How does recency influence the timing of purchasing smartphones? The moderating role of context

Abstract

Consumers replace their smartphones more frequently, with their increasing requirements for better functions, superior performance, and fancier appearance. The pursuit of replacement may also identify market opportunities and incur business risks. In line with the research streams in product replacement, this study focuses on the effect of recency on consumers’ purchase timing in the context of replacing smartphones. The framework of this research accordingly stems from the rationale of consumer psychology, with concentration on the moderation of consumers’ need for uniqueness, anticipated regret, and education level, besides the focal recency-purchase timing link. The corresponding hypotheses are therefore proposed. To test the hypotheses, we employed convenience sampling to collect data from a sample of smartphone users in China, and finally the valid sample size was 470. We analyzed the data collected in China for hypothesis testing. Based on the analysis of the data, our findings highlight the main contribution of this research, demonstrating that recency has a significant effect on the timing of consumers’ purchasing smartphones and this effect is moderated by consumers’ need for uniqueness and anticipated regret.

Keywords:

• recency
• purchase timing
• need for uniqueness
• education level
• anticipated regret
• moderation

1. Introduction

Technological advancements have changed the world, and smartphones have become an essential part of modern life. China is no exception. China boasted an 89% of smartphone penetration rate in 2018. The number of mobile phone users in China had reached 1.643 billion by the end of 2021, according to the Ministry of Industry and Information Technology of the People’s Republic of China (2022).

The more people rely on their smartphones, the more they increasingly seek value in using smartphones. An upgrade replacement decision refers to a consumer’s second or later purchase of an improved version of a product, implying the notion of product performance improvement (Kim & Srinivasan, 2009). The ongoing development of 5 G networks in China currently further prompts consumers to upgrade their smartphones. In 2018, 66% of Chinese consumers were willing to replace their smartphones within one year, and 32% of them would do so even within half a year.

China is the largest smartphone market in the world, according to a research report on global smartphone shipments (Counterpoint, 2022). Chinese smartphone consumption is currently facing saturation as consumers in China tend to lengthen the smartphone replacement cycle in recent years. The demand for new 5 G smartphones does not look as strong as expected, even though China still leads the world with over 84% 5 G penetration in January 2022. The increasing market saturation furthermore leads to fiercer business competition in China.

For smartphone manufacturers, the enormous opportunity to increase sales is to shorten the replacement cycle (Euromonitor, 2018). Purchases of high-technology durable goods are rarely one-time affairs because such purchases often involve a series of decisions of whether to retain an incumbent model or replace it with a new one that better meets a decision maker’s evolving needs and tastes (Cripps & Meyer, 1994). As for conspicuous durable goods, there is a salient difference between attracting a new consumer and persuading an existing consumer to replace his/her product. Accordingly, consumer smartphone replacement decisions have been the focus of business and a challenge for all smartphone manufacturers.

This research focuses on the relationship between consumers’ recency of smartphone purchases and the timing of replacement, as well as the moderating effects of the need for uniqueness, education level, and anticipated regret on the relationship. The two objectives of the study are stated as follows:

1. To investigate the recency of smartphone purchases on the purchase timing of future smartphone upgrades.

2. To explore the moderating effects of consumers’ need for uniqueness, education level, and anticipated regret on the recency-purchase timing link.

In terms of theoretical contributions, this research expands upon the extant literature by exploring the timing of purchasing durable goods, specifically smartphones, which differs from previous studies that emphasizes fast-moving consumer goods (FMCG). The research adopts a temporal approach, based on the RFM (Recency, Frequency, Monetary) model (Anitha & Patil, 2022; Chou & Chang, 2022; Hughes, 1996) and the consumer purchase cycle (Song et al., 2016), to stress the effect of recency on consumers’ purchase timing, rather than on their purchase intention or likelihood (Bitran & Mondschein, 1996; Bult & Wansbeek, 1995; Chen et al., 2023; Fader et al., 2005; Kim & Srinivasan, 2009; Okada, 2001; Rhee & McIntyre, 2008). By conducting empirical research in this context, this study bridges a gap in our understanding of the time to replace high-tech durables like smartphones.

Furthermore, this research contributes to the mechanism underlying how consumer-related factors moderate the decision-making of their purchase timing. On the basis of the previous studies, it specifically incorporates contextual variables such as personal traits of consumers (e.g., need for uniqueness), demographics (e.g., education level), and consumer psychology (e.g., anticipated regret) into a model for a better comprehension of such moderating effects. Firstly, the need for uniqueness, a vital personality dimension in the field of consumer behavior (Lynn & Harris, 1997; Lynn & Snyder, 2002; Ruvio, 2008; Tian et al., 2001), reflects the human need to maintain a balance between similarity and dissimilarity to other people according to uniqueness theory (Lynn & Snyder, 2002; Snyder & Fromkin, 2012). Consumers desire to see the differences between themselves and their peers or social groups to gai1966n meaningful self-identity (Abosag et al., 2020; Das et al., 2022). Social identity theory further indicates that consumers use brands for self-definition when they can satisfy their self-motivation (Ahmad et al., 2022; Farivar & Wang, 2022; Wolter et al., 2016). It is logical to predict an inextricable relationship between the need for uniqueness and consumer behavior. Secondly, education level has been a salient demographic variable for market segmentation research for a long time (Burton, 2002), influencing consumer mindsets, information acquisition, awareness levels, and ultimately, purchase behavior (Ajzen, 1991; Alba & Hutchinson, 1987; Bettman & Park, 1980; Burton, 2002; Granger & Billson, 1972; Hawkins et al., 1992, Levy, 1966; Mowen & Minor, 1995; Moyer & Hutt, 1978; Rao & Monroe, 1988; Russo et al., 1975; Wibowo et al., 2022). In this research, education level is regarded as a potential moderator variable rather than a control variable, as commonly treated in most marketing studies. Finally, anticipated regret represents a negative emotion that functions as a crucial prohibitive force in consumer decision-making (Anderson, 2003; Fredin, 2008). Unlike other negative emotions (Brewer et al., 2016; Gu et al., 2023), anticipated regret is one of the most influential negative emotions that promote or inhibit behaviors (Saffrey et al., 2008). Given the effect of anticipated regret on the consumer decision-making process (Ang et al., 2022; Baumeister et al., 2011; Brown & Daus, 2016; Jiang et al., 2017; Kim et al., 2022; Saffrey et al., 2008; Wang et al., 2020), consumers are more likely to suffer from anticipated regret for irreversible decisions, such as those related to durable goods like smartphones, compared to ordinary decisions (Zeelenberg, 1999). Consequently, anticipated regret is incorporated in the research as a moderator because it has a more substantial impact on consumption decisions for durable goods than on FMCG.

Smartphones are used as the research object. This research integrates recency and consumer purchase timing, along with consumer traits, consumer cognition, and consumer affect, in a more comprehensive and well-developed model. Specifically, it is mainly to explore the impact of purchase recency on consumers’ smartphone purchase timing, with the moderation of the need for uniqueness, education level, and anticipated regret. This research uses an online survey for data collection and moderated regression analyses for hypothesis testing. The managerial implications can hopefully provide some guidance for the marketing of smartphones.

The paper is organized as follows. In the next section, the relevant literature is reviewed in a logical manner to develop the theoretical model of this research. In the third section, hypotheses are further proposed based on the theoretical model. The fourth section presents the methodology and data collection. The fifth section is concerned with the data analysis and leads to the findings of this research. In the final section, the research addresses the theoretical contributions and management implications. The research limitations and direction for future research are discussed subsequently.

2. Literature review and theoretical development/model

Research framework

The demand for replacing existing products mainly drives new purchases in the consumer durable market (Oh, 2019). As a result, a model is proposed to explore a negative relationship between recency and consumers’ purchase timing, moderated by the need for uniqueness, education level, and anticipated regret. The rationale for such moderating effects is rooted in an interactionist perspective, which asserts that consumer behavior depends on (1) the context, (2) the individual differences, and (3) the interaction between the two (Dickson, 1982; Punj & Stewart, 1983). Recency can be defined as the time elapsed since the last time a customer purchased a product from a firm (Blattberg et al., 2008). In a replacement decision, the recency factor forms uncertainty about the future level of utility that may be derived from a currently owned product and a potentially new product (Cripps & Meyer, 1994). The conceptual framework of this research framework is illustrated in Figure 1.

Figure 1. Research framework

Purchase timing refers to the time it takes for consumers to make their next purchase (Kumar et al., 2008). This construct can be measured on a discrete or a continuous scale by nonnegative integers or real numbers (Kopperschmidt & Stute, 2009). Manufacturers need to understand consumers’ purchase timing to allocate resources more efficiently. Many firms try to earn higher profits by shortening consumers’ purchase timing. Several studies have accordingly investigated the effects of various marketing activities on purchase timing (Bucklin & Gupta, 1992; Bucklin et al., 1998; Gupta, 1988; Helsen & Schmittlein, 1993; Jain & Vilcassim, 1991). Such factors as price sensitivity (Bucklin & Gupta, 1992; Bucklin et al., 1998), product type (Bayus, 1988), and the speed of technological change (Grenadier & Weiss, 1997) may influence consumers’ purchase timing. A variety of models for predicting consumers’ purchase timing emerged in marketing research (Allenby et al., 1998; Boatwright et al., 2003; Bohling et al., 2013; Chintagunta, 1998; Kumar & Luo, 2017; Seetharaman & Chintagunta, 2003; Wheat & Morrison, 1990).

In the marketing literature, satisfaction is revealed as a determinant of repurchase intention, which concerns a repurchase of the same product (Szymanski & Henard, 2001). Therefore, this research also incorporates consumer satisfaction as a control variable into the operating model to avoid other confounding influences on dependent variables.

3. Hypothesis development

3.1. Recency and purchase timing

Recency refers to the interval from the last purchase to the current time (Bult & Wansbeek, 1995). The measurement of recency is usually calculated concerning how recently a customer has purchased from a company (Blattberg et al., 2008). Repurchasing durable products typically includes the length of time the current products can last and the length of time they will function properly (Brucks et al., 2000). According to Kumar et al. (2008), purchase timing refers to the time it takes for consumers to make their next purchase.

Cripps and Meyer (1994) indicated in their research that the length of time since consumers’ previous purchase significantly influences the decision to replace. Similarly, according to the RFM model (Fader et al., 2005), consumers with larger recency are more likely to repurchase. Given the constant and rapid improvement of high-tech products, the continuous consumption of product value in the greater recency context results in a lower remaining mental book value of the current reusable (Okada, 2001). Consumers who use older high-tech products purchased earlier are likely to believe that new and upgraded high-tech products can provide a more excellent promotion in task performance (Chow, 2016). As a result, they are driven by performance expectations to increase the intention to upgrade their smartphones and to purchase a new smartphone as soon as possible, thus shortening the purchase timing of replacement. Conversely, since consumers have to pay for upgrades of high-tech products, consumers who have purchased high-tech products recently may feel guilty about disposing of the recently purchased products and may be less willing to upgrade to improved products in general (Kim & Srinivasan, 2009). The low intention to upgrade in a short period extends the purchase timing.

From the product life cycle perspective, once people purchase costly high-end durable goods, such as household electric appliances and digital products, they may not purchase them again for a relatively long time (Song et al., 2016). However, for consumers, durables have a certain life cycle. We assume the product will be used at some rate, and the need for the product will continue in time (Herniter, 1971). As the duration since the last purchase increases, the time remaining in the current product life cycle will be shorter. It represents that it is closer to the timing point of the next purchase. In other words, consumers’ purchase timing is shorter as well. Hence, hypothesis 1 is proposed as follows:

H1:

Consumers’ recency has a negative relationship with their purchase timing.

3.2. Moderating effect of the need for uniqueness

The findings of extant research suggest that the relationship between recency and purchase likelihood may vary with consumer heterogeneity (Blattberg et al., 2008; Gönül & Shi, 1998; Gönül et al., 2000; Khan et al., 2009). Consumers’ need for uniqueness is one of the personal traits that reflects a manifestation of consumers’ individual differences (Snyder & Fromkin, 1977). The rationale for the moderation of the need for uniqueness stems from the fact that people purchase unique consumer durables like a smartphone for conspicuous consumption to the public (Snyder, 1992). Consumers are more likely to make corresponding purchases to show their uniqueness. According to the need for uniqueness theory (Snyder & Fromkin, 1977), when individuals feel that their uniqueness is threatened, there will be a need to differentiate themselves from others, and it will compete with other needs. In short, in the real world, individuals want to be able to show where they are different from others or better than others. This kind of demand is called the need for uniqueness. Uniqueness has various kinds of patterns and expressions. Snyder (1992) mentioned in the theory of uniqueness that, under the premise of avoiding severe social punishment for violating social norms, some forms of material expression can satisfy the need of individuals to express their uniqueness. Therefore, some consumption behaviors, such as the purchase of certain products, showing their unique consumption characteristics to others, have become typical manifestations of the individual to show unconventionality under the background of the market economy. For example, individuals can demonstrate their uniqueness by displaying all their belongings (Belk, 1988), by expressing their interpersonal interaction style (Maslach et al., 1985), or by manifesting their knowledge and expertise in a particular field (Holt, 1995). For consumers, Abosag et al. (2020) suggest that consumers convey the message “we are what we have” to others by purchasing products and services that reinforce or reflect their unique self-identities.

In the process of smartphone replacement, fewer consumers choose to upgrade their products in advance, and more consumers take a wait-and-see attitude, waiting for the evaluation and feedback from early users and choosing whether to upgrade or not. Consequently, product upgrade is a relatively unique and minority consumption choice that can meet consumers’ higher need for uniqueness better than most consumers’ choices. Compared to the average consumers, consumers with higher need for uniqueness consider less about purchase cycle of the product and more about whether the product is sufficiently unique. They will likely upgrade their smartphones when the uniqueness meets their needs, even if the recency is short. According to Kashi (2013) and Kumar et al. (2009), consumers’ need for uniqueness affects their purchase intention. Unique products positively influence the attitudes and behaviors of these consumers in the consumption process (Zhu et al., 2019). As a result, when consumers have a higher need for uniqueness, they are more likely to engage in smartphone upgrade behavior ahead of the average consumers for different phone styles and functions than other consumers, which shortens their next-time purchase timing. As a result, the second hypothesis is proposed as follows:

H2:

Consumers’ need for uniqueness strengthens the negative relationship between recency and purchase timing.

3.3. Moderating effect of education level

The education level of consumers reflects their relevant knowledge, which can influence human decisions or actions (Guo & Meng, 2008; McEachern & Warnaby, 2008). It means that consumers with different education levels will make different purchasing decisions due to their differences in awareness, information acquisition, product demand, and upgrade opportunity cost, etc. Education level is one of the most popular demographic variables for market segmentation. It possesses the advantages of observability and low acquisition cost (Smith, 1996). Consequently, we adopt it as a moderator in the model.

According to the cognitive-affect behavior (CAB) model (Breckler, 1984), decisions begin with cognitions (e.g., individual beliefs, thoughts and perceptions and meanings or attitudes towards an issue or object) and lead to behavior (including the intention to act or actual actions) (Babin & Harris, 2010; Hu & Tsai, 2009; Solomon, 2011). Consumers with higher education levels usually have more exposure to various facts and figures that trigger awareness levels. Therefore, the higher-educated group has shown a relatively higher mean of awareness than the lower-educated group (Ishak & Zabil, 2012). The ability of education provides consumers with enough information to be effective in the marketplace (Seitz, 1972). Education also allows consumers to learn how to evaluate their needs and resources to find the most suitable product when they are at a disadvantage in the marketplace (Langrehr & Mason, 1977). As one of the essential assets of consumers, the upgrade decision of smartphones can reflect consumers’ professional knowledge and expertise in electronic technology to a certain extent. Consumers with higher education levels will be equipped with more knowledge of high-tech durable goods in terms of new technology and new functions. Such knowledge will help them to take greater and more efficient advantage of technological changes in product upgrades, making it easier for them to realize the benefits brought by product upgrading.

However, according to previous research, consumers with insufficient knowledge and cognitive ability are more susceptible to simple cue effects, which include aversion to loss appeals (Bettman & Park, 1980; Park & Lessig, 1981). When consumers choose to replace their smartphones earlier, it implies a loss of functionality and value of the old smartphone, as well as the purchase of an upgrade product at a relatively higher price. These are a kind of cost of replacement for consumers, in other words, a replacement-generated loss, and this loss is what consumers with low knowledge level want to avoid. As a result, the following hypothesis is proposed in the research:

H3:

The negative relationship between recency and purchase timing is strengthened by consumers’ education level.

3.4. Moderating effect of anticipated regret

In consumer purchase decision-making, regret will not only occur when purchase is realized but may also appear before purchase. Researchers in psychology have found that when people face counterfactual comparisons, they learn to anticipate their future regrets and minimize them in various ways (Mellers et al., 1999; Zeelenberg & Beattie, 1997). This kind of regret of pre-purchase illusion is called anticipated regret, as opposed to experienced regret, which is felt through actual comparison after purchase. Janis and Mann (1977) used anticipated regret as an allusion to the main psychological effect of various worries about loss, which is the materialized loss that troubles decision-makers before their decisions. The psychology of anticipated regret occurs prior to making a choice (Somasundaram & Diecidue, 2017; Wong & Kwong, 2007). Since negative affect may act as a prohibitive force in consumer purchases (Anderson, 2003; Fredin, 2008), anticipated regret plays an essential role in the decision-making process, given the fact that individuals may seek to avoid choosing an option that would induce the experience of negative feelings (Brown & Daus, 2016). Unlike other negative emotions, anticipated regret strongly influences consumers’ decision-making (Brewer et al., 2016; Saffrey et al., 2008). The impact of anticipated regret has also been further discussed in the context of consumer purchase decisions (Kim et al., 2022; Wang et al., 2020).

Consumers usually have different purchase cycles for different products. With a certain level of recency, the advancement of purchase timing means that consumer behavior will appear earlier. For the purchase of smartphones, this represents that they will be earlier to experience the upgrade products, which brings the higher switching cost and usage risk to some extent. Consumers with a higher level of anticipated regret usually foresee their possible regret psychology more easily before consumption. While consumers are averse to regret, it is found that anticipated regret leads to increased efforts in consumers’ behavioral intentions (Novliadi et al., 2018). Anticipated regret is useful for increasing consumers’ understanding of the motivation to take a recommended step or make a careful decision (Ogbanufe & Pavur, 2022). They will take specific measures to cope with such regret to avoid this negative psychology (Khan et al., 2019). Bjälkebring et al. (2016) indicated in their research that ways to avoid future regret include avoiding feedback on the consequences of not purchasing a product, consciously anticipating regret, and delaying decision-making. Given a certain influence level of recency on purchase timing, the higher level of consumers’ anticipated regret, the more likely consumers are to assess the disadvantages of early purchase timing before consumption, and the more likely they are to delay their purchase timing. Therefore, the fourth hypothesis is proposed as follows:

H4:

The relationship between consumers’ recency and purchase timing is weakened by anticipated regret.

4. Research methods and data

4.1. Sample and data collection

This research takes smartphones as the research object. Online survey has the characteristics of low cost, high anonymity, high convenience, rapid diffusion, and high response speed (Wright, 2005). Considering the limitations of this research, such as time, human resources, financial resources, and some other aspects, the researchers chose the online survey as the survey methodology to test the proposed model. With 902 million visitors in 2017, WeChat has become one of the largest mobile traffic platforms and even the most prominent mobile Internet ecosystem in China. WeChat platform has a high coverage and utilization rate among Chinese consumers as a communication APP, so WeChat users also fully meet the conditions of owning smartphones. In consequence, researchers handed out a questionnaire on the WeChat platform. Participants were asked to complete questionnaires that included the measurements of all five constructs, as well as some smartphone usage behaviors and experiences. This questionnaire was first distributed on 1 March 2018. And then, 503 samples were collected on 6 March 2018, and 470 valid samples were used for subsequent data analysis.

Among the 470 available samples, 58.1% of the respondents are single, followed by married with children (34.5%), married without children (5.3%), divorced (1.3%), and widowed (0.9%). Regarding age, only 3.2% of the samples are under 20 years old, 56.2% are 21–30 years old, 9.8% are between 31 and 40, 18.7% are from 41 to 50, and 12.1% are above 50 years old. As for the educational background, 1.1% of participants have a junior high school education or below, 7.9% have a senior high school or technical secondary school education, 12.1% have a junior college degree, 37.9% (the most) of them have a bachelor degree, 36.4% have a master degree and even 4.7% have doctor degree. Moreover, in terms of occupations, office workers account for the largest proportion of the samples (47%); Students make up 38.9% of the overall participants; And the rest of the participants are self-employed entrepreneurs, freelance work, retirees, and others, which make up only a tiny proportion. The mode of the participants’ monthly income is below 8,000 yuan (about 1,165 USD), accounting for 81.4 percent of the samples, with 27 percent earning less than 2,000 yuan (about 291 USD). 32.1% of the samples earn between 2,000 yuan and 4,999 yuan (about 727 USD) each month, while 22.3% have a monthly income between 5,000 yuan and 7,999 yuan. The frequency distribution of specific demographic characteristics of the respondents is shown in Table 1.

Table 1. The frequency distribution table of the samples

		Frequency	Percentage
Marital status	Single	273	58.1%
	Married without children	25	5.3%
	Married with children	162	34.5%
	Divorced	6	1.3%
	Widowed	4	0.9%
Age	≤20	15	3.2%
	21–30	264	56.2%
	31–40	46	9.8%
	41–50	88	18.7%
	≥51	57	12.1%
Highest education level	Junior high school or below	5	1.1%
	High school/Technical secondary school	37	7.9%
	Junior college	57	12.1%
	Bachelor	178	37.9%
	Master	171	36.4%
	Doctor degree or above	22	4.7%
Current occupation	Students	183	38.9%
	Office workers	221	47.0%
	Self-employed entrepreneurs	10	2.1%
	Freelance work	26	5.5%
	Retirees	19	4.0%
	Others	11	2.3%
Monthly income	≤2000 RMB	127	27.0%
	2000–4999 RMB	151	32.1%
	5000–7999 RMB	105	22.3%
	8000–9999 RMB	36	7.7%
	10000–14999 RMB	30	6.4%
	15000–19999 RMB	5	1.1%
	20000–29999 RMB	8	1.7%
	30000–49999 RMB	3	0.6%
	≥50000 RMB	5	1.1%

According to the survey results, the most used brand by participants is Apple (48.5%), followed by Huawei (24%) and Xiaomi (10.9%). As for the operation systems, Android and iOS each account for half of the market. Android accounts for 50.4% and iOS for 48.1%, respectively. Regarding replacing smartphones, 96% of participants have the experience of replacing their smartphones, and the main replacement reason is “The old smartphone was broken,” accounting for about 44.3% of the samples. In addition, 21.3% of the participants say that they replaced the old version due to “The new smartphone has better specifications/performance,” followed by “The new smartphone has better/more functions,” which accounts for 15.7%. Finally, the samples with an average inter-purchase interval of 18–24 months represent the maximal proportion (33.2%), followed by 30–36 months (31.3%). More than 83.4% of the participants choose to replace their smartphones within an average of 18–48 months.

4.2. Measurement

We follow Churchill’s (1979) approach to measure all the constructs in the research hypotheses. The recency and purchase timing constructs use a ratio scale to collect responses, while the construct of education level uses a nominal scale. In addition, the constructs of the need for uniqueness and anticipated regret are answered on a five-point Likert scale. Five measures of interest are included in the final instrument. Among them, the measure for recency of purchase is adopted from Kumar and Shah (2009), while the measure of purchase timing is referred to Farías’s (2019) purchase recency description of measures. As for moderators, the need for uniqueness is measured by a 28-item scale based on the research of Knight and Kim (2007) and Tian and McKenzie (2001). A 16-item scale captures anticipated regret by adopting the study of Janis and Mann (1977). Whereas, for the measure of education level, we design a question by referring to the research of Craig and McCann (1978) and combining it with the general classification criteria of the current education level in China. In this research, satisfaction with the current brand is the control variable related to the consumers’ purchase and use of smartphones. For specific variable measurements, see appendix and Table A1.

5. Analysis and results

Factor analysis examines various dimensions of the need for uniqueness and anticipated regret. In line with previous research such as Knight and Kim (2007) and Tian and McKenzie (2001), the need for uniqueness is divided into three dimensions in factor analysis: creative choice counter-conformity, unpopular choice counter-conformity, and avoidance of similarity. As for anticipated regret, previous research was unable to distinguish its multi-dimensionality. A factor analysis of anticipated regret divided it into two dimensions: rational anticipated regret and perceptual anticipated regret. The original data is represented with fewer constructs, thus reducing the difficulty of data processing and interpretation in the subsequent. The factors extracted after factor analysis are used in the research to confirm the appropriateness of these two constructs for a specific measurement problem and to calculate Cronbach’s Alpha value of all constructs. Subsequently, the reliabilities of all construct scales are judged, and the consistency of the internal reliability of the questionnaire is tested.

After a series of confirmatory factor analyses (CFAs), reliability, convergent validity, and discriminant validity are established. The results of the specific reliability and validity tests are shown in Table 2. The CFAs use LISREL 8.80 to purify the reflective scales. According to Bagozzi et al. (1991), discriminant validity is assessed to unity for χ2 difference tests between the constrained and unconstrained models by constraining the estimated correlation between pairs of constructs. Then common method bias attributable to the measurement method is tested by using Harman’s single-factor test (Tehseen et al., 2017). The results of Harman’s single factor analysis show that a single factor can explain 36.8% of the variance (<50%), and there are seven factors with eigenvalues greater than one, so the data do not suffer from common method variance. All the multi-item measures have great Cronbach’s coefficient Alphas which are higher than .70, meaning that the measurements in this research have got sufficient reliability. Specifically, the measure of need for uniqueness has Cronbach’s alpha values of .957, .851 and .954 for the three sub-dimensions. The alpha values are .950 and .940 for the two traits of anticipated regret. Table 3 provides Pearson’s correlation matrix for all variables.

Table 2. Reliability and validity results

Construct		Item number	Cronbach’s alpha	CR	AVE	Square root of AVE
Need for uniqueness	Creative choice counter-conformity	11	0.957	0.987	0.729	0.854
	Unpopular choice counter-conformity	7	0.851
	Avoidance of similarity	10	0.954
Anticipated regret	Rational anticipated regret	10	0.950	0.981	0.767	0.876
	Perceptual anticipated regret	6	0.940

Table 3. Pearson’s correlation matrix

Variable	Coefficients of Pearson correlations
	Recency	Need for uniqueness	Education level	Anticipated regret	Purchase timing
Recency	1.000
Need for uniqueness	.012	1.000
Education level	−.004	−.124**	1.000
Anticipated regret	.057	.482**	−.148**	1.000
Purchase timing	−.172**	−.032	.001	.018	1.000

*p < .1; **p < .05; ***p < .01

This research uses Sharma et al. (1981) moderated regression analysis to test the above hypotheses. The model consists of the main effect of one control variable, one dependent variable, three moderating variables, and three relevant independent variable × moderating variable interactions. The control variable in this regression analysis is satisfaction with current brands. The corresponding questionnaire responses in the equally weighted scales are used to calculate the scores for all constructs. The independent variable and moderating variables, recency, need for uniqueness, education level and anticipated regret, have been mean-centered before establishing the interaction terms, so this research does not have a serious multicollinearity problem.

The results of the moderated regression model are shown in Table 4 below. In Model 1, the control variable accounts for 1.4% of the variance in the current brand satisfaction (F  = 6.420, p  < .05). Model 2 means the variance of the current brand satisfaction is 4.5%, with the F-value of 4.330 (p < .05). Between Model 1 and Model 2, Δ F equals 3.769 and p  < .01, with a significant difference in R², which means that the main effect does make sense. The effect of recency on purchase timing is negative and significant (β = −.165, p < .01), so Hypothesis 1 is supported.

Table 4. Results of moderated regression analysis

Dependent variable	Purchase timing
	Model 1	Model 2	Model 3
Control variable
Satisfaction with current brand (SCB)	.116**	.111**	.107**
Independent variable (Main effect)
Recency (R)		−.165***	−.151***
Moderator variables (Main effects)
Need of uniqueness (NU)		−.067	−.075
Education level (EL)		−.004	−.004
Anticipated regret (AR)		.073	.074
Moderating effects
R × NU			.098*
R × EL			−.062
R × AR			−.087*
R^2	.014	.045	.057
Δ R^2		.031	.013
F-value	6.420**	4.330***	3.498***
Δ F		3.769***	2.063

*Significant at p < .10; **Significant at p < .05; ***Significant at p < .01

The research uses an incremental F-test to validate the moderating effects in Hypotheses 2, 3, and 4 (Aiken et al., 1991). So Model 3 is compared with Model 2 here. Between Model 2 and Model 3, the Δ F-value is 2.063, with p > .10.According to the test of individual interaction terms, two of the three interaction terms are statistically significant. In the H2 mentioned above, we predict that the need for uniqueness strengthens the negative relationship between recency and purchase timing. The hypothesis is confirmed because the regression coefficient of the interaction term in H2 is significantly positive (β = .098, p < .10). As for H3, it predicts that consumers’ education level positively moderates the negative relationship between recency and purchase timing. The corresponding interaction term, recency × education level, is negative but not significant (β = −.062, p > .10), so H3 is rejected. Consistent with the expectation of H4, the negative relationship between recency and purchase timing decreases as anticipated regret increases. The research findings support it because the regression coefficient of the interaction term (recency × anticipated regret) is negative and significant, with β = −.087, p < .10. The specific interaction plots are presented in Figure 2.

Figure 2. Interaction plot of moderators

6. Discussion

6.1. Conclusion

The research findings confirm that recency can play an essential role in the changes in consumers’ purchase timing. The critical reasons for the significance of this correlation can be considered in terms of both the product’s value and the consumer’s purchase cycle. Greater recency means that the value difference between the old and new product increases (Okada, 2001) and also means that consumers can obtain a more substantial product upgrade by purchasing sooner, thus contributing to a shorter purchase timing. In addition, greater recency also implies that the product is getting closer to the end of its life. Consumers will have increased anxiety about the inconvenience of broken smartphones, thus contributing to shorter purchase timing.

The interaction between uniqueness needs and recency is significant, mainly because the faster one completes the upgrade of a mobile phone, the less likely one is to use the same style and function of the old phone with others, which can satisfy consumers’ uniqueness needs (Zhu et al., 2019). It contributes to a shorter purchase timing. In addition, Brown and Daus (2016) also emphasize that anticipated regret plays an essential role in consumers’ decision-making process. The psychology of anticipated regret has prohibitive forces on consumer behavior, leading consumers to postpone decisions to avoid future regret (Bjälkebring et al., 2016). The empirical results support, to some extent, the effect of the interaction terms of these variables with recency on purchase timing. The interaction between anticipated regret and recency negatively influences consumers’ purchase timing, while the interaction between recency and the need for uniqueness affects purchase timing positively.

The findings of this research confirm that there is no significant moderating effect of consumers’ education level on the relationship between recency and consumers’ purchase. It deviates from our hypothesis. We consider the possible reason is that, according to the survey results, more than 79% of the respondents have a bachelor’s degree or higher in terms of educational background. Since the respondents are concentrated in high education levels, this is very lacking in differentiation in the regression analysis, which is an important reason for the non-significant results.

6.2. Theoretical implications

In today’s technological environment, consumers must decide of whether to retain an incumbent model or replace it with a new one that better meets their evolving needs and tastes. This research contributes to the burgeoning literature on the relationship between recency and consumers’ likelihood of purchasing high-technology durables in upgrade decision contexts and investigates the moderating role played by need for uniqueness, education level, and anticipated regret in the relationship between recency and consumers’ purchase timing.

The research contributes to the marketing literature in several main respects. Firstly, this research studies the recency issue by from the traditional FMCG domain to the high-tech durable goods context. In contrast to utilitarian tradeoff approaches, the research mainly takes a temporal approach to how consumers make repurchase decisions over a while based on the RFM model (Hughes, 1996) and the consumer purchase cycle (Song et al., 2016). Accordingly, this research emphasizes the effect of recency on consumers’ purchase timing instead of that on purchase intention or purchase likelihood in the traditional literature (Bitran & Mondschein, 1996; Bult & Wansbeek, 1995; Fader et al., 2005; Kim & Srinivasan, 2009; Okada, 2001; Rhee & McIntyre, 2008). Such empirical exploration rarely exists in previous research.

Secondly, our findings are consistent with existing findings by showing how the impact of recency on purchase timing varies with different contextual conditions shaped by consumer characteristics, namely, the need for uniqueness and anticipated regret. The findings imply that consumers’ need for uniqueness represents consumers’ cognitive state that shapes a contingency in which the effect of recency on purchase timing is strengthened. In contrast, as an affective construct, anticipated regret negatively moderates the recency—purchase timing relationship. Rather than complement each other, recency and anticipated regret create negative interaction. The combination of the need for uniqueness and anticipated regret can, to a certain extent, reflect the efforts to build a comprehensive and integrative model.

6.3. Managerial implications

Our research, concentrating on upgrading decisions from temporal perspective, is supplemented by affective and cognitive perspectives. The research findings may offer several guidelines for managing and launching consumer high-technology products, mainly shed some light on several managerial issues: innovation management for matching the pace of technological advancement; competitive actions for preemptive advantage; new and old product management for cannibalization prevention; optimal timing and marketing communications for high-tech product launch and so on.

From the marketers’ perspective, understanding consumers’ upgrading behavior is essential to product planning. Product managers of high-tech companies would like to know what fraction of consumers would upgrade to the new product and improved versions at what time. In business practices, durable goods firms must keep launching new superior products, in the form of technological improvements and style changes, faster and more frequently than their competitors do to maintain market share. On the basis of our findings, marketers may help consumers manage the transition between generations of products by offering a migration path to some extent (Mohr, 2001). Consumers in technology-intensive markets must make essential decisions about if and when to adopt a new generation of technology (Castaño et al., 2008). In the extreme, they may leapfrog, or pass entirely on purchasing, a current generation of technology in anticipation of a new, better innovation coming down the pike soon.

6.4. Research limitations and future research

With the limited human, material, and financial resources, the research tries to complete step by step in a rigorous way. However, there is still room for improvement. Among them, the main limitation of this research stems from concerns about the representativeness of online samples in this study. There is more than 1.3 billion population in China. Such a substantial population represents an obstacle to random sampling. Because the target population is sufficiently computer-literate and the uses of smartphones and web media are highly overlapped, the use of the Internet-based survey mode is justifiable. However, it is a form of convenience sampling and, as such, the generalizability of the findings may be limited. In this research, questionnaires were distributed through the WeChat platform, from which the samples were inevitably students and young white-collar workers. As a result, demographic variables such as the age and occupation of the respondents could not be controlled entirely. Meanwhile, the proportion of respondents in this research with a bachelor’s degree or higher accounted for more than 70%, which was higher than the average education level in China. Hence this unequal distribution of education levels is one of the limitations worth noticing. Furthermore, consumers are affected by many situational factors when making upgrade decisions for smartphones. Mixed-mode surveys using Internet-based and traditional media are suggested for future studies in this area (Ilieva et al., 2002).

In this research, we mainly focus on the construct of recency that influences consumers’ purchase timing, with moderating effects of three variables, consumers’ need for uniqueness, education level, and anticipated regret. However, an investigation into the moderating effects of situational factors on the recency-purchase timing link is rooted in existing findings that the relationship of recency with purchase likelihood may differ by consumer-related contingencies (Khan et al., 2009). In other words, consumers are influenced by several situational factors when deciding to upgrade smartphones. Other important and relevant factors, such as brand image and price level, are not incorporated into the model. It can also be learned from some academic researches that when consumers make decisions to upgrade smartphones in real life, if the image of the manufacturer is damaged during their use, or the price of the new generation upgraded smart phone of the original brand is high, consumers may reject the product upgrade decision and even choose to switch brands. Therefore, this research is suggested to be extended in the future by including the factors above. Even more other influence factors can be included into the model to discuss. It may lead to more accurate results and is also a logical step to advance our knowledge.

Another area that deserves attention is that the main object of this research is the smartphone, consumer durable goods for daily use. Its excellent market penetration and salient product features make it different from other consumer durables. As such, the research findings might not be directly applied to the upgrade decisions of other high-tech products. It is logical to extend this research to other product category contexts, such as utilitarian and hedonic durable products, to explore consumers’ upgrade decisions and make the research findings more comprehensive and applicable. The logic of building up the models in this research may be applied to developing models depicting product differences. From a theoretical viewpoint, a more general model of recency may be established by integrating the models across product categories. Future research may try to extend the findings to a business marketing context in that the need to manage upgrade options can be particularly sensitive for business customers.

Finally, the respondents in this research are Chinese consumers. People in other countries may well think and behave differently from their Chinese counterparts. As such, it is possible that certain differences may exist in consumer durable replacement behavior, including the timing of purchasing smartphones, among a variety of countries or regions. In future research, we suggest to further explore the differences in smartphone replacement behavior across countries and incorporating the understanding of such differences into the formulation of corresponding marketing strategies. In this regard, one possible approach to the influences of different cultures on consumer behavior is to adopt Hofstede’s cultural dimensions theory (Hofstede, 1993). It advances our understanding of replacing consumer durables across across cultural and geographical contexts by incorporating the national culture constructs such as individualism/collectivism, power distance, and uncertainty avoidance. Deeper and valuable consumer insights can be gained into the mechanisms through which cultural differences shape the decision-making of replacing smartphones. This will lead to a formative model that may guide the formulation and implementation of international marketing strategies for high-technology products like smartphones.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes on contributors

Yuying Zheng

Yuying Zheng is a Lecturer at the Business School of NINGBOTECH UNIVERSITY. She holds a Ph.D. degree and a master degree from the National Chengchi University, and a bachelor degree from Renmin University of China. Her primary research interests include Logo design, high-tech marketing and marketing strategy. She has presented at many conferences such as AMA, EMAC and DSI.

Chien-Wei Chen

Chien-Wei Chen is currently Professor and Department Chair of the Department of International Business, College of Commerce, National Chengchi University. He received his Ph.D. degree from Warwick Business School, University of Warwick, U.K. His research interests cover a variety of areas including new product preannouncement, high-technology marketing, brand management, marketing communications, etc. He has published papers in international journals such as Marketing Letters, Journal of Business Research, Industrial Marketing Management, Australian Journal of Management, the Service Industries Journal, and Journal of Marketing Theory and Practices.

Appendix

This research adopts the method of a survey to conduct research. The questionnaire is divided into four parts. The specific order of the questionnaire, the number of questions in each dimension, the measurement items, references, and measurement scales are shown in the following table.

Table A1. Research questionnaire design

Part	Amounts of items	Construct	Reference	Measurement scale
Part 1	11 items	Smartphone purchase and use experience	N/A	Nominal scale, ratio scale, and five-scale Likert scale.
Part 2	28 items	Need for uniqueness	Knight and Kim (2007); Tian et al. (2001).	Five-scale Likert scale.
Part 3	16 items	Anticipated regret	Janis and Mann (1977).	Five-scale Likert scale.
Part 4	5 items	Demographic characteristics	N/A	Nominal scale and interval scale.

In this research, each construct involved in the survey is carefully considered. Some relevant questionnaires in previous research are referred to and then modified and redesigned according to the measurement requirements of this research. The specific items in the survey are listed below.

Part 1. Please answer the following questions based on your purchase or use experience in the past.

1. Do you have a smartphone now?

A)Yes. B) No (Finish answering the questionnaire. Please fill in your basic information directly.).

2. Which brand of smartphone do you use now?

A)Huawei. B) Oppo. C) Vivo. D) Mi. E) Apple. F) Samsung. G) ZTE.H) Lenovo. I) Gionee. J) Meizu. K) Le. L) Others _____.

3. Which is your current smartphone system?

A)iOS. B) Android. C) Windows. D) Others _____.

4. Have you ever replaced your smartphone?

A)Yes. B) No (Please skip to Question 6).

5. Why did you replace your smartphone last time? (Single choice)

A) The old smartphone was broken. B) You replaced the telecom contract when it expired and passingly chose to purchase a new smartphone. C) The new smartphone has better/more functions. D) The new smartphone has better specifications/performance.E) The new smartphone has a better appearance. F) The new smartphone is cheaper. G) People around you replaced new smartphones. H) It was given to you as a gift.I) You were attracted by advertisement/brand ambassadors/promotions. J) Others _____.

6. At what time did you last purchase your current smartphone?

___ year(s) ___ month(s) ago.

7. How long do you expect to “keep using” your current smartphone?

___ year(s) ___ month(s).

8. How often do you replace your smartphone?

___ year(s) ___ month(s).

9. How satisfied are you with your current smartphone brand?

A) Very unsatisfied. B) Unsatisfied. C) Neutral. D) Satisfied. E) Very satisfied.

10. How about your intention to purchase the newly launched smartphones?

A) Will not purchase. B) May not purchase. C) Neutral. D) May purchase.E) Will purchase.

11. How about your purchase intention for the latest smartphone launched by the “current brand”?

A)Will not purchase. B) May not purchase. C) Neutral. D) May purchase.E) Will purchase.

After collecting and discussing previous relevant pieces of literature, this research mainly refers to the research of Knight and Kim (2007). It is supplemented by the scale structure of Tian et al. (2001). A set of questions about consumers’ degree of need for uniqueness that meet the requirements of this research are modified and designed, including a total of 28 narrative questions. A five-scale Likert scale is used as the measurement index. Option 1 means strongly disagree; Option 2 means disagree; Option 3 means neutral; Option 4 means agree; Option 5 means strongly agree.

Part 2. In this part, please tick the option that fits you according to your personal subjective feeling.

1. I purchase unique products to tell people I am different.

2. I have sometimes purchased unique products or brands to create a more distinctive personal image.

3. I often look for one-of-a-kind products to create a style that is all my own.

4. Often, when buying merchandise, an important goal is to find something that communicates my uniqueness.

5. I often create a personal image for myself that unique products cannot duplicate.

6. I often try to find a more interesting product because I enjoy being original.

7. I actively seek to develop my personal uniqueness by buying unique products or brands.

8. Having an eye for products that are interesting and unusual assists me in establishing a distinctive image.

9. The brands and products that I like best are the ones that express my individuality.

10. I often think of the things I buy and do in terms of how I can use them to shape a more unusual personal image.

11. I am often on the lookout for new products or brands that will add to my personal uniqueness.

12. I have sometimes dared to use the products or brands others are likely to disapprove of.

13. Though it may seem strange to others, I still consider using a different product.

14. I rarely act in agreement with what others think are the right things to buy.

15. Concern for being out of place does not prevent me from using what I want.

16. When it comes to the products I buy and the situations in which I use them, I have often broken customs and rules.

17. I have often violated the understood rules of my social group regarding what to buy or own.

18. I enjoy challenging the prevailing taste of people I know by buying something they would not seem to accept.

19. I am often aware that others think my products are peculiar, but I don’t care.

20. When products or brands I like become extremely popular, I lose interest in them.

21. I avoid products or brands that have already been accepted and purchased by the average consumer.

22. When a product I own becomes popular among the general population, I use it less.

23. I often avoid the general population buying products or brands I know.

24. I am not too fond of products or brands everyone customarily purchases.

25. I stop using them when the products I use become popular with the general public.

26. The more commonplace a product or brand is among the general population, the less interested I am in buying it.

27. Products do not seem to hold much value for me when everyone often purchases them.

28. When a product I own becomes too commonplace, I usually quit using it.

After collecting and discussing previous relevant literature, this research designed a set of questions for consumers’ degree of anticipated regret that meet the requirements of this research. It includes 16 narrative questions, mainly based on Janis and Mann’s (1977) definition of anticipated regret and the main influencing factors of consumers’ anticipated regret psychology. A five-scale Likert scale is used as the measurement index. Option 1 means strongly disagree; Option 2 means disagree; Option 3 means neutral; Option 4 means agree; Option 5 means strongly agree.

Part 3. This part is to understand your views on using or purchasing a smartphone. Please tick the option that fits you according to your personal subjective feeling.

1. Before I bought a smartphone, I was worried that its appearance would become obsolete soon, and I thought I would regret buying it.

2. Before I bought a smartphone, I was worried that its design would become unfashionable soon, and I thought I would regret buying it.

3. Before I bought a smartphone, I was worried that it would fail to reflect my taste soon, and I thought I would regret buying it.

4. Before I bought a smartphone, I was worried that it would not match my status soon, and I thought I would regret buying it.

5. Before I bought a smartphone, I was worried that it would lose its conversational edge soon, and I thought I would regret buying it.

6. Before I bought a smartphone, I was worried that it would not satisfy me soon, and I thought I would regret buying it.

7. Before I bought a smartphone, I was worried that its value would decline soon, and I thought I would regret buying it.

8. Before I bought a smartphone, I was worried that it would become less cost-effective soon, and I thought I would regret buying it.

9. Before I bought a smartphone, I was worried that its market price would drop so quickly that I thought I would regret buying it.

10. Before I bought a smartphone, I was worried that the cost of repairing it would be higher than the cost of buying a new one soon, and I thought I would regret buying it.

11. Before I bought a smartphone, I was worried that its service life was getting shorter and shorter, and I thought I would regret buying it.

12. Before I bought a smartphone, I was worried that its functions would become obsolete soon, and I thought I would regret buying it.

13. Before I bought a smartphone, I was worried that its specifications would become obsolete soon and I thought I would regret buying it.

14. Before I bought a smartphone, I was worried that it would fall behind in performance soon, and I thought I would regret buying it.

15. Before I bought a smartphone, I was worried that it would be easily obsolete in terms of quality, and I thought I would regret buying it.

16. Before I bought a smartphone, I was worried that its operating system would be incompatible soon, and I thought I would regret buying it.

Part 4. Please answer the following questions according to your basic information.

1. Your marital status:

A) Single. B) Married without children. C) Married with children. D) Divorced.E) Widowed.

2. Your age:

A) 20 years old or less than 20 years old. B) 21–30 years old. C) 31–40 years old.D) 41–50 years old. E) 51 years old or over 51 years old.

3. Your highest education background:

A) Junior high school or below. B) High school/Technical secondary school. C) Junior college.D) Bachelor. E) Master. F) Doctor degree or above.

4. Your current occupation is:

A) Students. B) Office workers. C) Self-employed entrepreneurs. D) Freelance work.E) Retirees. F) Others _____.

5. Your monthly income is:

A) Less than 2000 RMB. B) 2000–4999 RMB. C) 5000–7999 RMB. D) 8000–9999 RMB.E) 10000–14999 RMB. F) 15000–19999 RMB. G) 20000–29999 RMB. H) 30000–49999 RMB.I) More than 50,000 RMB.