accounting for 65.6% of the overall cumulative variance.Two factors, product importance and hedonic value, are loaded on the same dimension and showed a statistically significant correlation (P < 0.001).This dimension may also be referred to as pleasure value, given that it incorporates consumers' ego-importance and the pleasure of consuming dairy products.Internal consistency reliability tests are assessed for the combination of items representing these 4 dimensions.The reliability coefficients (Cronbach's α) for these 4 dimensions are higher than or close to the recommended standard of 0.70 (Jarvis et al., 2003), ranging from 0.684 to 0.96.Harman's single-factor test with a nonrotated exploratory factor analysis is performed to examine common method bias (Podsakoff et al., 2003).As a result, the dominant factor explains only 30% of the total variance, which is below the threshold of 40%, meaning that there is no major concern with common method bias.
The average score for pleasure value is 5.63 (SD = 1.19), and the average score for symbolic value is 4.27 (SD = 1.53).This indicates that dairy, similar to other food products, is associated with a significant hedonistic or enjoyment value, and purchasing dairy products in China may serve as a personal social state signal to others.The average scores for the 2 subdimensions (risk importance and risk probability) of perceived risk are relatively low.Specifically, consumers believe they have a low likelihood of making a poor dairy choice (risk probability: average = 3.82; SD = 1.61), and it is notable that Chinese consumers attach less importance to the potential negative consequences of a poor decision (risk importance: average = 3.17; SD = 1.43).This indicates that purchasing dairy in China is not considered highly risky and includes few obstacles for consumers.

Cluster Analysis
Based on the respondents' pleasure value, symbolic value, risk importance, and risk probability scores, a 2-stage cluster analysis confirms 4 distinct consumer groups.The pleasure value shows the highest input (predictor) importance for clustering (1.00), followed by risk importance (0.96), and symbolic value (0.86; see Figure 3).It further demonstrates that the 4 cluster results are fair, because the average silhouette value is 0.3, which is above the cutoff value of 0.2 (the lowest fair threshold is 0.2).In the case of the ratio of sizes, our results show that the largest cluster and smallest cluster contain 301 and 235 respondents, respectively, and the ratio value is 1.28, which meets the criterion of below 2. The scores of 4 involvement factors among different clusters are presented in Table 3 with the results of the one-way ANOVA and the post hoc Tukey multiple comparison tests.
The first consumer cluster (26.1% of the sample) is identified as "face-concerned dairy lovers," who enjoy consuming or eating dairy products and do not perceive many risks in their purchasing process.This group has the highest score on symbolic value, which means that this group of consumers is sensitive to the personal symbolism attached to dairy products.The second cluster is the "carefree dairy consumer" (28.1% of the sample).This cluster's respondents report a low perceived risk of dairy purchase.They do not think the probability of making the wrong choice is high and do not take seriously the negative consequences of a mispurchase.Furthermore, these consumers report an extremely low score on symbolic value.The third cluster is identified as the "cautious dairy lover" (24.0% of the total sample).This group reports that purchasing and eating dairy products makes them feel happy and joyful, and they perceive dairy has express functions, signifying individuals' self-image in some extent.Furthermore, this group has the highest level of risk importance, meaning the respondents in this cluster are cautious about mispurchase and attach a great deal of weight to the eventual negative consequences of a bad purchase decision.The fourth cluster is the "confused dairy consumer" (21.9% of the total sample).This segment is not so concerned about making purchase mistakes but it does consider that purchasing dairy is a complicated and confusing task and finds it difficult to be certain about whether they make the right choice.In addition, this group does not experience pleasure value when eating or purchasing dairy and associates with relatively lower ego-importance to dairy.Face-concerned dairy lover: consumers who enjoy consuming or eating dairy products and do not perceive many risks in their purchasing process; carefree dairy consumer: consumers with a low perceived risk of dairy purchases; cautious dairy lover: consumers who feel happy when purchasing and eating dairy products; confused dairy consumer: consumers who are concerned about making purchase mistakes and view purchasing dairy as a complicated and confusing task. 1 = "fully disagree" to 7 = "fully agree."

Sociodemographic Effects on Cluster Analysis
Cross-tabulation assessment defines the profiles of each cluster, and a χ 2 statistic reveals whether there are substantial variations between the 4 clusters.According to the results, the 4 consumer groups differ considerably in relation to gender, income, product knowledge, number of children, and type of resident (urban or rural).However, no statistically significant differences exist in terms of age, education level, and overseas experience.
In relation to the effect of gender, there are more women in cluster 2, carefree dairy consumer (χ 2 = 12.33; P = 0.006), which has the lowest score for symbolic value.This is associated with a significantly lower factor score on symbolic value for females, according to the t-test (t = 2.59; P = 0.01).In addition, cluster 4, confused dairy consumer, has more men and the lowest factor score on pleasure value of all the consumer groups.This might be explained by males having a lower score than females on this factor (t = 2.31; P = 0.02).Additionally, data concerning the income level reveals that cluster 1, face-concerned dairy lover, includes a much higher proportion of dairy consumers (χ 2 = 24.10;P = 0.02) with an income of more than 15K CNY per month (18%), whereas for cluster 4, confused dairy consumer, the respective percentage (income of more than 15K CNY per month) is 12%.
Consumers who perceive themselves as knowledgeable about dairy products are significantly more represented in cluster 3, cautious dairy lover (χ 2 = 104.15;P = 0.00), associated with lower risk probability and higher risk importance.In addition, the confused dairy consumers of cluster 4 are represented by more consumers who are unfamiliar with product information and have the highest risk probability level.Thus, knowledge about dairy products tends to facilitate consumers in making the right decisions and not feeling puzzled when facing a variety of purchasing choices; however, the perceived knowledge does not mitigate risk perception in relation to the negative consequences of a purchase.
Families without children are clearly more represented in cluster 4 (χ 2 = 11.25;P = 0.01), and this group of consumers also includes more people living in rural areas (χ 2 = 11.62;P = 0.09) with lower incomes (χ 2 = 24.10;P = 0.02).The high representation of people living in rural areas might be why this consumer group is not motivated in dairy purchase decisions, in that most rural people (in China) do not consider dairy products part of their daily diet.This means that dairy products are relatively new to them and they consider the product information of dairy products is complex, making them feel confused.

Consequences of Consumer Behaviors
Extensiveness of Decision Making.Applied to purchasing dairy products, the scale for the extensiveness of decision making contains 4 items: alternatives comparison, time consumption, utilization of information, and consultation of the opinion of others.The KMO value is 0.72 with a significant Bartlett's test of sphericity, with χ 2 = 551.54(df = 6) and P = 0.000 (P < 0.05).The Cronbach's α of this scale is 0.66, which indicates that these 4 items measure one single construct.
Consequently, the average score on the 4 items can be interpreted as an indicator of the extensiveness of decision making and contrasted across the 4 customer groups.Table 4 reports the average scores of the 4 segments on the overall construct of extensiveness of decision making.The cluster 1 consumer group, faceconcerned dairy lovers, and cluster 3, cautious dairy lovers, are found to engage in the most extensive decision making for dairy products.When making dairy purchase decisions, these groups are willing to invest their time, and devote themselves to information searching, to make comparisons with similar alternatives, and seeking the advice of peers.The cautious dairy lover group, concerned about mispurchase and bad consequences, has the highest score in time consumption, searching information, and making comparisons.Decision making for dairy consumption seems far less crucial for the cluster 4 confused dairy consumer because they might have lower involvement in making complex dairy consumption decisions.
Trust in Information Source.The PCA confirms 2 principal components, with KMO = 0.893 and a significant Bartlett's test of sphericity results of χ 2 = 2,309.18(df = 36) and P = 0.000 (P < 0.05).Information derived from food bloggers, experts, salesperson recommendations, social media (e.g., Weibo, Xiaohongshu), celebrity recommendations, and advertisements, and these sources are loaded in the same factor and referred to as trust for social media (Cronbach's α = 0.76), which demonstrates that the consumers' judgments of all mass media sources are similar.The suggestion from family and friends are loaded into one factor referring as a trust for family or friend with Cronbach's α = 0.65.Generally, experts' suggestions are given the highest score for trust or social media (mean = 5.38).In contrast, salesperson recommendations (mean = 4.13) and celebrities (mean = 4.05) have a relatively lower effect on the consumers.In addition, opinions from family members (mean = 5.70) are more important than friends' suggestions (mean = 5.40) under the second dimension of "trust for family/friend."Discriminant analysis with ANOVA tests reveals significant differences in these 2 factors among the 4 clusters (see Table 4).For the factor of "trust for social mass media," the face-concerned dairy lover and cautious dairy lover consumer groups reveal a significantly stronger effect of mass media than for the carefree dairy consumer and confused dairy consumer groups (F = 59.44;P = 0.00).The cautious dairy lover group has the highest scores for experts' suggestions, which may be attributed to their perceived risk being reduced by information from experts.The face-concerned dairy lover consumer group, which enjoys the purchasing process, is more easily influenced by advertisements and social media information than any of the other 3 clusters.Finally, for the trust for family/friend dimension, the confused dairy consumer group has a significantly lower score because they have the lowest involvement with dairy consumption of all the consumer groups (F = 17.47;P = 0.000).
Cue Utilization.Three principal components are explored using PCA with KMO = 0.88, with significant Bartlett's test (χ 2 = 3338.39;P = 0.00).The first component of cue utilization is extrinsic cues (Cronbach's α = 0.81), which are cues that are not part of the physical product, for example, labeling information, CoO, packaging, ownership labeling, process information, retailer information, third-party certification, and organic labeling.The second component of cue utilization (Cronbach's α = 0.57) is intrinsic cues, which are productrelated attributes such as taste and texture.Product protein, fat, and calcium content are loaded in the same factor under nutrition cue, the third component of cue utilization (Cronbach's α = 0.70).Respondents generally score the highest in CoO labeling in the extrinsic cue dimension, product taste in the intrinsic dimension, and protein content in the nutrition dimension.
A one-way ANOVA employing a Tukey post hoc test shows that the 4 groups differ significantly in relation to cue utilization (see Table 4).The face-concerned dairy lover and cautious dairy lover consumers clearly have a higher interest in extrinsic and intrinsic cues.CoO information and third-party certification are the 2 product attributes they value most among the dimensions of extrinsic cues.In contrast, the carefree dairy consumer and confused dairy consumer groups judge quality according to packaging information.Finally, for the nutrition cues, the results of the 3 clusters other than the confused dairy consumer cluster are homogeneous, with high scores on this dimension.However, all 4 consumer groups give the same ranking for different nutrition cues: protein content ranks first, followed by calcium content, and fat content last.
Consumption Habits.Overall, the consumers' confidence in dairy products gradually recovered after the melamine-contaminated milk powder incident, with a lower score on "decreased dairy consumption in the recent past years" (mean = 3.35; SD = 1.76) in contrast with "decreased dairy consumption after melamine scandals" (mean = 4.10; SD = 1.83).Significant discrimination effects are discovered in relation to the reported consumption decreases due to the melamine scandal (F = 12.05; P = 0.00), with significantly more confused dairy consumers reporting unwillingness for purchase intention compared with the other 3 groups.In addition, there is evidence that the confused dairy consumer group is more likely than the other clusters to reduce dairy purchases in the future (F = 28.71;P = 0.00).The one-way ANOVA results for the items: "decreased dairy consumption after melamine scandals," "decreased dairy consumption in the recent past years," and "to reduce dairy purchase in the future" are presented in Table 5.  Scores within a row with a different superscript are significantly different at P < 0.05 (one-way ANOVA and post hoc Tukey multiple comparison test); "a" means the accordance value is relatively higher, followed by "b" and "c." 1 Face-concerned dairy lover: consumers who enjoy consuming or eating dairy products and do not perceive many risks in their purchasing process; carefree dairy consumer: consumers with a low perceived risk of dairy purchases; cautious dairy lover: consumers who feel happy when purchasing and eating dairy products; confused dairy consumer: consumers who are concerned about making purchase mistakes and view purchasing dairy as a complicated and confusing task. 1 = "fully disagree" to 7 = "fully agree." The highest category of consumption frequency was daily consumption, with 52.7% of respondents.This is followed by weekly consumption, 42.55%.The difference analysis (F = 72.55;P = 0.00) found that the confused dairy consumer group has weekly or monthly consumption habits, whereas the other clusters are associated with daily consumption tendency (see Table 6).
Finally, behavior patterns are also different in relation to the place of purchase (see Table 6).Within the entire sample, 48.14% of respondents indicate that ordinary supermarkets are their preferred supplier of dairy products, followed by online stores (23.33%), exclusive stores (11.82%), high-end import supermarkets (8.47%), and convenience stores (6.24%).The share of ordinary supermarket channels (such as RT-MART and Walmart) is higher among the carefree dairy consumer and confused dairy consumer groups.Second, preference for online shopping is higher among the carefree dairy consumer group, which could be attributed to lower perceived risk in relation to both risk importance and risk probability, and that this group may also not be afraid of the higher risk attached to online shopping.Finally, the face-concerned dairy lover and cautious dairy lover groups indicate a higher preference for the high-end import supermarket and exclusive stores when purchasing dairy products.

DISCUSSION
This paper aimed to explore the effect of consumer involvement on dairy product purchase behavior to demonstrate the value of using consumer involvement as a segmentation tool for the identification of different consumer groups in relation to the following dimensions of behavioral consequences: extensiveness of decision making, trust of information sources, cue utilization, and consumption habits.First, involvement with dairy products is measured by 4 subdimensional factors: pleasure value, symbolic value, risk importance, and risk probability, which are similar to other product categories (Verbeke and Vackier, 2004).Pleasure value is the dominant factor in dairy consumption, which reveals that emotional value and dietary health considerations are the main motivations driving dairy consumption (Robinson and Getz, 2016;Lee et al., 2019), even in uncertain or risky circumstances.Despite the importance of pleasure value, this factor does not fully explain why consumers become involved in dairy consumption.Risk importance and probabil- Scores within a row with a different superscript are significantly different at P < 0.05 (one-way ANOVA and post hoc Tukey multiple comparison test).
1 Face-concerned dairy lover: consumers who enjoy consuming or eating dairy products and do not perceive many risks in their purchasing process; carefree dairy consumer: consumers with a low perceived risk of dairy purchases; cautious dairy lover: consumers who feel happy when purchasing and eating dairy products; confused dairy consumer: consumers who are concerned about making purchase mistakes and view purchasing dairy as a complicated and confusing task. 1 = "fully disagree" to 7 = "fully agree."ity are also essential in determining the respondents' level of consumer involvement.This is the result of the frequent dairy scandals that have occurred in China, making domestic consumers concerned about risks associated with dairy consumption.Furthermore, unlike other food involvement research finding a weak role of symbolic value in influencing consumers' degree of involvement (Verbeke and Vackier, 2004), this paper demonstrates that symbolic value is an important factor in Chinese consumers' involvement with dairy products.This can be explained by the fact that dairy products are increasingly becoming a lifestyle choice and are a high-end trend in China (Sui, 2016;Ouyang et al., 2021).This means that face-consciousness Chinese consumers feel dairy consumption can affect their self-image and reflect their social status.
The study identified 4 consumer clusters according to individuals' dairy purchase involvement profiles.Each group has distinct characteristics in relation to sociodemographic variables and behavioral consequences related to dairy consumption.Most of existing studies that use consumer involvement as a segmentation tool focus on the ego-importance and interest aspect of involvement and neglect other subdimensions such as risk importance and symbolic value (Nassivera et al., 2014;Koksal, 2021;Santos et al., 2020).Such studies consider involvement as a one-dimensional construct and usually divide the sample into 3 subclusters of high, medium, and low involvement according to their score of average involvement.The present study furthers such research by identifying clusters according to their performance and characteristics on different subdimensional variables of involvement.This allows exploration of the relative heterogeneity and capture of the subtle nuances of the high level of consumer involvement with dairy purchases that may be driven by pleasure value, symbolic value, risk importance, or risk probability.
The first consumer segment, identified as the faceconcerned dairy lover, is motivated in their dairy decisions by the pleasure value they derive from dairy and believe this consumption improves their self-identity.This group includes more middle-aged men, most of whom are daily consumers with higher income levels.The members of this group also have a higher level of overseas experience than the other groups.This group of consumers clearly has the highest involvement with dairy products of all the consumer groups identified in this study.This finding aligns with previous studies that demonstrate high-involvement consumer groups are linked to higher income and requirements for high quality (Nassivera et al., 2014;Santos et al., 2020).In addition, the findings provide further evidence that Chinese men are eager to enhance their status through consumption behaviors (Mo, 2020).Motivated by high-er involvement, this group of consumers is interested in both social media information and family's or friends' suggestions when seeking high-quality dairy products.This group spends much time comparing alternatives, consulting opinions with friends, and actively searching for information.Furthermore, group members use taste, nutritional value, certification, and different labeling cues to make more reasonable decisions.
The second segment, identified as the carefree dairy consumer, has relatively few concerns and low perceived risk related to dairy consumption and safety.Still, high pleasure value indicates that this group's members enjoy eating dairy.Compared with the other segments, there are relatively more women in this group, and the members do not think that dairy consumption affects their self-image or conveys personal information.The melamine scandal did not greatly influence their purchase frequency, and they report the lowest intention to decrease consumption in the future.This is surprising given previous research findings that women are more risk-averse, as they expect greater outcome value (Eckel and Grossman, 2008).In addition, this group of consumers is not highly motivated or convinced by social media information or experts' recommendations.However, suggestions from relatives and friends can affect their dairy purchase decisions.According to their evaluation system, better taste and high nutritional content are essential signals for high-quality dairy products.Moreover, they enjoy purchasing dairy online, which may be explained by their low involvement, driving them to prefer to buy dairy products through more convenient channels.
The third segment, identified as cautious dairy lover, has a high level of risk importance and significant pleasure value in relation to dairy products.Interestingly, this group contains more young people (younger than 25 yr of age) without children, so their aversion to risk is not based on the consideration of children.In addition, the members of this group trust information from experts but have a neutral attitude to sales suggestions.In the quality evaluation process, labeling and government certification are not important cues for the members of this group; instead, they usually use taste and nutrient (protein) content to justify a purchase choice.Compared with the other groups, this group is the least likely to buy dairy products online and the most likely to buy dairy products in exclusive stores.This can be explained by the fact that it is difficult to identify product quality and risk is higher in online shopping, and the quality of dairy stores is more reliable.
The fourth segment, identified as confused dairy consumer, attaches the lowest pleasure value and interest in dairy products and feels less confident about their decision to purchase dairy products.Most of the Yin et al.: DAIRY PURCHASE BEHAVIORS members of this group consume dairy products weekly, and they have the lowest involvement level.Relatively more low-income consumers from rural areas are in this cluster, and this group has less knowledge about dairy products.Previous segmentation studies demonstrate that low-involvement consumers are more price sensitive and likely to select less-expensive items.They avoid sophisticated products with complex labeling (Balestrini and Gamble, 2006;Barber et al., 2007;Koksal, 2021).In addition, the members of this group are sensitive to the melamine scandals, have the highest probability of reducing their dairy consumption in reaction to the crisis, did not fully recover their confidence in dairy products after the scandals, and have a relatively higher intention to decrease their dairy purchase in the future.An ordinary supermarket rather than a high-end import supermarket is the preferred location to purchase dairy products.
The findings of this study make several important theoretical contributions.First, most research on the Chinese dairy market has been obtained using data on consumer preferences for product attributes (Wibowo et al., 2019), safety cues (Mirosa et al., 2020), trust in traceability systems (Li et al., 2021), and confidence in future consumption (Li et al., 2019), with little known about the nature of consumers, their motivations, or segmentation.Many researchers argue that consumers should not be treated as a homogeneous group (Charters and Pettigrew, 2006;Santos et al., 2020).This is particularly true in contexts where the domestic market is challenging and complex, as is the Chinese dairy market.Frequent food scandals have heavily undermined consumer trust in dairy products and enhanced their perceived risk.Furthermore, with the development of the dairy industry in China, consumers' purchasing habits have changed from a single product (i.e., liquid milk) to a range of dairy products, such as cheese, butter, and cream (Yang and Cheng, 2021).Thus, understanding the behavior of different dairy consumer segments helps dairy enterprises to grasp the market challenges, boost market growth, and attract certain types of dairy consumers through refining dairy products offered in the Chinese market.
Second, compared with traditional forms of segmentation based on sociodemographic variables, psychographic variables, such as consumer involvement, have great advantages in capturing insightful information about consumer clusters that facilitate understanding of consumer behaviors.One possible weakness of using a psychographic variable approach in consumer research is that it might be difficult for researchers to identify and access these different clusters (Charters and Ali-Knight, 2002;Molina et al., 2015).However, our study indicates that different consumer segments can be accessed according to purchase locations, for example, the face-concerned dairy lover shops at highend import supermarkets, the carefree dairy consumer shops online, the cautious dairy lover shops in exclusive stores, and the confused dairy consumer shops in ordinary supermarkets.Thus, by using this psychological variable, this research helps to identify how and where different consumer groups can be targeted.
The findings of this research also have important implications for marketers and managers in the dairy industry.That is, based on the characteristics of the 4 consumer segments, different marketing strategies could be targeted to different consumer groups.That is, if a dairy enterprise is self-positioned as high-end, then it should target the face-concerned dairy lover consumer.The provision of premium services or member cards that provide benefits and customized priorities, such as exclusive customer service and more human after-sales service, would strengthen the correlation between customers and enterprises.Following Nassivera et al. (2014), who suggest organizing tailor-made wine tourism to attract highly involved consumers, dairy enterprises could offer on-site tasting events for the promotion of the new products.In addition, offering supplementary services with higher charges might be an effective strategy for the face-concerned dairy lover segment.The carefree dairy consumer group has a high proportion of women, and this group prefers to shop online.Therefore, online commodity webpages on Taobao or Jingdong could embody elements that attract female consumers.Furthermore, more convenient search services are important factors in this cluster.In addition, the main task for the cautious dairy lover is to mitigate the perceived risk.As a significant proportion of young consumers belong to this group, dairy firms could offer some free on-site visits or educational seminars to introduce advanced production technology and production standards that are in line with international standards to eliminate the concerns of this group about the safety of dairy products.Finally, some valuable strategies from previous research studies (Koksal, 2021) could be used for the confused dairy consumer group, which has the lowest level of involvement.For example, dairy companies might provide basic and unsophisticated products and sell these products mainly in ordinary supermarkets to target this consumer group.In addition, marketers need to target their opinion leaders and organize group purchase activities that offer a relatively lower price.
The findings also provide useful insight for policymakers.After the melamine scandals in the Chinese dairy industry, efforts have been made to reduce consumers' risk perception and regain confidence, including regulating manufacturing processes, providing third-party certification, and improving retailer reputations (Li et al., 2019(Li et al., , 2021)).These strategies are particularly associated with the face-concerned dairy lover and cautious dairy lover groups, which represent almost half of the dairy market sample in this study.In addition, traditional marketing strategies, such as using celebrity and salesperson recommendations, may not work very well because online media is an increasingly important source of information in consumers' daily life.The importance of information from food bloggers and social media platforms such as Weibo and Xiaohongshu was identified by the face-concerned dairy lover and cautious dairy lover groups.Bruwer et al. (2014) argue that consumers who are inclined to seek more food information tend to consider social media as part of a wider channel and use cognitive efforts to justify the reliability of the messages sent by online media.The policy makers could deliver the relevant dairy information on the social platforms which could be noticed by highly involved consumers.
Figure 2. Schematic representation of the statistical method.
Figure 3. Two-stage cluster analysis result.
Yin et al.: DAIRY PURCHASE BEHAVIORS Yin et al.: DAIRY PURCHASE BEHAVIORS

Table 2 .
Yin et al.: DAIRY PURCHASE BEHAVIORS Factor analysis of the initial 13 involvement items (factor loadings from principal component analysis)

Table 3 .
Involvement-based dairy market segmentation (average scores on 7-point scales) 1 Scores within a row with a different superscript are significantly different at P < 0.05 (one-way ANOVA and post hoc Tukey multiple comparison test).1

Table 4 .
Effects of involvement on the extensiveness of decision making, trust of information sources, and cue utilization (average scores on 7-point scales) 1

Table 5 .
Yin et al.: DAIRY PURCHASE BEHAVIORS Effects of involvement on consumption habits (average scores on 7-point scales) 1

Table 6 .
Effects of involvement on consumption frequency and place of purchase (% of respondents)1,2