Uploaded by Scarlett Raincloud

Li et al.

advertisement
Heliyon 9 (2023) e14154
Contents lists available at ScienceDirect
Heliyon
journal homepage: www.cell.com/heliyon
Research article
Total reward satisfaction profiles and work performance: A
person-centered approach
Chunling Li *, Xueyun Duan, Xiangling Chu, Yong Qiu
Business School, Beijing Technology and Business University, Beijing, 100048, China
A R T I C L E I N F O
A B S T R A C T
Keywords:
A person-centered approach
Latent profile analysis (LPA)
Total reward satisfaction (TRS)
Work performance
It has recently become an incentive management challenge for organizations to implement a total
reward system. Existing variable-centered studies have neglected to explore the incentive effect of
a total reward system from the perspective of individual differences. Our study aimed to initially
investigate the profiles of total reward satisfaction (TRS) and the impact of these profiles on work
performance. Using a person-centered approach, two studies were conducted using retail industry
employees in China as samples. Study 1 identified the TRS profiles of 429 samples using latent
profile analysis. Study 2 replicated Study 1’s configuration of profiles and examined the rela­
tionship of these profiles with demographic variables and work performance using 885 samples.
Our results were as follows: (1) there were four quantitatively and qualitatively distinct profiles
(subpopulations) of TRS, namely, dissatisfied (DS), development and career opportunities
satisfied-dominant (DOS-dominant), work-life balance satisfied-dominant (WLS-dominant), and
compensation satisfied-dominant (CS-dominant); (2) demographic variables involving gender,
age, education, and position level affected the likelihood of membership in each TRS profile; and
(3) the four profiles predicted different levels of work performance, or more specifically, different
levels of task and contextual performance. The task and contextual performance of the four
subpopulations listed from best to worst were WLS-dominant, DOS-dominant, CS-dominant, and
DS. For practical management, organizations should customize a classified total reward system
according to employee subpopulations to improve work performance.
1. Introduction
With the current situation that organizations generally face financial difficulties due to the COVID-19 pandemic, improving
employee work performance has been a critical issue in organizational incentive management. Since a single financial reward is no
longer considered the only driver for employees, organizations attempt to improve work performance by providing the right total
reward system as an employee-driven system, especially when they face difficulties in offering higher financial rewards [1,2,3,4]. Total
reward refers to all forms of non-financial and financial returns offered by organizations to their employees [5]. At present, many
organizations across industries worldwide popularly implement a total reward system [3,6,7]. A total reward system is distinctly
characterized by a customized combination of rewards that can satisfy the needs or preferences of individuals [8,9]. Therefore, many
practitioners are increasingly concerned about how total reward satisfaction (TRS) promotes work performance. However, extant
research provides almost no guidance for management practice in this regard.
* Corresponding author.
E-mail address: licl@th.btbu.edu.cn (C. Li).
https://doi.org/10.1016/j.heliyon.2023.e14154
Received 2 June 2022; Received in revised form 21 February 2023; Accepted 23 February 2023
Available online 28 February 2023
2405-8440/© 2023 The Authors.
Published by Elsevier Ltd.
This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Heliyon 9 (2023) e14154
C. Li et al.
Although the existing research is persistently concerned about the relationship between reward satisfaction and work performance,
empirical investigations barely pay attention to TRS as a multidimensional construct [10,11]. Moreover, the extant research pays far
more attention to financial reward satisfaction than to non-financial reward satisfaction [3,7,10,12]. In addition, the conclusions of the
existing studies are controversial. Some studies regard certain financial reward (e.g., monetary) satisfaction as motivating [13,14],
whereas others see certain non-financial reward satisfaction as motivating [3,5,15]. The main reason for the above controversies stems
from the limitations of the variable-centered approach [16], which has resulted in most previous studies focusing on the effect of
certain TRS dimensions on work performance. This approach investigates the "average" and independent effects of one variable on
another by assuming that the population of a sample is homogeneous [16]. However, there are two weaknesses to the
variable-centered approach. One is that a sample’s population does not conform to the homogeneous assumption in reality [17]. The
other is that the use of this approach makes it difficult to interpret the complex interaction effects between variables [18]. Previous
studies have ignored the potential combination of TRS dimensions and failed to detect not only how individuals in multiple sub­
populations of a sample could experience differential satisfaction with various dimensions of total reward but also how the differences
in subpopulations affect performance [5,19]. Therefore, this research introduces a person-centered approach to complement the
shortcomings of a variable-centered approach [16].
Assuming the heterogeneous population of a sample [16,17], a study based on a person-centered approach can determine possible
profiles (subpopulations) consisting of variable dimensions by latent profile analysis (LPA) and explore the relation of profiles to
covariates using mixture models [20,21]. The driving of employee outcomes by total reward is a complex and dynamic mechanism,
and there are individual differences in the effect on individual behaviors of different forms of total reward [22]. Our research explores
whether TRS profiles (subpopulations) differ, how demographic variables predict these profiles, and how these profiles exhibit
different levels of work performance. To sum up, our research presents another way of looking at the same reality [20]; that is, it not
only advances the theoretical research on total reward incentive from the perspective of individual differences but also offers a
classification system for management based on the types of employee subgroups [17,23]. Therefore, our results can serve to guide
organizations in implementing classification management policies for total reward systems to improve work performance.
There are few existing studies on TRS profiles; thus, our study seems unsuitable for raising hypotheses about the configurations of
profiles and their relationship to work performance. Consequently, in short, our study pursues exploratory questions and tests them
with two samples of employees drawn from the retail industry in China. Study 1 examines whether TRS profiles (subpopulations)
differ; Study 2 replicates the configuration of profiles in Study 1 and explores how demographic variables affect these profiles and how
these profiles exhibit different levels of work performance.
2. Theoretical background
2.1. TRS profiles
TRS refers to the degree to which negative or positive attitudes are expressed by employees regarding their received rewards [3,5].
There are several views on the dimensions of TRS measurements [3,7]. Among measurements, the five-dimension measurement is
based on the generally recognized five components of total reward [22], whose reliability and validity were confirmed by Chinese
retail enterprise employees [3]. In summary, the present study adopts the five-dimension measurement.
From a person-centered perspective, various dimensions of TRS may form different configurations, i.e., latent profiles. Profile
indexes can qualitatively and quantitatively classify profiles [21,24,25]. Quantitative distinctions in profiles (level) refer to absolute
differences in profile indexes. In contrast, qualitative distinctions in profiles (shapes) refer to relative differences in profile indexes.
Thus, there may be quantitative differences among TRS profiles, i.e., all five dimensions of satisfaction in one profile are high, while all
five dimensions of satisfaction in another profile are low. In contrast, there may also be qualitative differences among TRS profiles, i.e.,
one or two dimensions of satisfaction are dominant in one profile but not in another. For instance, one profile has high compensation
satisfaction and low satisfaction in other dimensions, whereas another has high benefit satisfaction and low satisfaction in other di­
mensions. There have been only a few studies on TRS applying a person-centered approach to date. Their findings using cluster-wise
regression suggest the presence of different employee subgroups in the relation of TRS to turnover intention and job satisfaction [5,19].
Since cluster-wise regression is less accurate and efficient than LPA for identifying classification [20], these two studies contain
insufficient direct evidence for determining profiles. In conclusion, although we cannot assume the way in which five dimensions of
TRS can be combined in different profiles, we anticipate finding profiles that differ quantitatively and/or qualitatively. Thus, the
research question (RQ) we propose is as follows:
RQ 1.
Are there distinct TRS profiles with regard to quantity and/or quality?
2.2. Demographic variables and TRS profiles
According to need hierarchy theory, the level of people’s particular needs varies [26,27], which means that their need or value for a
particular type of total reward may differ [9]. It has been demonstrated in previous studies that employees of different gender, age,
marital status, children’s age and position levels had significant differences in the value they placed on different types of total rewards.
Men pay more attention to financial rewards, and women with children are more likely to value work-life balance; young employees
(especially non-managers) prefer career development more than older employees do, but older or female breadwinners are more likely
to value benefits [28]. Older employees prefer financial rewards more than younger employees do [29]. Generation Z employees (the
2
Heliyon 9 (2023) e14154
C. Li et al.
millennial generation) draw more attention to ability enhancement and support from the leader than monetary rewards [11,30].
Government employees in safe positions value intrinsic rewards more, while those in unsafe positions give greater value to extrinsic
rewards [31]. Lower-level managers attach more importance to bonuses than middle-level managers do, middle-level managers pay
more attention to promotion opportunities than high-level managers do, and lower-level managers attach more importance to
retirement plans than high-level managers do [32].
The study based on a person-centered approach assumes the heterogeneous population of a sample [17,19], and demographic
variables can be used as covariates to construct mixed regression models that predict similar individuals within each profile [21]. The
above literature shows that individuals with different demographic variables perceive different levels of satisfaction with the various
offered forms of total reward. We aim to assess how demographic variables determine whether similar individuals belong to a
particular profile. Therefore, we propose the following RQ:
RQ 2. How do demographic variables involving gender, age, education, and position level influence membership in each profile of
TRS? Or are similar individuals with specific demographic variables more likely to share a specific profile?
2.3. TRS profiles and work performance
Work performance indicates the observable behaviors demonstrating that employees are conducive to the organization or orga­
nizational goals [33]. Task and contextual performance are widely recognized types of work performance. Task performance can be
defined as those behaviors directly supporting the completion of job tasks and job responsibilities. In contrast, contextual performance
refers to those behaviors supporting a broader social, organizational, and psychological context in which job tasks can be attained [34].
Social exchange theory, which combines perspectives from psychology, management, and sociology, is prevalently used to un­
derstand organizational behaviors [35,36]. Blau [37] held that exchange activities bring rewards that have a dominant influence on
human behavior, and the rewards obtained by interacting with others are called social rewards, including both intrinsic and extrinsic
rewards. Social exchange can be defined as voluntary behavior that is stimulated by the rewards that people expect and receive from
interacting with other people and varies in fundamental ways from economic exchange. Economic exchange involves exact quantities
as defined by a formal contract and are transacted to ensure that both parties perform specific responsibilities [36]. Social exchange, by
contrast, entails unspecified obligations, so there is no contractual guarantee. Because certain job tasks that employees are responsible
for are not all specified in advance, the exchange between employees and their employing organization is somewhat closer to social
exchange [37]. According to these views, while an organization offers a total reward to its employees, employees need to invest their
time, energy, skills, and other input costs to achieve the work performance required by organizations as returns to their organization.
The employment relationship established between organizations and employees through a total reward system can be regarded as an
exchange relationship that includes both social and economic exchange [38,39,40].
Social exchange relationships can be classified into two types, given the causal perspective of forming exchange relationships [35,
41]. One is the positive hedonic value model. When positive initial behaviors benefit a target, positive exchange relationships occur.
Then, targets will tend to respond to positive initiating actions in kind with more positive and/or fewer negative behaviors. The other is
the negative hedonic value model. When negative initial behaviors harm a target, negative exchange relationships occur. Then, targets
tend to respond to negative initiating actions with more negative behaviors [35,42]. Based on the positive hedonic value model, as a
positive initiator, a good exchange relationship arises when an organization offers a total reward that its employees perceive as
satisfactory. In this context, employees are likely to reply with high work performance [43]. Employees tend to not only strive to
improve their task performance but also to undertake more contextual performance, such as teamwork and maintaining the organi­
zational reputation. Conversely, the negative hedonic value model also holds. As a result, different levels of TRS have different effects
on work performance.
Previous research has mainly investigated the separate effect of each TRS dimension on work performance from the perspective of a
variable-centered study. Nevertheless, these research conclusions have been inconsistent to date. A meta-analysis suggested that pay
level satisfaction has a weak positive impact on work performance [44]. Moreover, some studies have suggested that pay satisfaction
and benefit satisfaction significantly enhance task or contextual performance [13,14,45]. Others found that compensation (pay)
satisfaction or benefit (material reward) satisfaction had no significant effect on task or contextual performance [3,5,46,47], and De
Gieter and Hofmans [5] found that financial reward satisfaction significantly negatively impacted task performance. Furthermore,
there is evidence that satisfaction with not only one’s work-family balance [3,48,49] but also performance and recognition (psy­
chological reward) [3,50,51] significantly positively impact work performance or task performance. However, other studies have
suggested that satisfaction with one’s psychological reward, development and career opportunities has no significant effect on one’s
task performance [3,5] or contextual performance [47] and even significantly negatively influences one’s task performance [47].
No studies have yet examined the relation of different TRS profiles to work performance using LPA. A study using cluster-wise
regression found no relationship between different subgroups of TRS and task performance [5]. Consequently, it is difficult for our
study to assume how the profiles of TRS influence work performance. However, the existing conclusions also indicate that the five
dimensions of TRS might have different effects on work performance to a certain extent. Individuals’ experiences of satisfaction with
the five types of total reward may vary, which could generate interactions among the five dimensions of TRS. We aim to discover the
complex mechanism between the interaction of each of these five dimensions and work performance. Therefore, we propose the
following RQs:
RQ 3. Do TRS profiles show (predict) the different levels of work performance?, RQ 3-1: Do TRS profiles show (predict) the different
levels of task performance?, RQ 3-2: Do TRS profiles show (predict) the different levels of contextual performance?
3
Heliyon 9 (2023) e14154
C. Li et al.
Following the work of Wang and Hanges [20], Meyer and Morin [21], and Howard and Hoffman [16], we establish the mixed
model of TRS profiles and covariates (predictors, or outcomes), as shown in Fig. 1, in which demographic variables are predictors of
TRS profiles and work performance is the outcome variable of TRS profiles.
3. Study 1
3.1. Method of study 1
3.1.1. Sample and data collection
In Study 1, with the help of a questionnaire survey company, we randomly sampled employees from several retail formats (e.g.,
department stores, convenience stores, specialty stores, supermarkets, and B2C e-commerce) in the eastern and central cities of China.
The pre-survey including 186 valid questionnaires showed that the reliability of each modified variable scale was acceptable. In LPA,
large sample sizes (>500) are generally considered to be the best fit for mixed models [16], but smaller profiles that lack practical
significance may be identified using substantial sample sizes (>1000) [21]. According to the above requirement of LPA samples, we
conducted the formal questionnaire by conducting online surveys in February 2019 and randomly sampled 511 questionnaires with
the help of the same questionnaire survey company as that which assisted us with the pre-survey. A period of 2 s is the minimum time
needed to answer each item under conservative estimation [52]. Therefore, we considered that a participant who completed our
questionnaire within 1 min had provided poor quality data and excluded them. Finally, there were 429 valid questionnaires, with an
effective rate of 83.95%.
The samples included 45.5% male and 54.5% female. A total of 5.1% were less than 23 years old, 38.5% were 23 to 29 years old,
40.1% were 30 to 39 years old, and 16.3% were 40 years old or above. Regarding education, technical secondary or high school
accounted for 31.0%, junior college accounted for 37.5%, and bachelor’s degree accounted for 31.5%. Regarding position level, 56.2%
were ordinary employees, 35.4% were from low management (e.g., team leaders), and 8.4% were from middle management (e.g.,
department managers/store managers). Beijing Technology and Business University granted ethical approval of our research. Par­
ticipants were assured of confidentiality and anonymity in the informed consent form and in each questionnaire, and we promised to
use all survey data for our study only.
3.1.2. Variable measurement
TRS was measured with a modified scale using five dimensions taken from De Gieter and Hofmans [5], Payne et al. [22], and Li and
Yin [3]. Compensation satisfaction (CS) was measured by a 4-item subscale (e.g., "My current base pay"; α = 0.886). Benefit satisfaction
(BS) was measured by a 4-item subscale (e.g., "Public insurance and housing funds contributed by my company"; α = 0.706). Work-life
balance satisfaction (WBS) was measured by a 4-item subscale (e.g., "My company facilitates employees caring for their families"; α =
0.742). Performance and recognition satisfaction (PRS) was measured by a 3-item subscale (e.g., "My supervisor communicates
regarding my performance"; α = 0.702). Development and career opportunity satisfaction (DOS) was measured by a 4-item subscale (e.
g., "Promotion opportunities offered by my company"; α = 0.728). A Likert 5-point scale was applied to score all items (1 = "very
dissatisfied"; 5 = "very satisfied"). The Cronbach’s α of the total scale was 0.817.
3.1.3. Statistical analysis strategy
Study 1 used LPA in Mplus 7.4 to identify different profiles. LPA is a typical analytical method for person-centered approaches.
Compared with cluster-wise regression, LPA is more conducive to exploring the relationship between variables in each profile [20]. To
detect the optimal profile model, we report six indicators: the Akaike information criterion (AIC), the Bayesian information criterion
(BIC), the sample-size adjusted BIC (ABIC), the Lo-Mendell-Rubin likelihood ratio test (LMR), the bootstrapped likelihood ratio test
(BLRT), and entropy [53,54]. The profile model fits better when the values of AIC, BIC, and ABIC are lower. LMR and BLRT are used to
compare a chi-variance between a k-profile model and a k-1-profile model. It is better to use a k-profile model if the p values of LMR
and BLRT are significant; otherwise, it is better to use a k-1-profile model. The entropy value (0–1) represents a classification error in a
profile model, where a value closer to 1 indicates that the classification error of the model is small. Choosing the right profile structure
requires the consideration of the profiles’ theoretical significance [53], and each profile should have a sample size of at least 5% of the
total number of samples [55].
Fig. 1. The mixture model.
4
Heliyon 9 (2023) e14154
C. Li et al.
3.2. Results of study 1
3.2.1. Descriptive statistics and correlations
Table 1 exhibits descriptive statistics and bivariate correlations of the research variables. The mean for each of the five TRS di­
mensions was just above 3, indicating that employees in the sample were satisfied with the five elements of total reward.
3.2.2. LPA
In response to RQ 1, Study 1 implemented LPA in Mplus 7.4 to identify the different profiles. Table 2 presents multiple fit indicators
to test possible profile models. From the one-profile to five-profile model, the AIC, BIC, and ABIC values gradually declined, which
indicated the model fit’s improvement; each entropy value exceeded 0.80, indicating that the model was accurate. The LMR p value of
the four-profile model was significant (p < 0.001), showing that it significantly outperformed the three-profile model. The LMR p value
of the five-profile model was not significant (p = 0.7055 > 0.05), showing that it did not significantly outperform the four-profile
model, and the model maximum is five profiles. In conclusion, the four-profile model outperformed the others, and was thus cho­
sen as study 1’s model.
Table 3 presents the five-dimensional means of TRS for the four-profile model, and their standardized mean plots are displayed in
Fig. 2. Table 3 and Fig. 2 reflect distinct quantitative and qualitative differences in the four profiles, which can be labeled as follows: (a)
dissatisfied profile (DS), (b) development and career opportunities satisfied-dominant profile (DOS-dominant), (c) work-life balance
satisfied-dominant profile (WLS-dominant), and (d) compensation satisfied-dominant profile (CS-dominant). Employees who belonged
to DS (N = 53; 12.4%) experienced dissatisfaction with all five dimensions, and the mean of each dimension was just above 2. Em­
ployees who belonged to the DOS-dominant profile (N = 131; 30.5%) experienced satisfaction with the available development and
career opportunities (M = 3.82) and neither satisfaction nor dissatisfaction with the remaining four dimensions, of which the means
were just above 3. Employees who belonged to the WLS-dominant profile (N = 117; 27.3%) experienced satisfaction with work-life
balance (M = 4.10) and neither satisfaction nor dissatisfaction with the remaining four dimensions, of which the means were
approximately 3. Employees who belonged to the CS-dominant profile (N = 128; 29.8%) experienced satisfaction with compensation
(M = 4.22) and neither satisfaction nor dissatisfaction with the remaining four dimensions, of which the means were approximately 3.
Thereby, these findings prove the importance of RQ 1, suggesting that four qualitatively and quantitatively differentiated profiles of
TRS exist. In other words, there are four distinct subpopulations of employees in TRS.
4. Study 2
4.1. Method of study 2
4.1.1. Sample and data collection
In Study 2, with the HR Department’s assistance, we invited 1980 employees of 61 stores of a supermarket in Beijing to participate
in a formal online survey in April 2019. According to the requirements of LPA samples [16,21] and based on the number of partici­
pants, we randomly sampled approximately 49% of the total participants in each store to ensure that the total sample included all
stores and that the participants in each store were equally likely to be sampled. We sampled a total of 974 participants and asked them
to fill out the questionnaire. We explained the purpose of our investigation to satisfy their doubts in advance, emailed each participant
the survey link, and restricted each device to one submission. Finally, we excluded 89 participants who completed the questionnaire in
less than 90 s, so there was a final total of 885 valid questionnaires (90.9% of the valid rate).
The final samples included 40.6% males and 59.4% females. A total of 18.9% were less than 23 years old, 25.1% were from 23 to 29
years old, 36.8% were from 30 to 39 years old, and 19.2% were 40 years old or above. Regarding education, technical secondary school
or high school accounted for 47.1%, junior college accounted for 35.2%, and bachelor’s degree accounted for 17.7%. Regarding
position level, 43.1% were ordinary employees, 36.3% were from low management (e.g., team leaders), and 20.6% were from middle
management (e.g., department managers/store managers). Beijing Technology and Business University granted ethical approval for
our research. Participants were assured of confidentiality and anonymity in the informed consent form and in each questionnaire, and
we promised to use all survey data for our study only.
Table 1
Descriptive analysis and correlations among variables of study 1.
CS
BS
WLS
PRS
DOS
MEAN
SD
CS
BS
WLS
PRS
DOS
3.05
2.99
3.10
2.92
2.92
1.03
0.74
0.87
0.81
0.83
1
0.473**
− 0.118*
0.007
− 0.045
1
0.226**
0.271**
0.400**
1
0.596**
0.311**
1
0.471**
1
Note: *p < 0.05, **p < 0.01; SD, standard deviation; CS, compensation satisfaction; BS, benefit satisfaction; WLS, work-life balance satisfaction; PRS,
performance and recognition satisfaction; DOS, development and career opportunities satisfaction.
5
Heliyon 9 (2023) e14154
C. Li et al.
Table 2
Model fit indicators for LPA of study 1.
No. of profiles
AIC
BIC
ABIC
Entropy
p for LMR
p for BLRT
1
2
3
4
5
21915.900
20806.251
19914.616
19196.603
19055.417
22053.989
21017.447
20198.918
19554.011
19485.931
21946.094
20852.430
19976.780
19274.751
19149.550
0.896
0.950
0.952
0.935
<0.001
0.0025
<0.001
0.7055
<0.001
<0.001
<0.001
<0.001
Table 3
The five-dimensional means of TRS for the four latent profiles of study 1.
CS
BS
WLS
PRS
DOS
DS (n = 53; 12.4%)
DOS-dominant (n = 131; 30.5%)
WLS-dominant (n = 117; 27.3%)
CS-dominant (n = 128; 29.8%)
1.42
1.82
1.94
1.99
2.16
3.14
3.44
3.18
3.23
3.82
2.42
2.83
4.10
3.43
2.87
4.22
3.16
2.58
2.51
2.35
Fig. 2. Standardized means of TRS by four profiles of study 1. Note: The mean of a dimension score in a profile is 0, and the standard deviation is 1.
A score closer to 0 shows that the individuals’ dimension satisfaction level in the profile is closer to the overall sample’s mean for that dimension.
4.1.2. Variable measurement
TRS was measured as in Study 1. The Cronbach’s α of the CS, BS, WLS, PRS, and DOS subscales was 0.864, 0.920, 0.849, 0.821, and
0.873, respectively. The Cronbach’s α of the total scale was 0.956.
Work performance was measured with a modified scale of two dimensions taken from Motowidlo and Van Scotter [34] and Yu [56].
Task performance (TP) was measured by a 4-item subscale (e.g., "I am one of the best employees in the entire department"; α = 0.841).
Contextual performance (CP) was measured by a 10-item subscale (e.g., "I can always offer to help my colleagues if they need it"; α =
0.904). A Likert 5-point scale was applied to score all items (1 = "strongly disagree "; 5 = "strongly agree "). The Cronbach’s α of the
total scale is 0.923.
Demographic variables might be related to profiles and work performance. These included gender, age, education, and position level
as antecedents to test the demographic characteristics that predict each profile of TRS.
4.1.3. Statistical analysis strategy
Study 2 implemented LPA and BCH commands in Mplus 7.4 and performed multinomial logistic regression (MLR) analysis in SPSS
23.0. First, Study 2 determined different profiles by applying LPA, as in Study 1.
Second, Study 2 employed MLR to test the profiles’ predictability by demographic variables (predictors). The final model including
all predictors was significant compared to the intercept-only model, indicating that the predictors have explanatory power. The
regression coefficient was adopted to evaluate whether the rise of predictors would lead to an individual’s higher probability of
belonging to a profile other than the reference profile [57]. The odds ratio (OR) was a specific value used to reflect a change in
probability. Age was used as a continuous variable. The other demographic variables are categorical variables, and were coded into
dummy variables such as the following: gender, 1 = female, and 0 = male; education, 1 = less educated (technical secondary school or
high school), and 0 = highly educated (junior college and bachelor’s degree); position level, 1 = ordinary employees, and 0 = low and
middle management (e.g., team leaders, department managers/store managers).
Third, Study 2 applied the BCH command to test the relationship between profiles and outcome variables [58] so that the profiles
6
Heliyon 9 (2023) e14154
C. Li et al.
do not shift once BCH is employed. First, the BCH command uses the two covariates of task and contextual performance as the distal
outcomes for comparisons between profiles. Then, the BCH command uses Wald’s chi-square test to generate an overall test and
pairwise comparisons between profiles of covariates [59].
4.2. Results of study 2
4.2.1. Descriptive statistics and correlations
Table 4 exhibits the descriptive statistics and bivariate correlations of the research variables. Each mean for the five dimensions of
TRS was just above 3, indicating that employees in the sample were satisfied with the five elements of total reward. Each mean for task
and contextual performance was also just above 3, indicating that work performance was at a medium level. The correlations showed
that compensation satisfaction was not significantly related to task and contextual performance, while the remaining four dimensions
of TRS were all significantly related to task and contextual performance.
4.2.2. LPA
In response to RQ 1, Study 2 implemented LPA in Mplus 7.4 to test the profiles of Study 1. Table 5 presents multiple fit indicators to
test the possible profile models. From the one-profile to five-profile model, the values of AIC, BIC, and ABIC gradually declined,
indicating the model fit’s improvement; each entropy value exceeded 0.80, indicating that the model was accurate. The LMR p value of
the four-profile model was significant (p < 0.001), showing that it significantly outperformed the three-profile model. The LMR p value
of the five-profile model was not significant (p = 0.1758 > 0.05), showing that it did not significantly outperform the four-profile
model, and the model has a maximum of five profiles. In conclusion, the four-profile model outperformed the others, and was cho­
sen as study 2’s model.
Table 6 presents the five-dimensional means of TRS for the four-profile model, and their standardized mean plots are displayed in
Fig. 3. Table 6 and Fig. 3 reflect distinct quantitative and qualitative differences among the four profiles, which can be labeled as
follows: (a) DS, (b) DOS-dominant, (c) WLS-dominant, and (d) CS-dominant. Employees who belonged to DS (N = 88; 10.0%)
experienced dissatisfaction with all five dimensions, and the mean of each dimension was just above 2. Employees who belonged to the
DOS-dominant profile (N = 115; 13.0%) experienced satisfaction with development and career opportunities (M = 4.28) and neither
satisfaction nor dissatisfaction with the remaining four dimensions, of which the means were just above 3. Employees who belonged to
the WLS-dominant profile (N = 282; 31.9%) experienced satisfaction with work-life balance (M = 4.22) and neither satisfaction nor
dissatisfaction with the remaining four dimensions, of which the means were approximately 3. Employees who belonged to the CSdominant profile (N = 400; 45.1%) experienced satisfaction with compensation (M = 4.00) and neither satisfaction nor dissatisfac­
tion with the remaining four dimensions, of which the means were approximately 3. These findings again prove the importance of RQ
1, suggesting that four qualitatively and quantitatively differentiated profiles of TRS exist. In other words, there are four distinct
subpopulations of employees in TRS.
4.2.3. Predictors of profile membership
In response to RQ 2, Study 2 performed MLR analyses to verify how demographic characteristics predicted TRS profiles. The
formula was organized as
( )
Pi
logit
= α + β1 Age + β2 Education + β3 Gender + β4 Position level,
Pj
where Pi is the probability of profile i, Pj is the probability of reference profile j, logit(Pi /Pj ) is the logit function of the chosen two
profiles, and β1 , …, β4 are the coefficients of the responding predictors.
The model with demographic variables (− 2 log likelihood = 413.474, c2 = 283.423, df = 12, p < 0.001) had a significantly lower
chi-square value (− 2 log likelihood = 696.897), which indicated that demographic variables such as gender, age, education, and
position level were statistically significant. Table 7 provides the MLR results for these predictors’ effects on profile membership and
demographic variables, compared across profiles.
For female employees, the probability of belonging to DS was significantly higher than that of belonging to the DOS-dominant
profile (B = 1.454, p < 0.001, OR = 4.280), the WLS-dominant profile (B = 0.702), p = 0.020, OR = 2.018) or the CS-dominant
Table 4
Descriptive analysis and correlations among the variables of study 2.
CS
BS
WLS
PRS
DOS
TP
CP
MEAN
SD
CS
BS
WLS
PRS
DOS
TP
CP
3.27
3.00
3.33
3.00
3.09
3.23
3.32
0.87
0.89
1.02
0.86
0.91
0.60
0.60
1
0.26**
− 0.15**
0.18**
0.10**
0.03
0.05
1
0.14**
0.28**
0.34**
0.18**
0.20**
1
0.31**
0.21**
0.31**
0.39**
1
0.32**
0.22**
0.29**
1
0.25**
0.21**
1
0.68**
1
Note: **p < 0.01.
7
Heliyon 9 (2023) e14154
C. Li et al.
Table 5
Model fit indicators for LPA of study 2.
No. of profiles
AIC
BIC
ABIC
Entropy
p for LMR
p for BLRT
1
2
3
4
5
52590.354
42260.916
38193.706
37184.76
36326.727
52772.206
42538.48
38566.982
37653.747
36891.427
52651.526
42354.284
38319.27
37342.519
36516.683
0.991
1
0.954
0.962
0.0004
0.0004
<0.001
0.1758
<0.001
<0.001
<0.001
<0.001
Table 6
The five-dimensional means of TRS for the four latent profiles of study 2.
CS
BS
WLS
PRS
DOS
DS (n = 88; 10.0%)
DOS-dominant (n = 115; 13.0%)
WLS-dominant (n = 282; 31.9%)
CS-dominant (n = 400; 45.1%)
2.19
2.13
2.02
2.08
2.26
3.21
3.52
3.19
3.24
4.28
2.60
2.92
4.22
3.15
3.00
4.01
3.10
3.03
3.03
2.99
Fig. 3. Standardized means of TRS by four profiles of study 2. Note: The mean of a dimension score in a profile is 0, and the standard deviation is 1.
A score closer to 0 shows that the individuals’ dimension satisfaction level in the profile is closer to the overall sample’s mean for that dimension.
Table 7
MLR for the predictors’ effects on profile membership.
DS vs. DOS-dominant
Gender
Age
Education
Position level
DS vs. WLS-dominant
OR
Coefficient (SE)
OR
Coefficient (SE)
OR
1.454 (0.357) ***
− 1.825 (0.204) ***
1.851 (0.356) ***
2.958 (0.412) ***
4.280
0.161
6.363
19.255
0.702 (0.301) *
− 1.116 (0.166) ***
0.532 (0.277)
1.136 (0.308) ***
2.018
0.328
1.703
3.113
1.037 (0.301) **
− 1.398 (0.168) ***
0.805 (0.279) **
1.610 (0.308) ***
2.822
0.247
2.238
5.004
DOS-dominant vs.
WLS-dominant
Gender
Age
Education
Position level
DS vs. CS-dominant
Coefficient (SE)
DOS-dominant vs.
CS-dominant
WLS-dominant vs.
CS-dominant
Coefficient (SE)
OR
Coefficient (SE)
OR
Coefficient (SE)
OR
− 0.752 (0.243) **
0.709 (0.138) ***
− 1.318 (0.266) ***
− 1.822 (0.311) ***
0.472
2.032
0.268
0.162
− 0.416 (0.224)
0.428 (0.131) **
− 1.045 (0.251) **
− 1.347 (0.301) ***
0.659
1.533
0.352
0.260
0.335 (0.163) *
− 0.282 (0.083) **
0.273 (0.160)
0.475 (0.160) **
1.398
0.754
1.314
1.607
Note: *p < 0.05, **p < 0.01, ***p < 0.001; SE, standard error; OR: Odds Ratio. The coefficients and OR indicate the predictors’ effects on the
likelihood of membership in the former profile relative to the latter.
profile (B = 1.037, p = 0.001, OR = 2.822). The probability of belonging to the DOS-dominant profile was significantly lower than that
of belonging to the WLS-dominant profile (B = − 0.752, p = 0.002, OR = 0.472). The probability of belonging to the WLS-dominant
profile was significantly higher than that of belonging to the CS-dominant profile (B = 0.335, p = 0.040, OR = 1.398). There was no
significant difference between the probability of belonging to the DOS-dominant profile and belonging to the CS-dominant profile. In
8
Heliyon 9 (2023) e14154
C. Li et al.
summary, compared with male employees, female employees were most likely to belong to the DS profile, followed by the WLSdominant, DOS-dominant, and CS-dominant profiles, in order.
For older employees, the probability of belonging to the DS profile was significantly lower than that of belonging to the DOSdominant profile (B = − 1.825, p < 0.001, OR = 0.161), the WLS-dominant profile (B = − 1.116, p < 0.001, OR = 0.328) or the
CS-dominant profile (B = 0–1.398, p < 0.001, OR = 0.247). The probability of belonging to the DOS-dominant profile was significantly
higher than that of belonging to the WLS-dominant profile (B = 0.709, p < 0.001, OR = 2.032) or the CS-dominant profile (B = 0.428,
p = 0.001, OR = 1.533). The probability of belonging to the WLS-dominant profile was significantly lower than that of belonging to the
CS-dominant profile (B = − 0.282, p = 0.001, OR = 0.754). In summary, older employees most likely belonged to the DOS-dominant
profile, followed by the CS-dominant, WLS-dominant, and DS profiles, in order.
For less-educated employees, the probability of belonging to the DS profile was significantly higher than that of belonging to the
DOS-dominant profile (B = 1.851, p < 0.001, OR = 6.363) or the CS-dominant profile (B = − 0.897, p = 0.001, OR = 0.408). The
probability of belonging to the DOS-dominant profile was significantly lower than that of belonging to the WLS-dominant profile (B =
0.805, p = 0.004, OR = 2.238) or the CS-dominant profile (B = − 1.045, p < 0.001, OR = 0.352) profiles. In summary, compared with
highly educated employees, less-educated employees most likely belonged to the DS or WLS-dominant profile, followed by the CSdominant and DOS-dominant profiles, in order.
For ordinary employees, the probability of belonging to the DS profile was significantly higher than that of belonging to the DOSdominant profile (B = 2.958, p < 0.001, OR = 19.255), the WLS-dominant profile (B = 1.136, p < 0.001, OR = 3.113) or the CSdominant profile (B = 1.610, p < 0.001, OR = 5.004). The probability of belonging to the DOS-dominant profile was significantly
lower than that of belonging to the WLS-dominant profile (B = − 1.822, p < 0.001, OR = 0.162) or the CS-dominant profile (B =
− 1.347, p < 0.001, OR = 0.260). The probability of belonging to the WLS-dominant profile was significantly higher than that of
belonging to the CS-dominant profile (B = 0.475, p = 0.003, OR = 1.607). In summary, compared with low and middle management,
ordinary employees most likely belonged to the DS profile, followed by the WLS-dominant, CS-dominant, and DOS-dominant profiles,
in order.
In conclusion, demographic characteristics differentiate the difference TRS profiles. The highest likelihood of membership in the DS
profile belonged to female, less-educated, youngest or ordinary employees. The highest likelihood of membership in the DOSdominant profile belonged to male, highly educated, older employees or to low- and middle-level management. The highest likeli­
hood of membership in the WLS-dominant profile belonged to less-educated employees, next came female, younger, or ordinary
employees. The highest likelihood of membership in the CS-dominant profile belonged to male employees, followed by less-educated,
older, or ordinary employees.
These results confirm the importance of RQ 2, indicating that demographic variables involving gender, age, education, and position
level affect each the likelihood of membership in each profile of TRS. Alternatively, similar individuals with specific demographic
variables are more likely to share a specific profile.
4.2.4. Outcome variable analysis of profiles
In response to RQ 3, Study 2 used the BCH command [58] to evaluate how each of the four profiles influence work performance.
The results are shown in Table 8. Task and contextual performance were significantly different in the four TRS profiles (χ2 = 128.564,
p < 0.001; χ2 = 133.648, p < 0.001).
In the DS, DOS-dominant, WLS-dominant, and CS-dominant profiles, the mean values of task performance were M = 2.136, M =
3.069, M = 3.627, and M = 2.858, respectively; meanwhile, the mean values of contextual performance were M = 2.091, M = 3.175,
M = 3.453, and M = 2.906, respectively. Employees in the WLS-dominant profile presented significantly higher task and contextual
performance than those in the other three profiles. Employees in the DOS-dominant profile presented significantly lower task and
contextual performance than those in the WLS-dominant profile, but significantly higher performance than those in the CS-dominant
and DS profiles. Employees in the CS-dominant profile presented significantly lower task and contextual performance than those in the
WLS-dominant and DOS-dominant profiles but showed significantly higher performance than those in the DS profile. Taken together,
the WLS-dominant profile was linked with the highest level of task and contextual performance, followed by the DOS-dominant, CSdominant, and finally, the DS profiles.
Combined, these results confirm the importance of RQ 3-1, RQ 3-2, and RQ 3, illustrating that the each of the four TRS profiles
predict different levels of task and contextual performance, i.e., different levels of work performance.
Table 8
Outcome means and pairwise comparisons between profiles.
Profile means
DS
TP
2.136
DOSdominant
3.069
CP
2.091
3.175
Profile comparisons
Differences between
profiles
WLSdominant
3.627
CSdominant
2.858
1 vs. 2
1 vs. 3
1 vs. 4
2 vs. 3
2 vs. 4
3 vs. 4
***
***
***
***
**
***
χ2 = 128.564;
3.453
2.906
***
***
***
*
**
***
χ2 = 133.648;
Note: *p < 0.05, **p < 0.01, ***p < 0.001.1, DS; 2, DOS-dominant; 3, WLS-dominant; 4, CS-dominant.
9
3 > 2>4 > 1
3 > 2>4 > 1
Heliyon 9 (2023) e14154
C. Li et al.
4.2.5. Supplemental analysis using a variable-centered approach
To further exhibit the unique value of introducing a person-centered approach, Study 2 tested the impact of TRS on work per­
formance by applying hierarchical regression analysis to an overall sample. The results are shown in Table 9. For task performance,
satisfaction with one’s work-life balance, performance and recognition, and development and career opportunities had significant
positive predictive effects, while satisfaction with compensation and benefit had no significant predictive effect. For contextual
performance, satisfaction with benefit, work-life balance, and performance and recognition had significant positive predictive effects,
while satisfaction with compensation and development and career opportunities had no significant predictive effect.
These results only demonstrate that the effect of each dimension of TRS on work performance is independent. Therefore, by
assuming a homogeneous sample population, the variable-centered study could not detect the different combinations of TRS di­
mensions or reveal how the profiles influence work performance from the perspective of individual differences.
5. Discussion
We introduced a person-centered approach and performed an LPA to discover TRS profiles with the use of two samples. Based on
constructing mixture models of profiles, we showed the predictability of demographic variables on profile similarity and the differ­
ential effects of TRS profiles on work performance. Our exploratory findings confirm the importance of RQ 1, RQ 2, and RQ 3 (RQ 3-1,
RQ 3-2).
Our study determined that there are four TRS profiles (subpopulations) and they differ quantitatively and qualitatively. Employees
of DS subpopulation experienced dissatisfaction with all five dimensions. Employees of the DOS-dominant, WLS-dominant, and CSdominant subpopulations respectively experienced satisfaction with development and career opportunities, work-life balance, and
compensation, while they experienced neither satisfaction nor dissatisfaction with the remaining four dimensions. Our result supports
the hypothesis of sample heterogeneity in the person-centered study and conforms to the LPA standard [17,21,24,25]. Compared with
the results of Hofmans et al. [19] and De Gieter and Hofmans [5] that use traditional cluster-wise analysis, LPA based on probabilistic
classification can more accurately identify the classification of subpopulations [20]. Therefore, our study captured the unobserved
heterogeneity in the TRS population, while previous variable-centered studies found it difficult to identify such individual differences.
Our study suggested that similar individuals with specific demographic variables are more likely to share a specific profile of TRS.
The DS profile was most likely to consist of female, less-educated, youngest, or ordinary employees. The DOS-dominant profile was
most likely to consist of male, highly educated, older employees, or low- and middle-level management. The WLS-dominant profile was
most likely to consist of less-educated employees, followed by female, younger, or ordinary employees. The CS-dominant profile was
more likely to consist of male employees who were less-educated, followed by older, or low- and middle-level management. Our result
confirms a study using a person-centered approach can use demographic variables as predictors to more richly describe the nature of
each profile [21]. In contrast, previous variable-centered studies have shown that individuals with different demographic charac­
teristics attach distinct value to the types of total rewards [11,28,29,30,31,32]. However, in those studies, demographic variables are
used as control variables to reduce their impact on the relationships between other research variables but are usually not reported as
study results. Therefore, our study verified that demographic variables are sensitive to complex combinations among the dimensions of
TRS.
Our study uncovered four subpopulations that differentially predict work performance. On one hand, our results confirm the core
view of social exchange theory. Based on social exchange theory, social exchanges require openness and trigger adaptability and trust,
whereas economic exchanges tend to be based on explicit exchange conditions and produce less trust and more supervision [36].
Work-life balance and development and career opportunities are more likely to promote social exchange than compensation and
benefit [37]. Therefore, according to the positive-negative hedonic value model [35,41,42], employees in the WLS-dominant and
DOS-dominant profiles exhibited a higher work performance than CS-dominant employees, while employees in the DS profile
exhibited the lowest work performance.
On the other hand, compared with the study of De Gieter and Hofmans [5], our study more accurately reveals the differences in the
impact of TRS on work performance among employee subgroups because LPA is superior to traditional cluster analysis. Moreover, a
supplemental analysis of our study demonstrated that certain dimensions of TRS have a significant positive effect or no significant
Table 9
Results of regression analysis of the overall sample predicting work performance.
TP
Gender
Age
Education
Position level
CS
BS
WLS
PRS
DOS
CP
В
SE
P Value
В
SE
P Value
0.027
0.108***
0.064*
− 0.001
− 0.005
0.037
0.143***
0.059*
0.065**
0.039
0.020
0.026
0.017
0.023
0.024
0.020
0.024
0.023
0.485
0.000
0.014
0.944
0.838
0.129
0.000
0.015
0.006
0.032
0.096***
0.058*
0.008
0.012
0.051*
0.184***
0.097***
0.013
0.037
0.019
0.024
0.016
0.022
0.023
0.019
0.023
0.022
0.382
0.000
0.018
0.615
0.593
0.027
0.000
0.000
0.549
Note: *p < 0.05, **p < 0.01, ***p < 0.001.
10
Heliyon 9 (2023) e14154
C. Li et al.
effect on work performance, which is similar to the findings of previous variable-centered studies [3,5,13,14,44,45,46,47,48,49,50,
51]. All of these findings illustrate the limitations of exploring the independent effects of various dimensions of TRS on work per­
formance. However, based on the person-centered approach, our findings suggest that combinations of various TRS dimensions have a
different impact on work performance from the perspective of individual differences rather than from that of a single dimension. Thus,
our results can help to overcome the contradictory and difficult theoretical explanation of previous findings based on the
variable-centered approach [21,60].
6. Conclusion
Our findings demonstrated that there are four profiles or subpopulations of TRS and that they quantitatively and qualitatively
differed. DS varied quantitatively (in level), whereas the DOS-dominant, WLS-dominant, and CS-dominant profiles varied qualitatively
(in shape). Moreover, demographic variables, including gender, age, education, and position level, significantly predicted TRS profile
membership. Finally, TRS profiles exhibited different levels of work performance, that is, different levels of task and contextual
performance. The WLS-dominant profile displayed the highest task and contextual performance, followed by the DOS-dominant, CSdominant, and DS profiles.
6.1. Theoretical contributions
Based on a person-centered approach and by conducting LPA, our exploratory study examined TRS profiles and their incentive
effects, which fills the gap in previous research based on the use of a variable-centered approach and traditional cluster-wise analysis.
First, the four profiles or subpopulations of TRS are more accurate and consistent with reality, enriching the literature on TRS research.
Second, similar individuals with specific demographic variables were more likely to share a specific profile, which is an essential
complement to the findings of previous variable-centered studies. Third, the relationship between TRS profiles and work performance
offers new insight into the incentive effect of total rewards from the perspective of individual differences, which promotes theoretical
research in this field.
6.2. Practical implications
When using a person-centered approach, one needs to prove that the profiles not only are of theoretical significance but also have
practical value [57]. Our study sampled employees of retail enterprises suffering from labor costs. Therefore, our results offer theo­
retical guidance for organizations with regard to customizing the policies of a total reward system, with the aim of achieving a
breakthrough in incentive management given the difficulties of the current financial situation generally faced by organizations due to
the COVID-19 pandemic.
Our study supports the idea of classification management, which means fundamentally switching between analytical mindsets and
in turn having an important impact on the design of management policies [17,61]. Thereby, organizations need to transform the
concept of management into that of classification management for designing the total reward system. The various TRS subpopulations
should be considered in the policies of total reward to motivate target employees to improve work performance. Regarding the en­
terprises similar to the sample of this study, a differentiated and classified total reward system should be customized and implemented.
For the DS subpopulation, an enterprise should make improvements to the five types of total reward. For the WLS-dominant sub­
population, an enterprise should provide a more robust work-life balance policy. For the DOS-dominant subpopulation, an enterprise
should furnish more career development opportunities. Finally, for the CS-dominant subpopulation, an enterprise should offer better
monetary rewards.
6.3. Limitations and future research
First, our sample is taken from employees of the retail industry in China, which limits the applicability of the findings in other
countries and industries. The profile classification may vary from sample to sample, so future studies should use samples from different
countries (e.g., Europe, North America) and industries (e.g., information industry) to test the consistency of TRS profiles.
Second, this study only used cross-sectional data for testing, which cannot assess the stability of potential profiles and is not
conducive to reflecting the causal relationship between profiles and work performance. Longitudinal data and latent transition analysis
(LTA) should be employed to assess the stability of latent profiles at different points in time and whether latent profiles and outcome
variables have a true causal relationship [16,20]. Therefore, future research should investigate the development and stability of TRS
profiles based on longitudinal data to infer whether TRS profiles and work performance have a causal relation.
Third, all variables were measured by the self-reported ratings of employees, which may result in common method bias. The
distortion of a statistical analysis produced by common method bias is less for studies using LPA than for variable-centered studies
[21]. However, the possible impact on this study of common method bias cannot be ruled out. Future research should consider
employee-supervisor dyads in which superiors evaluate the work performance of their subordinates.
11
Heliyon 9 (2023) e14154
C. Li et al.
Declarations
Author contribution statement
Chunling Li: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Contributed
reagents, materials, analysis tools or data; Wrote the paper.
Xueyun Duan: Performed the experiments; Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or
data; Wrote the paper.
Xiangling Chu: Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data.
Yong Qiu: Analyzed and interpreted the data; Wrote the paper.
Funding statement
This research was supported by the National Social Science Fund of China (Grant No. 19CGL028) and the Beijing Municipal Social
Science Foundation [Grant No. 15JGB061]. The sponsors played no role in research design, collecting and analyzing data, drawing
research results, preparing the manuscript, or even in publication decisions.
Data availability statement
Data will be provided upon request.
Declaration of interests statement
The authors declare no conflicts of interest.
Additional information
This article does not contain additional information.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.heliyon.2023.e14154.
References
[1] R.L. Heneman, Implementing Total Rewards Strategies: a Guide to Successfully Planning and Implementing a Total Rewards System, 2007. Retrieved from,
http://www.shrm.org/hrdisciplines/benefits/documents/07rewardsstratreport.pdf.
[2] A.T. Landry, A. Whillans, The power of workplace rewards: using self-determination theory to understand why reward satisfaction matters for workers around
the world, Compensat. Benefit Rev. 50 (3) (2018) 123–148, https://doi.org/10.1177/0886368719840515.
[3] C. Li, L. Yin, Can total rewards satisfaction improve employees’ job performance in retail enterprises? J. Chin. Univ. Labor. Relat. 33 (4) (2019) 88–100.
[4] A. Rai, P. Ghosh, T. Dutta, Total rewards to enhance employees’ intention to stay: does perception of justice play any role? Evid-Based. Hrm. 7 (3) (2019)
262–280, https://doi.org/10.1108/EBHRM-07-2018-0045.
[5] S. De Gieter, J. Hofmans, How reward satisfaction affects employees’ turnover intentions and performance: an individual differences approach, Hum. Resour.
Manag. J. 25 (2) (2015) 200–216, https://doi.org/10.1111/1748–8583.12072.
[6] WorldatWork, 50 Total Rewards Programs and Practices: a Survey of what Is in Use Today, WorldatWork J, 2015. Retrieved from, https://www.worldatwork.
org.
[7] J. Sarkar, Measuring total rewards satisfaction: a scale development study, Indian J. Ind. Relat. 57 (3) (2022) 430–449, https://doi.org/10.31124/
advance.14909223.v1.
[8] M. Bussin, D.J. Van Rooy, Total rewards strategy for a multi–generational workforce in a financial institution, SA J. Hum. Resour. Manag. 12 (1) (2014) 1–11,
https://doi.org/10.4102/sajhrm.v12i1.606.
[9] C. Hartmann, J. Stuedemann, B. Viney, The right stuff: exploring total rewards in 2021 and beyond, Workspan 63 (6) (2020) 12–17.
[10] A.M. Peluso, L. Innocenti, M. Pilati, Pay is not everything: differential effects of monetary and non-monetary rewards on employees’ attitudes and behaviours,
Evid-Based. Hrm. 5 (3) (2017) 311–327, https://doi.org/10.1108/EBHRM-07-2015-0031.
[11] J. Tarigan, J. Cahya, A. Valentine, S. Hatane, F. Jie, Total reward system, job satisfaction and employee productivity on company financial performance:
evidence from Indonesian generation z workers, J. Asia. Bus. Stud. 16 (6) (2022) 1041–1065, https://doi.org/10.1108/JABS-04-2021-0154.
[12] D.B. Balkin, S. Werner, Theorizing the relationship between discretionary employee benefits and individual performance, Hum. Resour. Manag. Rev. 33 (1)
(2023), 100901, https://doi.org/10.1016/j.hrmr.2022.100901.
[13] S.C. Currall, A.J. Towler, T.A. Judge, L. Kohn, Pay satisfaction and organizational outcomes, Person. Psychol. 58 (3) (2005) 613–640, https://doi.org/10.1111/
j.1744–6570.2005.00245.x.
[14] K. Khalid, The impact of managerial support on the association between pay satisfaction, continuance and affective commitment, and employee task
performance, Sage Open 10 (1) (2020) 1–13, https://doi.org/10.1177/2158244020914591.
[15] B. Lyu, W.W. Li, M.Y. Xu, H. Chen, Y.C. Yang, All normal occupations are sunny and joyful: qualitative analysis of Thai ladyboys’ occupational wellbeing,
Psychol. Res. Behav. Manag. 14 (2021) 2197–2208, https://doi.org/10.2147/PRBM.S340209.
[16] M.C. Howard, M.E. Hoffman, Variable–centered, person–centered, and person–specific approaches: where theory meets the method, Organ. Res. Methods 21 (4)
(2018) 846–876, https://doi.org/10.1177/1094428117744021.
12
Heliyon 9 (2023) e14154
C. Li et al.
[17] A.J.S. Morin, A. Bujacz, M. Gagné, Person–centered methodologies in the organizational sciences: introduction to the feature topic, Organ. Res. Methods 21 (4)
(2018) 803–813, https://doi.org/10.1177/1094428118773856.
[18] H. Aguinis, R.K. Gottfredson, Best–practice recommendations for estimating interaction effects using moderated multiple regression, J. Organ. Behav. 31 (6)
(2010) 776–786, https://doi.org/10.1002/job.686.
[19] J. Hofmans, S. De Gieter, R. Pepermans, Individual differences in the relationship between satisfaction with job rewards and job satisfaction, J. Vocat. Behav. 82
(1) (2013) 1–9, https://doi.org/10.1016/j.jvb.2012.06.007.
[20] M. Wang, P.J. Hanges, Latent class procedures: applications to organizational research, Organ. Res. Methods 14 (1) (2011) 24–31, https://doi.org/10.1177/
1094428110383988.
[21] J.P. Meyer, A.J.S. Morin, A person–centered approach to commitment research: theory, research, and methodology, J. Organ. Behav. 37 (4) (2016) 584–612,
https://doi.org/10.1002/job.2085.
[22] S.C. Payne, A.L. Cook, M.T. Horner, M.K. Shaub, W.R. Boswell, The relative influence of total rewards elements on attraction, motivation and retention,
WorldatWork J. 20 (1) (2010), 6–21, Retrieved from, https://www.worldatwork.org.
[23] D. Bouckenooghe, D. De Clercq, U. Raja, A person–centered, latent profile analysis of psychological capital, Aust. J. Manag. 44 (1) (2019) 91–108, https://doi.
org/10.1177/0312896218775153.
[24] A.S. Gabriel, M.A. Daniels, J.M. Diefendorff, G.J. Greguras, Emotional labor actors: a latent profile analysis of emotional labor strategies, J. Appl. Psychol. 100
(3) (2015) 863–879, https://doi.org/10.1037/a0037408.
[25] H.W. Marsh, O. Lüdtke, U. Trautwein, A. Morin, Classical latent profile analysis of academic self–concept dimensions: synergy of person– and variable–centered
approaches to theoretical models of self–concept, Struct. Equ. Model. 16 (2009) 191–225, https://doi.org/10.1080/10705510902751010.
[26] T. Bridgman, S. Cummings, J. Ballard, Who built maslow’s pyramid? a history of the creation of management studies’ most famous symbol and its implications
for management education, Acad. Manag. Learn. Educ. 18 (1) (2019) 81–98, https://doi.org/10.5465/amle.2017.0351.
[27] C. Montag, C. Sindermann, D. Lester, K.L. Davis, Linking individual differences in satisfaction with each of Maslow’s needs to the big five personality traits and
Panksepp’s primary emotional systems, Heliyon 6 (7) (2020), https://doi.org/10.1016/j.heliyon.2020.e04325, 1, 9.
[28] M. Leaf, R. Ryan, Beyond Compensation: How Employees Prioritize Total Rewards at Various Life Stages, WorldatWork J, 2010. Retrieved from, https://www.
worldatwork.org.
[29] M. Von Bonsdorff, Age–related differences in reward preferences, Int. J. Hum. Resour. Manag. 22 (6) (2011) 1262–1276, https://doi.org/10.1080/
09585192.2011.559098.
[30] W.L. Su, B. Lyu, H. Chen, Y.Z. Zhang, How does servant leadership influence employees’ service innovative behavior? the roles of intrinsic motivation and
identification with the leader, Baltic J. Manag. 15 (4) (2020) 571–586, https://doi.org/10.1108/BJM-09-2019-0335.
[31] P.E. French, M.C. Emerson, Assessing the variations in reward preference for local government employees in terms of position, public service motivation, and
public sector motivation, Public. Perform. Manag. 37 (4) (2014) 552–576, https://doi.org/10.2753/PMR1530-9576370402.
[32] A.J. Dubinsky, R.E. Anderson, R. Mehta, Importance of alternative rewards: impact of managerial level, Ind. Market. Manag. 29 (5) (2000) 427–440, https://
doi.org/10.1016/S0019–8501(99)00070–x.
[33] J.P. Campbell, B.M. Wiernik, The modeling and assessment of work performance, Annu. Rev. Organ. Psych./Organ. Behav. 2 (1) (2015) 47–74, https://doi.org/
10.1146/annurev-orgpsych-032414-111427.
[34] S.J. Motowidlo, J.R. Van Scotter, Evidence that task performance should be distinguished from contextual performance, J. Appl. Psychol. 79 (4) (1994)
475–480, https://doi.org/10.1037/0021–9010.79.4.475.
[35] R. Cropanzano, E.L. Anthony, S.R. Daniels, A.V. Hall, Social exchange theory: a critical review with theoretical remedies, Acad. Manag. Ann. 11 (1) (2017)
479–516, https://doi.org/10.5465/annals.2015.0099.
[36] C. Porter, Long live social exchange theory, Ind. Organ. Psychol. 11 (3) (2018) 498–504, https://doi.org/10.1017/iop.2018.102.
[37] P. Blau (Ed.), Exchange and Power in Social Life, Routledge, New York, 1986, https://doi.org/10.4324/9780203792643.
[38] M. Madanoglu, Theories of economic and social exchange in entrepreneurial partnerships: an agenda for future research, Int. Enterpren. Manag. J. 14 (2018)
649–656, https://doi.org/10.1007/s11365-018-0515-6.
[39] C.M. Alcover, M.J. Chambel, Y. Estreder, Monetary incentives, motivational orientation and affective commitment in contact centers. a multilevel mediation
model, J. Econ. Psychol. 81 (2020) 1–11, https://doi.org/10.1016/j.joep.2020.102307.
[40] R. Yasin, G. Jan, A. Huseynova, M. Atif, Inclusive leadership and turnover intention: the role of follower-leader goal congruence and organizational
commitment, Manag. Decis. (2023), https://doi.org/10.1108/MD-07-2021-0925 ahead-of-print.
[41] B. Kuvaas, L.M. Shore, R. Buch, A. Dysvik, Social and economic exchange relationships and performance contingency: differential effects of variable pay and
base pay, Int. J. Hum. Resour. 31 (3) (2020) 408–431, https://doi.org/10.1080/09585192.2017.1350734.
[42] A. Thomas, V. Gupta, Social capital theory, social exchange theory, social cognitive theory, financial literacy, and the role of knowledge sharing as a moderator
in enhancing financial well-being: from bibliometric analysis to a conceptual framework model, Front. Psychol. 12 (2021) 1–16, https://doi.org/10.3389/
fpsyg.2021.664638.
[43] L.M. Chen, Y.R. Guo, L.J. Song, B. Lyu, From errors to OCBs and creativity: a multilevel mediation mechanism of workplace gratitude, Curr. Psychol. 41 (9)
(2021) 6170–6184, https://doi.org/10.1007/s12144-020-01120-5.
[44] M.L. Williams, M.A. McDaniel, N.T. Nguyen, A meta–analysis of the antecedents and consequences of pay level satisfaction, J. Appl. Psychol. 91 (2) (2006)
392–413, https://doi.org/10.1037/0021–9010.91.2.392.
[45] H. Fang, Y. Ge, The impact of science and technology enterprise employees’ payment satisfaction on their work performance from the view of perceived
organizational support, Technol. Innov. Manag. (4) (2016) 411–416, https://doi.org/10.14090/j.cnki.jscx.2016.0413.
[46] L.H. Faulk, Pay Satisfaction Consequences: Development and Test of a Theoretical Model, PhD’s Thesis, Louisiana State University and Agricultural and
Mechanical College, 2002.
[47] B.D. Edwards, S.T. Bell, W. Arthur Jr., A.D. Decuir, Relationships between facets of job satisfaction and task and contextual performance, Appl. Psychol. 57 (3)
(2008) 441–465, https://doi.org/10.1111/j.1464–0597.2008.00328.x.
[48] I.O. Ganiyu, Z. Fields, S.O. Atiku, Work–life balance strategies, work–family satisfaction and employees’ job performance in Lagos, Nigeria’s manufacturing
sector, J. Contemp. Manag. 14 (1) (2017) 441–460, https://doi.org/10.10520/EJC–8c772cc8f.
[49] J.H. Wayne, M.M. Butts, W.J. Casper, T.D. Allen, In search of balance: a conceptual and empirical integration of multiple meanings of work–family balance,
Person. Psychol. 70 (2017) 167–210, https://doi.org/10.1111/peps.12132.
[50] S. De Gieter, R. De Cooman, R. Pepermans, M. Jegers, Manage through rewards, not only through pay: establishing the Psychological Reward Satisfaction Scale
(PRSeSS), in: M. Vartiainen, C. Antoni, X. Baeten, R. Lucas (Eds.), Reward Management: Facts and Trends in Europe, Pabst Science Publishers, Lengerich, 2008,
pp. 97–117.
[51] L. Hou, L.J. Song, G.Y. Zheng, B. Lyu, Linking identity leadership and team performance: the role of group-based pride and leader political skill, Front. Psychol.
12 (2021), 676945, https://doi.org/10.3389/fpsyg.2021.676945.
[52] M.K. Ward, A.W. Meade, Applying social psychology to prevent careless responding during online surveys, Appl. Psychol. 67 (2) (2018) 231–263, https://doi.
org/10.1111/apps.12118.
[53] T. Zhong, H. Wang, Motivation profiles for physical activity among office workers, Front. Psychol. 10 (2019) 1–8, https://doi.org/10.3389/fpsyg.2019.01577.
[54] A.J. Morin, J. Morizot, J.S. Boudrias, I. Madore, A multi–foci person–centered perspective on workplace affective commitment: a latent profile/factor mixture
analysis, Organ. Res. Methods 14 (2011) 58–90, https://doi.org/10.1177/1094428109356476.
[55] Y. Lee, Identifying latent profiles in work-to-family conflict and family-to-work conflict, Hum. Resour. Dev. Q. 29 (2018) 203–217, https://doi.org/10.1002/
hrdq.21312.
[56] T.C. Yu, The Influence of Human Side System Factors on the Work Performance in a Context of Quality Management, PhD’s Thesis, National Sun Yat–sen
University, 1996.
13
Heliyon 9 (2023) e14154
C. Li et al.
[57] J.P. Meyer, A.J. Morin, S.A. Wasti, Employee commitment before and after an economic crisis: a stringent test of profile similarity, Hum. Relat. 71 (9) (2018)
1204–1233, https://doi.org/10.1177/0018726717739097.
[58] T. Asparouhov, B. Muthén, Auxiliary variables in mixture modeling: using the BCH method in Mplus to estimate a distal outcome model and an arbitrary
secondary model, Mplus Web Notes (2020). Retrieved from, http://www.statmodel.com.
[59] A. Mäkikangas, Job crafting profiles and work engagement: a person–centered approach, J. Vocat. Behav. 106 (2018) 101–111, https://doi.org/10.1016/j.
jvb.2018.01.001.
[60] A.S. Gabriel, J.T. Campbell, E. Djurdjevic, R.E. Johnson, C.C. Rosen, Fuzzy profiles: comparing and contrasting latent profile analysis and fuzzy set qualitative
comparative analysis for person–centered research, Organ. Res. Methods 21 (4) (2018) 877–904, https://doi.org/10.1177/1094428117752466.
[61] J. Howard, M. Gagné, A.J.S. Morin, A. Van den Broeck, Motivation profiles at work: a self-determination theory approach, J. Vocat. Behav. 95–96 (2016) 74–89,
https://doi.org/10.1016/j.jvb.2016.07.004.
14
Download