What drives job satisfaction among community pharmacists? An application of relative importance analysis

Abstract

Background

Pharmacy employers want to improve pharmacists' job satisfaction, but ratings of job satisfaction are highly subjective, as evaluating job satisfaction involves weighing simultaneously the importance of multiple correlated determinants that are often perceived unequally.

Objectives

To 1) describe the application of relative importance analysis in estimating the predictive ability of correlated determinants of job satisfaction, and to rank the determinants in order of relative importance, and 2) explore how the perceived relative importance of job satisfaction predictors may vary across community pharmacists' age, gender, and work setting categories.

Methods

Data were obtained from the 2019 National Pharmacy Workforce Survey administered to 96,110 licensed U.S. pharmacists. Multiple regression analysis (MR) and relative weight analysis (RWA) were used to assess the predictive ability of determinants to explain pharmacists' job satisfaction. Subgroup analyses were performed to explore variations in the perceived relative importance of predictors across pharmacists' age, gender and work setting categories.

Results

Over the entire sample of community pharmacists, no personal experience of workplace discrimination [RW = 0.0613, rank = 1] and less reported engagement in advanced dispensing activities [RW = 0.0235, rank = 2] were most associated with greater job satisfaction, as both predictors jointly accounted for 67.5% of the predicted criterion variance (R²). Pharmacists' compensation was observed to have the lowest perceived relative importance for predicting job satisfaction [RW = 0.0005, rank = 6], accounting for 0.5% of R². Between-group comparisons showed that, no personal experience of workplace discrimination had the highest perceived relative importance for job satisfaction across pharmacists' age groups, among women, and across most work settings except small chain pharmacies. Within-group comparisons showed that pharmacists' compensation was significantly more important than professional interactions (Δ_RW(PC-PPI) 0.2900 [0.0637, 0.5360]) for job satisfaction among pharmacists in large chain pharmacies, while patient-care services was significantly more important than compensation for job satisfaction in independent (Δ_RW(PPS-PC) 0.1761 [0.0017, 0.5980]) and health system retail pharmacists (Δ_RW(PPS-PC) 0.4190 [0.0444, 0.8303]).

Conclusions

Relative importance analysis corroborated multiple regression and provided a more interpretable presentation of variable influence on community pharmacists job satisfaction as the importance of personal and workplace characteristics in how pharmacists evaluate their job satisfaction varied across age, gender and work setting categories.

1. Introduction

Job satisfaction refers to workers' evaluation of how their current job realities compare with their expectations, and can be defined as how much positive affect workers have towards their jobs.^1,2 In pharmacy, high job satisfaction has been linked to pharmacists work engagement, organizational commitment, and reduced worker turnover.3, 4, 5 Job satisfaction has also been observed to influence the delivery and quality of health care, and to mediate the relationship between some work characteristics and patient outcomes.^6,7 Considering the implications of job satisfaction for organizational and patient outcomes, there is a need for pharmacy professionals and organizations to understand the determinants of pharmacists' job satisfaction, and how these determinants work in relation to one another to influence pharmacists' job satisfaction. Previous research studies have shown that job satisfaction in pharmacists is determined by personal characteristics like age, gender, and level of education; and work characteristics like workload, work-life balance, work setting, job autonomy, peer and supervisor support, compensation, opportunities to provide patient-care services, and career advancement.^2,8, 9, 10, 11, 12

Although job satisfaction is arguably one of the most researched constructs in the pharmacy workforce literature, ratings of job satisfaction are highly subjective in an operational context. Pharmacists within the same work setting or exposed to the same set of work conditions may evaluate their individual job satisfaction differently. For instance, pharmacists working in an independent community pharmacy setting may prioritize task and skill variety over compensation as a determinant of job satisfaction, while the converse may be true for pharmacists working in a chain pharmacy.¹¹ This suggests that when evaluating their job satisfaction, pharmacists consider multiple personal and workplace characteristics simultaneously, and seldom just choose between factors that are considered to be more favorable and those that are considered to be less favorable. That is, reaching an assessment of job satisfaction involves tradeoffs and weighing the importance of multiple correlated determinants that are often perceived unequally.12, 13, 14, 15

Conventionally, social science researchers have used regression weights (or coefficients) as a measure of variable importance. However, regression weights, which represent the unique contribution of predictors to explaining the variance of a criterion, may fail to accurately partition the criterion variance to predictors when predictors are correlated.¹⁶ Regression weights also may not fully capture the real-world context of the relationship between predictor variables and a criterion where variables often exert their effects simultaneously. This is reflected in the interpretation of the regression weight of a predictor in which the researcher assumes that there is no change in other predictor variables as indicated by the caveat “holding the effect of all other variables constant” that is included in reports of regression results.¹⁷ When predictors are correlated, the predictive ability of variables becomes more of a shared property. In such a case, adding or removing a variable from the regression model could produce substantial changes in the size of regression weights, since high correlation between variables creates a situation where a change in one variable results in a change in both the criterion and the other correlated predictors. This makes the separation of the unique individual contribution of predictors to explaining the criterion variance using regression coefficients alone difficult.^17,18

Conceptually and empirically correlated variables are pervasive in social and administrative pharmacy research. To address the problem of predictor comparison when variables are correlated, relative importance analysis was developed to complement traditional multiple regression analysis. Relative importance analysis refers to a set of statistical techniques, e.g., relative weight analysis¹⁹ and dominance analysis,²⁰ that are used to determine the contribution of a predictor to the predicted criterion variance (R²) when the predictor is considered by itself and in combination with other predictors in a model, and to provide estimates of variable importance for a set of multiple correlated predictors.²¹ Consider a pharmacy manager interested in improving the job satisfaction of his pharmacist employees. The pharmacists in this scenario have many aspects of their job to consider. They are less likely to judge their job satisfaction by evaluating each aspect of their job as an independent factor, but rather, would probably consider all relevant job characteristics simultaneously while implicitly attributing varying degrees of importance to each aspect. The pharmacy manager needs to know which job characteristics are important to, for instance, men pharmacists compared to women pharmacists, or older pharmacists relative to younger pharmacists, to guide the allocation of resources or the design and implementation of relevant interventions.¹¹ Empirically, relative importance can be estimated by dividing the explained variance (R²) of a criterion among predictors while including all the measured variables in the model using relative weight analysis or dominance analysis. That is, relative importance analysis reduces the need for a caveat when interpreting model results as its estimation procedure takes the correlation between predictors into account, and also that variables typically exert their effects concurrently in a real-world context absent of experimental controls and manipulation.²² In other words, relative importance analysis allows the manager or researcher to determine the proportionate contribution of a predictor to explaining the variance of an outcome after correcting for the intercorrelations among predictors. The theoretical underpinnings and statistical procedures for conducting relative importance analysis are described elsewhere.^16,18,19,23

In this study, the objectives were 1) to describe the application of relative importance analysis to address the issue of correlated predictors in estimating the predictive ability of some determinants of job satisfaction in a national sample of community pharmacists, and to rank order these determinants in terms of their relative importance, and 2) to explore how the perceived relative importance of job satisfaction predictors may vary across community pharmacists' age groups, gender, and work settings. Although, relative weight analysis (RWA) and dominance analysis (DA) yield almost identical results, we used RWA estimation procedures in the present study as RWA is more computationally efficient than DA for assigning importance weights to a set of predictors. Moreover, it is beyond the scope of the present study to assess how the predictive ability of a predictor varies across all possible subsets of predictors in the model, for which DA will be better suited.²⁴

2. Method

2.1. Sampling and data collection

Data were collected on the 2019 National Pharmacy workforce Survey (NPWS). IRB approval for the survey was granted by the University of Iowa. Witry et al. (2021) reported a full description of survey methods, nonresponse bias, and limitations.²⁵ Briefly, the NPWS was sent electronically to a random sample of 96,110 licensed U.S. pharmacists. Responses were considered usable for the present study if respondents identified as practicing community pharmacists, and were sufficiently complete for key demographic variables (age, gender, and work setting) and for the job satisfaction items.

2.2. Measures

2.2.1. Outcome variable: job satisfaction

Community pharmacists' job satisfaction was assessed on a 3-item 4-point scale (1 = very dissatisfied to 4 = very satisfied) adapted from Bacharach et al.'s (1991) 5-item 5-point scale²⁶ by dropping 2 items and the midpoint response for parsimony. Pharmacists were asked to indicate, in general, how satisfied are you with […]. Items on the adapted job satisfaction scale were “the chance your job gives you to do what you are best at doing”, “your present job in light of your career expectations”, and “your present job when you consider the expectations you had when you took the job.” The adapted pharmacists' job satisfaction scale was found to have high internal consistency reliability with a Cronbach's alpha of 0.91. A dependent Job Satisfaction variable was created for use in regression and relative weight analyses by taking the average score across the three job satisfaction items for each respondent.

2.2.2. Determinants of community pharmacists' job satisfaction

Personal characteristics and four major domains of community pharmacy work characteristics served as determinants of job satisfaction. Personal characteristics assessed in this study were pharmacists' age, gender, and work setting. The work characteristics domains included a) pharmacists' patient-care services, b) pharmacists' compensation, c) pharmacists' professional interactions, and d) no personal experience of workplace discrimination. Previous research on job satisfaction has identified patient-care services and compensation as important predictors of job satisfaction.^7,8 Professional interactions with healthcare providers and other health professional was expected to contribute to job satisfaction for pharmacists.²⁷ Research outside of pharmacy indicated that discrimination and perceived inequities in hiring practices, compensation, work assignments, and promotion negatively impact worker satisfaction.^28,29

Pharmacists' Patient-Care Services were assessed using a set of fifteen community pharmacy services indicators. Pharmacists were asked “from the list below, please indicate which services you have personally provided over the last 6 months.” The list included a) pharmacogenomics, b) disease management, c) opioid deprescribing, d) comprehensive medication management, e) point-of-care testing, f) medication reconciliation, g) medication therapy management, h) vaccine administration, i) patient medication assistance, j) medication synchronization, k) naloxone dispensing, l) special compounding, m) adherence packaging, n) durable medical equipment, and o) injection administration. Responses were binary (1 = yes and 0 = no).

Pharmacists' Compensation was assessed using a set of five community pharmacists' compensation indicators. Pharmacists were asked “did you have any of the following additional earnings during the past year from your primary employment?” The forms of additional earnings assessed were a) profit sharing, b) bonus/incentive pay, c) stock options, d) overtime. Responses were binary (1 = yes and 0 = no). Pharmacists were also asked “what was [your] percentage change in base pay from last year?”

Pharmacists' Professional Interaction refers to professional alliances and networks that community pharmacists maintain with other pharmacists and healthcare providers outside their place of primary employment. Pharmacists' professional interaction was assessed using three items; a) “have you had direct communication with a primary care practitioner in the last month that you worked?”, b) “is your practice site part of a Community Pharmacy Enhanced Services Network (CPESN)?”, and c) “do you manage patient medication therapy under a signed collaborative practice agreement?” Responses were collected on a binary scale (1 = yes and 0 = no).

No personal experience of workplace discrimination was assessed by asking pharmacists “have you personally experienced discrimination at your workplace?” Responses were binary (1 = no and 0 = yes).

2.3. Analysis

2.3.1. Descriptive analysis

The mean age of community pharmacists in the sample was determined. The sample distribution of pharmacists across gender and work setting categories and key variables were calculated. Correlations between the explanatory variables and job satisfaction were computed.

2.3.2. Principal component analyses

Separate principal component analyses (PCAs) of binary data were conducted on the 15 pharmacists' patient-care services indicators, the 5 pharmacists compensation indicators, and the 3 pharmacists professional interaction indicators. While the initial formulation of PCA was based on multivariate normality with continuous data, the requirement for data to have normal distribution limited the application of PCA to non-continuous data.³⁰ The generalized linear model framework has been used to extend the initial formulation of PCA for continuous data to binary data, discrete data, nominal data, and combinations of data types. For binary data, PCA uses the generalized linear model framework to make low-dimensional projections of the natural parameters (the logit of the probability of a value of one/yes/true) of the data from the Bernoulli distribution. Dimension reduction is achieved by a linear combination of the projected natural parameters represented by principal component scores.^31,32 In the present study, tetrachoric correlation matrices (binary variables) were used in the PCAs for pharmacists' patient-care services and pharmacists' professional interactions, while a point-biserial correlation matrix (binary and continuous variables) was used in the PCA for pharmacists' compensation. The scores for the extracted principal components from the separate PCAs were used in subsequent analyses.

The Kaiser-Meyer-Olkin (KMO) measure was used to determine whether the sampling of items to measure the underlying construct was adequate for PCA. A cut point of KMO values >0.50 was used to determine satisfactory sampling adequacy.³³ Bartlett's test of sphericity was used to determine that the indicators were sufficiently correlated for PCA. Separate initial PCAs were used to extract principal components equal to the original number of indicators for pharmacists' patient-care services, pharmacists' compensation, and pharmacists' professional interactions, respectively. The eigenvalues of these initial principal components were examined using a cut point of greater than or equal to 1, in combination with parallel analysis and visual inspection of scree plots, to determine how many principal components to extract in the final PCAs.^34,35 Principal component scores for each component extracted in the final PCAs for pharmacists' patient-care services, pharmacists' compensation, and pharmacists' professional interactions, respectively, were used in further analyses.

2.3.3. Multiple regression and relative importance analysis

First, pharmacists' age, gender, no personal experience of workplace discrimination, and the principal components extracted from the separate PCAs of pharmacists' patient-care services, pharmacists' compensation, and pharmacist professional interactions were entered into a multiple linear regression model to identify significant predictors of job satisfaction. The square root of variance inflation factor (√VIF) was used to assess the regression model for the presence of significant multicollinearity. Next, relative weight analysis was used to assess variable importance, and bootstrapped confidence intervals and tests of significance of relative weights were conducted using the procedure described by Johnson (2004) and Tonidandel et al. (2009).^18,36 Pairwise comparisons of the relative weights of components within pharmacists' patient-care services were done using bootstrapped confidence intervals around the difference between the relative weights of any two components being compared to evaluate statistical significance.¹⁸ Higher-order relative weights for pharmacists' patient-care services, obtained by specifying that the subdimensions be combined into the higher-order dimension in the analysis procedure, was used in subgroup analyses. Subgroup analyses were done to compare the relative importance of predictors in the model across pharmacists' age, gender, and work setting categories.

Subgroups were created by dividing the sample of pharmacists by age (younger, middle-aged, and older), gender (men and women), and work setting categories (independent, small chain, large chain, mass merchandizer, and health system retail). Relative weight analysis was done for each subgroup. Statistical difference in relative weights between groups was established when the confidence intervals of predictor relative weights for subgroups compared did not overlap. Contrasts were also used in the subgroup relative weights analysis to determine whether there were significant within-group differences in the relative importance of predictors for each age, gender, and work setting category. All statistical analyses were done using R v4.2.1. Principal component analysis was done using the psych package for R,³⁷ and relative weights analysis was done using the relaimpo package for R.³⁸

3. Results

3.1. Descriptive analysis

94,803 confirmed unique pharmacists' email addresses received the survey. 5467 responses were obtained to give an overall response rate of 5.8% for the NPWS. 2162 respondents identified as community pharmacists, and all 2162 community pharmacist respondents were included in the present analyses as missing data across the key demographic and outcome variables were negligible.³⁹ The mean age for community pharmacists was 44 years (SD = ±13), and a significantly higher proportion (p < 0.05) of study participants were of middle-age (36–54 years). The sample distribution across pharmacists' age groups, gender, and work setting categories are shown in Table 1. There were significant correlations between job satisfaction and its determinants that were assessed in this study (Table 2).

Table 1.

Sample distribution of community pharmacists (n = 2162).

Sample Characteristics	n (%)	Job Satisfaction Mean (SD)
Age (years)
≤35 (Younger Pharmacists)	793 (36.7)	2.27 (0.87)
36–54 (Middle-aged Pharmacists)1	850 (39.3)	2.16 (0.89)
≥55 (Older Pharmacists)	519 (24.0)	2.44 (0.96)
		⁎p < 0.05
Gender
Men	782 (36.2)	2.36 (0.94)
Women2	1374 (63.6)	2.21 (0.89)
Non-binary3	6 (0.3)	–
		⁎p < 0.05
Work Setting
Independent	400 (18.5)	2.93 (0.87)
Small Chain	63 (2.9)	2.78 (0.94)
Large Chain	942 (43.6)	1.97 (0.79)
Mass Merchandizer	381 (17.6)	2.16 (0.83)
Supermarket	320 (14.8)	2.24 (0.85)
Health System Retail	56 (2.6)	2.73 (0.96)
		⁎p < 0.05
Job Satisfaction
Dissatisfied	1273 (58.9)
Satisfied	888 (41.1)
Have you personally experienced discrimination at your workplace (No_Dis)?
Yes	753 (34.8)
No	1408 (65.2)

^⁎p < 0.05 indicates a statistically significant difference in group means.

¹Age group proportion is significantly greater than the proportion of other age groups:

²Proportion of women is significantly greater than the proportion of men:

³Excluded from regression and relative weight analyses due to low sample size.

Table 2.

Tetrachoric and point-biserial correlations between community pharmacists' job satisfaction and its determinants.

	Age2	Gender1	No_Dis1	Pharmacists' Professional Interaction2,3	Pharmacists' Compensation2,3	Medication management Services2,3	Advanced Dispensing Activities2,3	Other Services2,3
Gender1	−0.199⁎⁎
No_Dis1	0.012	−0.113
Pharmacists' Professional Interaction2,3	0.022	−0.008	−0.052⁎
Pharmacists' Compensation2,3	−0.168⁎⁎	0.064⁎⁎	−0.006	0.006
Medication mgt Services2,3	−0.002	−0.036	−0.010⁎⁎	0.347⁎⁎	0.064!
Adv. Dispensing Activities2,3	−0.194⁎⁎	0.058⁎	−0.083⁎⁎	0.084⁎⁎	0.213⁎⁎	0.147⁎⁎
Other Services2,3	0.019	−0.058⁎	0.026	0.178⁎⁎	−0.151⁎⁎	0.109⁎⁎	0.006
Job Satisfaction2	0.095⁎⁎	−0.075⁎⁎	0.250⁎⁎	0.086⁎⁎	−0.033	0.069⁎⁎	−0.165⁎⁎	0.129⁎⁎

** p < 0.001, * p < 0.05; ^!P = 0.05.

No_Dis: No personal experience of workplace discrimination.

Tetrachoric: correlations between binary variables; Point-Biserial: correlations between binary and continuous variables.

¹Binary variables; ²Continuous variables; ³Principal components extracted from separate PCAs of pharmacists' work characteristics with scores on a continuous scale.

³See Appendix A for original set of indicators from which principal components were extracted.

3.2. Principal component analyses (PCAs)

The KMO value for the pharmacists' patient-care services PCA was 0.76, indicating sampling adequacy, and the corresponding Bartlett's test of sphericity was significant (χ²₍₁₀₅₎ = 3244.15, p < 0.001), indicating sufficient variables correlation for PCA. Three dimensions of pharmacists' patient-care services were extracted from the PCA through parallel analysis, Eigen values, and visual inspection of scree plot. The three dimensions identified were medication management services, advanced dispensing activities, and other services. Items (a) to (g) were categorized as medication management services, items (h) to (k) were categorized as advanced dispensing activities, while items (l) to (n) were categorized as other services. Item (o) - injection administration- was excluded from all three categories due to having a factor loading below the cut-point of 0.40.³⁴ The factor loadings and the grouping of pharmacists' patient-care services indicators under their respective principal components are shown in Appendix A.

The KMO value for the pharmacists' compensation PCA was 0.53, indicating sampling adequacy, and the corresponding Bartlett's test of sphericity was significant (χ²₍₁₀₎ = 352.98, p < 0.001), indicating sufficient variables correlation for PCA. One dimension of pharmacists' compensation was identified from the PCA through parallel analysis, Eigen values, and visual inspection of scree plot. The factor loadings of the different modes of compensation assessed are shown in Appendix A. Percent change is base pay was excluded from subsequent analyses due to having a factor loading below the cut-point of 0.40.³⁴

The KMO value for the pharmacists' professional interactions PCA was 0.54, indicating sampling adequacy, and the corresponding Bartlett's test of sphericity was significant (χ²₍₃₎ = 206.74, p < 0.001), indicating sufficient variables correlation for PCA. One dimension of pharmacists' professional interactions was identified from the PCA through parallel analysis, Eigen values, and visual inspection of scree plot. The factor loadings of the different forms of professional interaction assessed are shown in Appendix A.

3.3. Multiple regression and relative importance analysis

No significant multicollinearity was observed between the explanatory variables to warrant excluding any variable from the model as the square root of the variance inflation factor (√VIF) values for the explanatory variables were <2 (range: 1.018 to 1.083).⁴⁰ However, significant correlations between the explanatory and the dependent variables were observed to justify the use of relative weight analysis (Table 2). The regression coefficients and ranks, the raw and rescaled relative weights, and the relative importance rank of each predictor are shown in Table 3. Based on regression coefficients, all the explanatory variables except pharmacists' compensation were significant predictors of job satisfaction (p < 0.05). However, based on relative weights, all the explanatory variables included in the model were significantly relatively important predictors of job satisfaction as the confidence intervals around their respective relative weights (Raw RW) did not include zero. Compared to younger community pharmacists (≤35 years), middle-aged pharmacists (36–54 years) reported lower job satisfaction while older pharmacists (≥55 years) reported higher job satisfaction. Advanced dispensing activities was negatively associated with community pharmacists' job satisfaction. The explanatory variables in the model collectively predicted 12.6% of the criterion variance (R²) of community pharmacists' job satisfaction. Also note that the predicted criterion variance (R²) for the regression is the same as the total relative weights (ΣRawRW). The most important predictor of pharmacists' job satisfaction was no personal experience of workplace discrimination (RI rank = 1) which accounted for as much as 48% of the predicted criterion variance (R²). This was followed by advanced dispensing activities (RI rank = 2) which accounted for as much as 18% of the predicted criterion variance (R²). Pharmacists' compensation (RI rank = 6) ranked the lowest in terms of variable importance accounting for 0.5% of the predicted criterion variance (R²).

Table 3.

Multiple linear regression (MR) and Relative weights analysis (RWA) of the relationship between pharmacists' age, gender, workplace characteristics and job satisfaction.

Explanatory Variables	β̂	Raw RW [95% CI]	Rescaled RW (% of R2 explained)	RI Rank
1Age (years)		0.0130 [0.0057, 0.0226]	10.4	3
36–54 [middle-aged pharmacists]	−0.149**
≥ 55 [older pharmacists]	0.104*
2Gender		0.0025 [0.0004, 0.0086]	1.9	5
Women	−0.024
No personal experience of discrimination	0.476**	0.0613 [0.0473, 0.0837]	48.8	1
Pharmacists' Patient-care Services			33.3!	2
Medication management services	0.088**	0.0056 [0.0007, 0.0147]	(4.4)
Advanced dispensing activities	−0.160**	0.0235 [0.0127, 0.0414]	(18.7)
Other services	0.104**	0.0129 [0.0058, 0.0291]	(10.2)
Pharmacist’ Compensation	0.017	0.0005 [0.0003, 0.0025]	0.5	6
Pharmacists Professional Interactions	0.074**	0.0065 [0.0020, 0.0138]	5.1	4
	R2 = 12.6%	ΣRawRW = 0.1258 [12.6%]	Σ = 100%

** p < 0.001, * P < 0.05; RW = Relative weight; Rescaled RW = [(Raw RW/ΣRawRW) × 100].

RI Rank = Rank of predictors in terms of relative importance [1 (most important) – 6 (least important)].

¹Reference age category: younger pharmacists [≤ 35 (years)].

²Reference gender category: men.

^!Sum of rescaled RW for medication management services (4.4), advanced dispensing activities (18.7), and other services (10.2). See Appendix A for full list of items.

Pairwise comparisons showed that the relative weight of medication management services was significantly lower than the relative weight of advanced dispensing activities (Δ_RW −0.0179 [−0.0327, −0.0092]) in explaining community pharmacists' job satisfaction. There was no statistically significant difference in relative weights between other services and either advanced dispensing activities or medication management services (p > 0.05). For model parsimony and to allow for simpler between- and within-group comparisons, we used the higher-order dimension of pharmacists' patient-care services, comprising medication management activities, advanced dispensing activities, and other services (Appendix A), in subgroup analyses.

As reported in Table 4, subgroup relative weights analyses showed that no personal experience of workplace discrimination, pharmacists' patient-care services (PPS), pharmacists' compensation (PC), and pharmacists' professional interactions (PPI) were consistently significantly important predictors of job satisfaction across age, gender, and work setting categories as the confidence intervals around their relative weights did not include zero. Between-group comparisons of relative importance ranks showed that no personal experience of discrimination was the most important predictor (RI rank = 1) of job satisfaction across all age groups, in women, and across community pharmacists work settings except small chain pharmacies. Pharmacists' patient-care services (PPS) was the most important predictor (RI rank = 1) of job satisfaction in men and small chain pharmacists (Table 4). However, there were no statistically significant between-group differences in the relative importance of predictors across pharmacists' age, gender, and work setting categories as the confidence intervals around their respective relative weights overlapped (Table 4). Variations in the relative importance weights of predictors across pharmacists' age, gender, and work setting categories are graphically illustrated in Appendix B to show how community pharmacists evaluating their job satisfaction may perceive the importance of multiple correlated determinants differentially.

Table 4.

Rescaled relative weights and relative importance ranks of predictors of pharmacists' job satisfaction across age, gender, and work setting categories.

	PC rRW [95% CI], (RI Rank)	PPS rRW [95% CI], (RI Rank)	PPI rRW [95% CI], (RI Rank)	No_Dis rRW [95% CI], (RI Rank)
Age
≤35 (Younger)	0.0579 [0.0008, 0.1315], (4)	0.3314 [0.1712, 0.6004], (2)	0.2016 [0.0390, 0.3372], (3)	0.4091 [0.2205, 0.6321], (1)
36–54 (Middle-aged)	0.0054 [0.0025, 0.0531], (4)	0.3984 [0.2343, 0.5837], (2)	0.0283 [0.0010, 0.0867], (3)	0.5679 [0.3699, 0.7274], (1)
≥55 (Older)	0.0529 [0.0084, 0.1568], (3)	0.4056 [0.2402, 0.5446], (2)	0.0347 [0.0056, 0.1175], (4)	0.5068 [0.3654, 0.6774], (1)
Gender
Men	0.0079 [0.0028, 0.1171], (4)	0.4629 [0.2968, 0.6961], (1)	0.0763 [0.0118, 0.2384], (3)	0.4529 [0.2546, 0.6390], (2)
Women	0.0042 [0.0022, 0.0480], (4)	0.3442 [0.2202, 0.4977], (2)	0.0444 [0.0054, 0.1282], (3)	0.6072 [0.4560, 0.7235], (1)
Work Setting
Independent	0.0011 [0.0004, 0.2247], (4)	0.1771 [0.0669, 0.5680], (2)	0.0351 [0.0018,0.1472], (3)	0.7867 [0.4045, 0.9063], (1)
Small Chain	0.0957 [0.0050, 0.4105], (3)	0.6078 [0.1037, 0.9555], (1)	0.0231 [0.0015, 0.2347], (4)	0.2734 [0.0068, 0.7329], (2)
Large Chain	0.3195 [0.0374, 0.6127], (2)	0.0999 [0.0167, 0.3150], (3)	0.0295 [0.0023, 0.1619], (4)	0.5510 [0.1332, 0.7977], (1)
Mass Merchandizer	0.1888 [0.0290, 0.5224], (2)	0.1637 [0.0554, 0.5203], (3)	0.0574 [0.0057, 0.1983], (4)	0.5902 [0.2080, 0.7669], (1)
Supermarket	0.0836 [0.0023, 0.3002], (3)	0.0889 [0.0335, 0.4177], (2)	0.0053 [0.0020, 0.1344], (4)	0.8222 [0.4569, 0.9090], (1)
Health System Retail	0.0158 [0.0064, 0.3184], (4)	0.4348 [0.0633, 0.9350], (2)	0.0232 [0.0064, 0.4078], (3)	0.5263 [0.0230, 0.8139] (1)

PC = Pharmacists' compensation; PPS = Pharmacists' Patient-care services [medication management services, advanced dispensing activities, and other services]; PPI = Pharmacists Professional Interactions; No_Dis = No personal experience of workplace discrimination; rRW = Rescaled Relative weight [(Raw RW/ΣRawRW)]; RI Rank = Relative importance rank of variables [1 (most important) – 4 (least important)].

As reported in Table 5, within-group pairwise comparisons of the relative weights of predictors showed that no personal experience of workplace discrimination and pharmacists' patient-care services were significantly more relatively important than pharmacists' compensation for job satisfaction across community pharmacists' age and gender categories. With the exception of younger pharmacists and pharmacists in health systems retail pharmacies, no personal experience of workplace discrimination was significantly more relatively important than professional interactions for predicting job satisfaction across subgroups. The relative importance of pharmacists' patient-care services was significantly higher than that of pharmacists' compensation across pharmacists age and gender categories, and among pharmacists in independent and health system retail pharmacies. Only among pharmacists in large chain pharmacies was a pharmacists' compensation significantly more important than professional interactions for predicting job satisfaction (Table 5).

Table 5.

Within-group pairwise comparisons of the relative importance weights of predictors of pharmacists' job satisfaction.

Pharmacists' Characteristics	Contrasts [Xa – Xb]	Difference in rRW [95% CI]
Age
≤35 (Younger Pharmacists)	No_Dis - PC	0.3512 [0.0877, 0.5823]
	PPS – PC	0.2734 [0.0643, 0.4928]
36–54 (Middle-aged Pharmacists)	No_Dis – PC	0.5625 [0.3152, 0.7481]
	No_Dis – PPI	0.5397 [0.2679, 0.7452]
	PPS - PC	0.3930 [0.1999, 0.5945]
	PPS – PPI	0.3702 [0.1756, 0.5804]
≥55 (Older Pharmacists)	No_Dis – PPI	0.4720 [0.2411, 0.6290]
	No_Dis – PC	0.4539 [0.2344, 0.6107]
	PPS – PPI	0.3709 [0.2021, 0.5288]
	PPS – PC	0.3527 [0.1770, 0.5409]
Gender
Men	No_Dis – PPI	0.3766 [0.0601, 0.5913]
	No_Dis – PC	0.4450 [0.2576, 0.6321]
	PPS – PPI	0.3867 [0.1042, 0.5785]
	PPS – PC	0.4550 [0.2508, 0.6030]
Women	No_Dis – PPI	0.5628 [0.2851, 0.7276]
	No_Dis – PC	0.6030 [0.3549, 0.7440]
	PPS – PPI	0.2998 [0.1249, 0.5254]
	PPS – PC	0.3400 [0.1827, 0.5325]
Work Setting
Independent	No_Dis - PPI	0.7517 [0.1891, 0.9296]
	No_Dis - PC	0.7857 [0.3117, 0.9399]
	PPS - PC	0.1761 [0.0017, 0.5980]
Large Chain	No_Dis – PPI	0.5215 [0.0024, 0.7934]
	PC – PPI	0.2900 [0.0637, 0.5360]
Small Chain	PPS – PPI	0.5847 [0.0617, 0.9177]
Mass Merchandizer	No_Dis – PPI	0.5328 [0.0887, 0.7478]
	No_Dis – PPS	0.4265 [0.0539, 0.6931]
Supermarket	No_Dis – PPI	0.8169 [0.3428, 0.9064]
	No_Dis – PPS	0.7334 [0.0084, 0.8277]
Health System Retail	PPS – PC	0.4190 [0.0444, 0.8303]

PC = Pharmacists' compensation;

PPS = Pharmacists' Patient-care services [medication management services, advanced dispensing activities, and other services];

PPI = Pharmacists Professional Interactions; No_Dis = No personal experience of workplace discrimination; rRW = Rescaled relative weights;

Contrasts Xa – Xb assess whether there is a difference in the relative importance of Xa and Xb such that negative values indicate that Xa is less important than Xb, while positive values indicate that Xa is more important than Xb.

Only statistically significant contrasts are reported, that is, confidence intervals around the value of Xa – Xb do not include zero.

4. Discussion

The present study showed how relative weight analysis can be used to complement multiple regression to estimate the predictive ability of correlated determinants of job satisfaction in community pharmacists. In addition, relative weight analysis provided a more contextually relevant description of the construct relations between job satisfaction and its determinants for community pharmacists.

In terms of both the multiple regression and the relative importance analyses (Table 3), no personal experience of workplace discrimination made the largest positive contribution by itself and in combination with other predictors to predicting job satisfaction in community pharmacists. Workplace discrimination refers to the prejudiced treatment of people or people groups based on characteristics like age, appearance, disabilities, race, religion, language, nationality, and gender.^41,42 In pharmacy practice, pharmacists may experience discrimination from supervisors, peers, and even patients.^43,44 Regardless of how and from whom discrimination is experienced at work, workplace discrimination has negative implications for pharmacists' well-being and job satisfaction,^43,45 and efforts should be made to eliminate discrimination from the workplace.⁴⁶

Exploring this finding further using subgroup analyses provided additional insight. We found evidence that the relative importance of not personally experiencing workplace discrimination as a predictor of job satisfaction varied across pharmacists' work-setting and gender categories. No personal experience of workplace discrimination ranked higher in relative importance for job satisfaction in women pharmacists than in men pharmacists and ranked as the most important determinant of job satisfaction across most pharmacy work settings. This finding is consistent with other studies that have shown that women are more likely to personally experience workplace discrimination than men in healthcare. Workplace gender discrimination against women occurs in various forms like sexist remarks and unwanted sexual attention,^47,48 lower performance evaluations and delayed career advancement,⁴⁹ or expectations that women should be responsible for planning events, catering for meals, and cleaning up common spaces at the workplace.⁵⁰ Although workplace discrimination is associated with negative affect and well-being, and an unfavorable assessment of job satisfaction for those on the receiving end,51., 52, 53 workplace discrimination persists due to the power imbalance that often exists between the perpetrators and the victims, which makes reporting or seeking redress for discrimination less likely.^51,54 Nonetheless, findings from this study indicate that no personal experience of workplace discrimination was the most relatively important predictor of community pharmacists' job satisfaction across age categories, work settings, except in small chain pharmacies and especially for women pharmacists, and therefore warrants more attention.⁴⁶

Pharmacists' patient-care services, which includes medication management activities, advanced dispensing activities, and other services (Appendix A), was the second most relatively important predictor of job satisfaction. Advanced dispensing activities accounted for at least half of the predictive ability of this group of predictors (Table 3). Interestingly, more engagement in advanced dispensing activities was associated with lower job satisfaction for community pharmacists. Although dispensing in pharmacy has been traditionally described as the filling of prescriptions, in the present study, advanced dispensing activities comprised vaccine administration, patient medication assistance, medication synchronization, and naloxone dispensing. The observed negative relationship between these newer advanced dispensing activities and community pharmacists' job satisfaction may be because pharmacists are overwhelmed and spend a significant portion of their time at work doing dispensing. So, while previous studies have reported pharmacists' desire for more clinically meaningful engagement with patients, these newer advanced dispensing activities may not be the meaningful work that pharmacists desire.^12,55 Also, adding these advanced dispensing tasks to an already high prescription volume without a commensurate increase in staffing, and training on how to perform new workflows and roles, is likely to increase pharmacists work stress and burnout.^10,56 When the effects of other predictors are accounted for, advanced dispensing activities contributed as much as 18% of the predicted criterion variance in community pharmacists' self-evaluation of their job satisfaction (Table 3). This suggests that pharmacists attributed a substantial degree of importance to considerations of task and skill variety and job meaningfulness when evaluating their job satisfaction.¹⁰ This finding shows that there is much to be gained in terms of enhancing job satisfaction by reducing how much medication dispensing activities pharmacists do and increasing work attributes that pharmacists find more satisfying.

Overall, pharmacists' professional interactions ranked higher than only pharmacists' compensation in relative importance for predicting job satisfaction (Table 3). Based on the relative importance rank of predictors for subgroups shown in Table 4, professional interactions consistently ranked higher than compensation as a determinant of job satisfaction among men and women pharmacists, but slight variations were observed across work settings. While professional interactions ranked higher than compensation among pharmacists working in independent and health system retail pharmacies, the converse was observed among pharmacists in chain, mass merchandizers, and supermarket pharmacies. Similarly, although pharmacists' compensation ranked the lowest in terms of relative importance for predicting job satisfaction over the entire sample of community pharmacists in this study (Table 3), subgroup analysis showed that the importance of compensation for pharmacists' job satisfaction varied depending on place of employment. Large chain and mass merchandizer pharmacies were the only work settings where pharmacists' compensation ranked higher in relative importance than pharmacists' patient-care services for predicting job satisfaction (Table 4).

The different relative importance ranks of predictors observed when assessed for the entire sample compared to subgroups could be due to the unique characteristics of demographic groups and the peculiarities of work across community pharmacy settings. For instance, the operating model of most chain pharmacies emphasizes efficiency and prioritizes profit maximization. In an industry where dispensing still accounts for a substantial share of pharmacy revenue, and reimbursements for patient care services are highly variable or not available, the profit maximization operating model of chain pharmacies tends to provide less incentive for pharmacy employers and pharmacists to perform clinically meaningful engagements with patients.^57,58 Although pharmacists face clinical care decisions every time they engage in dispensing, the need for efficiency creates time constraint such that pharmacists have to explicitly or implicitly choose between clinical care services like medication therapy counseling or simply dispensing. This situation is further complicated when pharmacists are compensated for efficiency with bonuses, or by the pharmacy employer extending participation in profit sharing schemes to pharmacists.^59,60 Even if pharmacists are not directly rewarded for efficiency through extra compensation, employers' emphasis on efficiency and profits that is pervasive in large chain pharmacies may compel pharmacists to work at a fast pace to meet company targets or to secure their jobs.⁶⁰ Taken together, the potential for extra income for being the “efficient” worker, or the threat of job loss for underperforming, and the associated financial strain, may make pharmacists focus on meeting company sales or dispensing targets at the expense of their professional obligations of clinical care, as high compensation is a way for pharmacists to resolve the dissonance and make sense of the high work stress.⁵⁷

4.1. Implications for pharmacy practice and research

The study finding that no personal experience of workplace discrimination was the most important predictor of job satisfaction in community pharmacists makes a unique contribution of identifying workplace discrimination as an important aspect of pharmacy workplace evaluations that has been little explored in community pharmacy research. Increasing investment by pharmacy stakeholders to address workplace discrimination and working to promote practice settings that are inclusive and equitable may have important implications for pharmacists' job satisfaction.

A major advantage of relative importance analysis is that it provides estimates that serve as intuitive and easily interpretable heuristics for making sense of the determinants of a criterion, especially when multiple correlated variables are involved. Concerning job satisfaction, findings from the present study may help pharmacy practice managers to understand how the influence of work-related factors are perceived differentially by pharmacists. This insight may aid pharmacy practice managers in designing interventions to improve job satisfaction for different demographic groups of pharmacists employed in their organizations. Recognizing that resources are limited, and that pharmacy practice managers may be interested in knowing which job aspects to prioritize for attention, guiding empirical evidence in the form of pairwise comparisons of predictors that show the differences in the relative importance of predictors that are not due to chance are provided (Table 5). Overall, relative importance analysis provides useful insights for decision making, and has the potential to move the perimeters of pharmacy research closer to practical relevance. It affords pharmacy researchers the capability to deal with the issue of multiple correlated predictors that is pervasive in pharmacy practice research, thus allowing research designs to more closely approximate reality.

5. Limitations and future research

The findings of this study should be interpreted within the context of its limitations. Data for this study were obtained in 2019 before the COVID-19 pandemic. During and following the pandemic, the nature of pharmacists' work has changed. Also, how pharmacists evaluate their work in light of their career expectations may have changed with the rise of virtual and remote work. Hence, future research should replicate this study to account for the influence of the COVID-19 pandemic on pharmacists' job satisfaction judgments. While differences in relative weights across pharmacists' age, gender, and work setting categories were not statistically significant in the present analyses, note that the emphasis in relative importance analysis is less on statistical significance but more on being able to estimate more accurately and realistically the contribution of explanatory variables to predicting criterion variance, especially when multiple variables are correlated. Moreover, the small size of observed relative weights was due to the small yet significant correlations between the variables. Larger relative weights can be expected as the correlations between variables become larger. Another limitation was that the explanatory variables included in the present study explained only 12.6% of the variance in community pharmacists' job satisfaction. Although the evidence in literature suggests that other factors like workload, staffing levels and role conflict may be salient determinants of job satisfaction, the selection of explanatory variables in the present study included predictors like no personal experience of workplace discrimination and pharmacists' professional interactions that have been less studied in pharmacy practice research. Moreover, the study design represents perhaps the first time this combination of variables were studied in context. Nonetheless, we acknowledge that the relative importance findings in the present study are indeed based on only the variables included and should be interpreted as such. More studies are needed to explore the relative importance of other predictors in context. Future research should include qualitative and mixed methods investigations to assess the underlying mechanisms of how pharmacists differentially perceive the factors affecting their job satisfaction.

6. Conclusion

Multiple personal and workplace variables influence pharmacists' job satisfaction. Not personally experiencing workplace discrimination had a substantial impact on job satisfaction for pharmacists. The addition of newer advanced dispensing activities to community pharmacy workflow did not yield higher job satisfaction for pharmacists. The relative importance of personal and workplace characteristics in how pharmacists evaluate their job satisfaction varied across age, gender and work setting categories as these variables are often conceptually and empirically related. While multiple regression analysis provides estimates of the unique contribution of variables to predicting job satisfaction, regression weights can be biased when multiple predictors are correlated. Relative importance analysis, which can be used as a supplement to multiple regression, addresses this issue by allowing the pharmacy manager or researcher to determine the proportionate contribution of a predictor to explaining the variance of an outcome after correcting for the intercorrelations among predictors.

Funding

This study was supported by the Pharmacy Workforce Center, Inc.

CRediT authorship contribution statement

Olajide O. Fadare: Conceptualization, Data curation, Methodology, Formal analysis, Writing – original draft. Matthew J. Witry: Conceptualization, Writing – review & editing, Supervision. Caroline A. Gaither: Methodology, Writing – review & editing. William R. Doucette: Writing – review & editing. Jon C. Schommer: Writing – review & editing.

Declaration of Competing Interest

None.

Appendix A

Results of principal component analysis showing dimension reduction of pharmacists' work characteristics.

	Work Characteristics		Factor Loadings
	Pharmacists Patient-care Services		Medication Management Services	Advanced Dispensing Activities	Other Services
(a)	Pharmacogenomics		0.74	−0.16	0.09
(b)	Disease Management		0.74	0.06	0.07
(c)	Opioid deprescribing		0.71	−0.06	−0.20
(d)	Comprehensive Medication Management		0.66	0.15	0.20
(e)	Point-of-Care Testing (POCT)		0.57	0.22	−0.42
(f)	Medication Reconciliation		0.57	0.02	0.32
(g)	Medication Therapy Management		0.46	0.33	0.08
(h)	Vaccine Administration		0.00	0.89	−0.16
(i)	Patient Medication Assistance		0.04	0.64	0.22
(j)	Medication Synchronization		−0.02	0.60	0.52
(k)	Naloxone Dispensing		0.28	0.51	−0.28
(l)	Special Compounding		0.14	−0.10	0.66
(m)	Adherence Packaging		0.36	−0.20	0.56
(n)	Durable Medical Equipment		0.11	0.27	0.45
(o)	⁎Injection Administration		0.23	0.30	0.31
	Pharmacists' Compensation		Factor Loadings
	Bonuses and Incentives		0.62
	Profit sharing		0.61
	Stock options		0.82
	Overtime pay		0.49
	!Percent change in base pay		0.19
	Pharmacists Professional Interactions		Factor Loadings
	Collaborative practice agreement		0.83
	CPESN membership		0.80
	Direct communication with primary care providers		0.60

The sign of factor loadings indicates the direction of relationship between indicators and principal components and not the magnitude of the relationship

^⁎injection administration not assigned to any principal component due to factor loading <0.40; PC = Principal component.

^!Percent change in base pay excluded from subsequent analysis due to factor loading <0.40.

Appendix B

Variations in the relative importance of job satisfaction predictors across community pharmacists' age, gender, and work setting categories.

[Legend: Rx = Pharmacists, PC = Pharmacists' Compensation, PPI = Pharmacists Professional Interactions, PPS=Pharmacists Patient-care Services, No_Dis = No personal experience of discrimination].