SAY, STAY, STRIVE: An Ordered Logistic Regression Analysis of Human Resource Management Practices and Employee Engagement

Jeden O. Tolentino De La Salle University

Say, Stay, Strive

2

Abstract

Using ordered logistic regression via maximum likelihood estimation, this study examined the relationship between the human resource management practices of firms and the level of engagement of their employees. More than 1,000 employees from 13 different companies belonging to the BusinessWorld Top 1000 Corporations in the Philippines responded to an Employee Opinion Survey developed by Edralin (2007), which was based on the best employer characteristics and attitudes identified by Hewitt Associates. Estimates were generated for three models, each representing the three elements of employee engagement (SAY, STAY, and STRIVE). Results indicate that among eight human resource management functions, the communication of business strategies, compensation, and training and development were the best predictors of employee engagement. Managerial implications were then discussed.

Say, Stay, Strive

3

SAY, STAY, STRIVE: An Ordered Logistic Regression Analysis of Human Resource Management Practices and Employee Engagement

Organizational behavior rests on an interdisciplinary foundation of fundamental concepts about the nature of people and organizations. One of the basic approaches of organizational behavior is the human resources approach, wherein employee growth and development are encouraged and supported (Newstrom, 2007). Ensuring such growth and development is one of the major roles of the leaders of an organization. This role is facilitated by the process of human resource management, which involves attracting, developing, and maintaining a talented and energetic workforce. The basic goal of human resource management is to build organizational performance capacity by raising human capital, to ensure that highly capable and enthusiastic people are always available (Schermerhorn, 2005). Firms who are able to effectively fulfill this role are often considered “best employers.” Being a best employer requires a holistic approach to building a work environment in which employees are constantly engaged and committed to business success. These firms employ leaders who are passionate about people, who are clear and concise when communicating strategic direction, and who possess a sound ability to execute on their people practices (Looi, Marusarz, & Baumruk, 2004).

This study aims to examine the relationship between the human resource management practices of a firm and the level of engagement of its employees.

Say, Stay, Strive

4

Review of Related Literature

A key challenge for managers is dealing with employees who increasingly expect to have concern shown for their attitudes and feelings as well as to receive rewards. Effective behavioral management that continuously works to build a supportive human climate in an organization can help produce favorable attitudes (Newstrom, 2007). This study focuses on two specific organizational behavior concepts: human resource management and employee engagement.

Human Resource Management

According to Noe, Hollenbeck, Gerhart, and Wright (2004), human resource management is the policies, practices, and systems that influence employees’ behavior, attitudes and performance. They emphasized that there are several important human resource management practices: (1) analyzing work and designing jobs; (2) attracting potential employees; (3) choosing employees; (4) teaching employees how to perform their jobs and preparing them for the future; (5) evaluating their performance; (6) rewarding employees; (7) creating a positive work environment, and (8) supporting the organization’s strategy. Schermerhorn (2005) echoed these practices by listing the three major responsibilities of human resource management: (1) attracting a quality workforce, which involves human resource planning, and employee recruitment and selection; (2) developing a quality workforce, which involves employee orientation, training and

Say, Stay, Strive

5

development, and performance appraisal; and (3) maintaining a quality workforce, which involves career development, work/life balance, compensation and benefits, retention and turnover, and labor-management relations. DeCenzo and Robbins (2005) proposed that human resource management consists of four basic functions: (1) staffing; (2) training and development; (3) motivation; and (4) maintenance. In less academic terms, they said that human resource management is made up of four activities: (1) hiring people; (2) preparing them; (3) stimulating them; and (4) keeping them.

Employee Engagement

According to Newstrom (2007), employee attitudes are important to monitor, understand, and manage. They develop as the consequences of the feelings of equity or inequity in the reward system, as well as from supervisory treatment. Managers are particularly concerned with four types of attitudes – job satisfaction, job involvement, organizational commitment, and work mood. Combinations and variations of these four types of attitudes lead to employee engagement. Endres and Mancheno-Smoak (2008) described engaged employees as those who work with passion and feel a profound connection with their company. They drive innovation and move the organization forward. In contrast, not-engaged employees sleepwalk through their workday, putting time – but not energy and passion – into their work.

Say, Stay, Strive

6

Figure 1. The Hewitt engagement model.

In the Hewitt Engagement Model proposed by Hewitt Associates (2003) (see Figure 1), engagement is affected by several factors, and is defined as “the emotional and intellectual commitment employees have to their organization” (p. 11). Specifically, Hewitt Associates stated that engagement is when employees (1) consistently speak positively about the organization to co-workers, potential employees, and, most critically, customers (current and potential); (2) have an intense desire to remain with the organization; and (3) are committed, exert extra effort, and are engaged in work that contributes to business success. They summarized these three actions as “SAY,” “STAY,” and “STRIVE.”

Say, Stay, Strive

7

Linking Human Resource Management Practices to Employee Engagement

To examine the relationship between the human resource management practices of a firm and the level of engagement of its employees, several models from the literature were gathered. The important elements of these models will be used as the basis of analysis.

Figure 2. A conceptual model of best employer studies.

Figure 2 represents the conceptual model proposed by Joo and McLean (2006) to describe best employer studies. The model is based on the Hewitt Engagement Model and has eight propositions corresponding to eight relationships between five specific factors. Of particular interest in this study are propositions 2 (P2) and 3 (P3). P2 states that “[f]or employees to be engaged in organization and work, the business strategy should be communicated and shared with individual employees” (p. 21). P3, meanwhile, states that “[b]etter HR practices will have a positive impact on employee engagement” (p. 22).

Say, Stay, Strive

8

Figure 3. A schematic drawing of a best employer.

Looi, Marusarz, and Baumruk (2004), in a report commissioned by Hewitt Associates, described what best employers do to foster an environment that engages employees (see Figure 3). Although they found no specific formula, they discovered that in placing a high value on employees’ needs, best employers inspire people to do their best work, motivate them to stay with the company, and cause them to promote the company to their friends, family, and customers. Essentially, because best employers create an environment in which employees’ needs are met, employees are more engaged. This is a feat to which many companies, especially those that desire to survive in the long term, aspire.

Say, Stay, Strive

9

Figure 4. Characteristics of model employer firms

Figure 4 illustrates the relationships of the major variables that were investigated by Edralin (2007) in her study on model employers in the Philippines. This framework

Say, Stay, Strive

10

takes the human resource management functions with their corresponding indicators as the characteristics of the model employer firms. The benefits of being a model employer were also included in the model, one of which is engaged and committed employees.

Figure 5. The IES engagement model.

Finally, the engagement model proposed by Robinson, Perryman, and Hayday (2004) illustrates the strong link between feeling valued and involved and employee engagement (see Figure 5). Their findings suggest that many of the drivers of engagement will be common to all organizations, regardless of sector. However, some variability is likely, and the relative strength of each driver is also likely to be contingent upon the organization.

Say, Stay, Strive

11

Theoretical Framework

The framework of this study is presented in Figure 6. The independent variables in the framework are Communication of Business Strategies and seven human resource management

functions (Recruitment, Training

and

Development,

Performance

Management, Compensation, Benefits, Work/Life Balance, and Labor Relations). The dependent variables, on the other hand, are the three elements of employee engagement according to Hewitt Associates (2003): SAY, STAY, and STRIVE.

Figure 6. Operational framework of human resource management practices and employee engagement.

Say, Stay, Strive

12

Independent Variables

Communication of business strategies. In this study, good communication practices involve (1) the leaders of the firm communicating the company's business strategy; (2) the employees having a good understanding of the company's business goals and objectives; (3) the leaders utilizing every communication channel possible to help employees understand the company's direction; (4) the leaders providing frequent and continuous communication regarding their expectations on the employees; (5) the leaders consulting employees when major changes are made in the company; (6) the company employing a purpose, mission or vision statement with the organization's value proposition with which the employees can also identify; (7) the company having a clear cut communication flow between managers and subordinates; and (8) the company being open to criticism (Edralin, 2007).

Recruitment. In this study, good recruitment practices include selecting for fit, which involves (1) the leaders having highly selective recruiting programs; (2) the company recruiting the people based on the right fit; (3) the company recruiting people who share the same set of values and beliefs of the company. These practices also include better screening, which involves (1) the company keeping practices across the organization mostly consistent; and (2) the company finding new workers through referrals from existing employees (Edralin, 2007).

Say, Stay, Strive

13

Training and development. In this study, good training and development practices include training for soft skills in addition to technical skills, which involves (1) the company emphasizing learning and development for cultural behaviors and values than technical skills training; and (2) the employees being encouraged to take some responsibility for their own development (Edralin, 2007).

Performance management. In this study, good performance management practices include rigorous and effective performance management; which involves (1) the company preferring continual coaching rather than over-reliance on formal performance evaluation; (2) the managers providing constructive feedback on employee performance; and (3) the company providing an opportunity for employees to evaluate their managers and their peers (Edralin, 2007).

Compensation. In this study, good compensation practices include pay-forperformance, which involves (1) the company providing financial rewards other than salary; (2) the company giving cash incentives not only to recognize good performance but also to encourage employees; and (3) the company properly acknowledging and adequately compensating overtime. These practices also include incentive plans, which involve (1) the company being likely to offer incentive or variable pay; and (2) the company providing profit sharing programs (Edralin, 2007).

Benefits. In this study, good benefits practices include flexible benefits, which involve (1) the company offering flexible benefits that are tailored-fit to the diverse

Say, Stay, Strive

14

needs of the employees; and (2) the company offering high package fringe benefits that can be converted to cash. These practices also include convenience, which involves (1) the company offering on-site personal services such as ATMs, dry cleaners, banking services, etc.; (2) the company offering casual or business casual dress code; and (3) the company offering wellness programs for the employees (Edralin, 2007).

Work/life balance. In this study, good work/life balance practices include work flexibility, which involves (1) the company offering family welfare programs by providing activities that enhance working and family relationships; (2) the company offering sports programs; and (3) the company offering recreational programs (Edralin, 2007).

Labor relations. In this study, good labor relations practices include management support, which involves (1) the company recognizing the right of employees to self-organization; and (2) the company strictly following the law regarding labor relations. These practices also include harmonious dispute resolution, which involves (1) the company ensuring that due process and sanctions are extended to alleged employer-employee violators; (2) the company considering both formal and informal modes of grievance machinery as an avenue for consultation; (3) the management and the union closing the collective bargaining agreements based on reasonable compromise and mutual respect; and (4) the company advocating education on human rights, labor code, work attitudes, family, and health care in the collective bargaining agreement (Edralin, 2007).

Say, Stay, Strive

15

Dependent Variables: Elements of Employee Engagement

In this study, employee engagement involves (1) the employees consistently speaking positively about the organization to co-workers, potential employees and customers (SAY); (2) the employees having an intense desire to be a member of the organization (STAY); and (3) the employees exerting extra effort and engaging in behaviors that contribute to business success (STRIVE).

Empirical Limitations

This study focuses mainly on the human resource management practices of a firm and their effect on employee engagement. It was not able to capture the demographics of the respondents of the survey, particularly age, sex, and years spent with the company.

Methodology

1,003 employees from 13 different companies belonging to the BusinessWorld Top 1000 Corporations in the Philippines responded to an Employee Opinion Survey developed by Edralin (2007), which was based on the best employer characteristics and attitudes identified by Hewitt Associates. The survey questionnaire was accomplished by supervisory and rank-and-file employees.

Say, Stay, Strive

16

Using a five-point Likert scale, the respondents rated their level of agreement or disagreement on three statements: (1) the employees consistently speak positively about the organization to co-workers, potential employees and customers; (2) the employees have an intense desire to be a member of the organization; and (3) the employees exert extra effort and engage in behaviors that contribute to business success. Statement 1 is linked to the SAY element of employee engagement, Statement 2 is linked to the STAY element, and Statement 3 is linked to the STRIVE element. Then, the respondents rated the implementation of different human resource management practices by their respective organizations. The items on the questionnaire included aspects on (1) Communication of Business Strategies, (2) Recruitment, (3) Training and Development, (4) Performance Management, (5) Compensation, (6) Benefits, (7) Work/Life Balance, and (8) Labor Relations. Finally, to examine the link between the human resource management practices of a firm and the level of engagement of its employees, ordered logistic regression via maximum likelihood estimation was used.

Results and Discussion

Maximum likelihood estimation was run on the three models (SAY, STAY, STRIVE), and the results are presented in turn.

Say, Stay, Strive

17

Model 1: SAY

When asked whether they consistently speak positively about the organization to co-workers, potential employees, and customers, a fifth (18 percent) of the 1,003 respondents said they “strongly agree,” more than half (56 percent) said they “agree,” about a quarter (23 percent) were “neutral,” while the rest (three percent) said they either “disagree” or “strongly disagree.” Using this data and the employees’ ratings of the eight human resource management practices, maximum likelihood estimation yielded that following output:

Table 1 Estimation Output for the SAY Statement Dependent Variable: SAY Method: ML - Ordered Logit (Quadratic hill climbing) Included observations: 1003 Number of ordered indicator values: 5 Convergence achieved after 5 iterations Covariance matrix computed using second derivatives Coefficient Std. Error Communication 0.940343 0.248217 Recruitment 0.864995 0.207754 Training and Development 0.477193 0.146224 Performance Management 0.339376 0.163284 Compensation 0.678208 0.156884 Benefits -0.218945 0.201047 Work/Life Balance -0.102554 0.145446 Labor Relations 0.101935 0.196318 Limit Points LIMIT_2:C(9) 5.308301 0.726695 LIMIT_3:C(10) 7.497607 0.619215 LIMIT_4:C(11) 10.42308 0.651690 LIMIT_5:C(12) 13.86126 0.719112 Akaike info criterion 1.804809 Schwarz criterion Log likelihood -893.1115 Hannan-Quinn criter. Restr. log likelihood -1107.608 Avg. log likelihood LR statistic (8 df) 428.9930 LR index (Pseudo-R2) Probability(LR stat) 0.000000

z-Statistic 3.788396 4.163557 3.263439 2.078433 4.322987 -1.089024 -0.705099 0.519237

Prob. 0.0002*** 0.0000*** 0.0011*** 0.0377* 0.0000*** 0.2761 0.4807 0.6036

7.304712 12.10824 15.99393 19.27551

0.0000 0.0000 0.0000 0.0000 1.863561 1.827135 -0.890440 0.193657

Say, Stay, Strive

18

The Log likelihood of Model 1 (SAY) is -893.1115. This is used in the Likelihood Ratio (LR) Chi-Square test of whether all predictors’ regression coefficients in the model are simultaneously zero and in tests of nested models (UCLA Academic Technology Services, n.d.). The LR statistic obtained (with eight degrees of freedom) is 428.9930. The p-value of this LR stat is very low, which leads to the conclusion that at least one of the predictors' regression coefficient is not equal to zero in the model. The McFadden pseudo-R2 of the model is 0.1937.

Interpretation of the Logit Parameter Estimates

Of the eight human resource management practices, each of six variables have a positive effect on the ordered log-odds of being in a higher SAY response category while the other variables in the model are held constant. The other two variables (Benefits and Work/Life Balance), however, have a negative effect, which contradicts a priori expectations. Taking the absolute value, the variable with the highest ordered logodds regression coefficient is Communication (0.9403), followed by Recruitment (0.8650). Labor Relations, on the other hand, has the lowest ordered log-odds regression coefficient (0.1019). Using the Z-statistic, the coefficients of Benefits, Work/Life Balance, and Labor Relations are not significant, the coefficient of Performance Management is significant at the 0.05 level of significance, and the rest are significant at the 0.005 level of significance.

Say, Stay, Strive

19

Interpretation of the Ancillary Parameter Estimates

The limit points refer to the thresholds used to differentiate the adjacent levels of the response variable, which is, in this case, SAY. The coefficients of these ancillary parameters are the estimated cut points used to differentiate a response in the Likert scale from those responses higher in the scale when the values of the predictor variables are evaluated at zero (UCLA Academic Technology Services, n.d.). The respondents that have a value of 5.30 or less on the underlying latent variable that gave rise to the SAY variable would be classified as having said that they “strongly disagree” with the SAY statement given that they rated all eight human resource management practices zero. The respondents that have a value of 7.50 or less would be classified as having said that they “strongly disagree” or “disagree” with the SAY statement given that they rated all eight human resource management practices zero. The respondents that have a value of 10.42 or less would be classified as having said that they “strongly disagree” or “disagree” or having stayed “neutral” with the SAY statement given that they rated all eight human resource management practices zero. The respondents that have a value of 13.86 or less would be classified as having said that they “strongly disagree” or “disagree,” or having stayed “neutral,” or having said that they “agree” with the SAY statement given they rated all eight human resource management practices zero. All coefficients are statistically significant.

Say, Stay, Strive

20

Interpretation of the Odds Ratios

The proportional odds ratios for the ordered logit model can be obtained by the exponentiation of the ordered logit coefficients (i.e. ecoef.). The ordered logit model estimates a single equation over the levels of the dependent variable. If the change in levels are viewed in a cumulative sense and if the coefficients are interpreted in odds, comparisons are made between the respondents who are in groups greater than k and those who are in groups less than or equal to k, where k is the level of the response variable. The interpretation would be that for a one unit change in the predictor variable, the odds for respondents in a group that is greater than k versus less than or equal to k are the proportional odds times larger (UCLA Academic Technology Services, n.d.).

Table 2 Odds Ratios for Model 1: SAY Variable

Odds Ratio

Communication

2.560860

Recruitment

2.374994

Compensation

1.970344

Training and Development

1.611544

Performance Management

1.404071

Say, Stay, Strive

21

The variable with the highest proportional odds ratio is Communication (see Table 2). For a one unit increase in a respondent’s evaluation of Communication practices in a firm, the odds of strongly agreeing to the SAY statement versus the combined four other response categories are 2.56 times greater, given the other variables are held constant in the model. Likewise, for a one unit increase in a respondent’s evaluation of Communication practices in a firm, the odds of the combined “strongly agree” and “agree” responses to the SAY statement versus the combined three other response categories are 2.56 times greater, given the other variables are held constant. The variable with the next stronger influence is Recruitment (2.37).

Model 2: STAY

When asked whether they have an intense desire to be a member of the organization, a fifth (20 percent) of the 1,003 respondents said they “strongly agree,” about three-fifths (59 percent) said they “agree,” another fifth (19 percent) were “neutral,” while the rest of the respondents (two percent) said they either “disagree” or “strongly disagree.” Using this data and the employees’ ratings of the eight human resource management practices, maximum likelihood estimation yielded that following estimation output:

Say, Stay, Strive

22

Table 3 Estimation Output for the STAY Statement Dependent Variable: STAY Method: ML - Ordered Logit (Quadratic hill climbing) Included observations: 1003 Number of ordered indicator values: 5 Convergence achieved after 5 iterations Covariance matrix computed using second derivatives Coefficient Std. Error Communication 1.178454 0.256756 Recruitment 0.199019 0.213586 Training and Development 0.547139 0.150654 Performance Management 0.311657 0.173602 Compensation 0.502448 0.162127 Benefits 0.199650 0.206790 Work/Life Balance 0.488450 0.152091 Labor Relations 0.271185 0.197123 Limit Points LIMIT_2:C(9) 7.402324 0.729504 LIMIT_3:C(10) 8.983455 0.671423 LIMIT_4:C(11) 12.48404 0.721083 LIMIT_5:C(12) 16.41380 0.807573 Akaike info criterion 1.591241 Schwarz criterion Log likelihood -786.0072 Hannan-Quinn criter. Restr. log likelihood -1038.616 Avg. log likelihood LR statistic (8 df) 505.2182 LR index (Pseudo-R2) Probability(LR stat) 0.000000

z-Statistic 4.589789 0.931797 3.631765 1.795241 3.099106 0.965471 3.211558 1.375713

Prob. 0.0000*** 0.3514 0.0003*** 0.0726 0.0019*** 0.3343 0.0013*** 0.1689

10.14706 13.37973 17.31291 20.32485

0.0000 0.0000 0.0000 0.0000 1.649993 1.613568 -0.783656 0.243217

The Log likelihood of Model 2 (STAY) is -786.0072. The LR statistic obtained (with eight degrees of freedom) is 505.2182. The p-value of this LR stat is very low, which leads to the conclusion that at least one of the predictors' regression coefficient is not equal to zero in the model. The McFadden pseudo-R2 of the model is 0.2432.

Interpretation of the Logit Parameter Estimates

Each of the eight human resource management practices have a positive effect on the ordered log-odds of being in a higher STAY response category while the other

Say, Stay, Strive

23

variables in the model are held constant, confirming a priori expectations. The variable with the highest ordered log-odds regression coefficient is Communication (1.1785), followed by Training and Development (0.5471). Benefits, on the other hand, has the lowest ordered log-odds regression coefficient (0.1997). Using the Z-statistic, the coefficients of Recruitment, Performance Management, Benefits, and Labor Relations are not significant, while the rest are significant at the 0.005 level of significance.

Interpretation of the Ancillary Parameter Estimates

The respondents that have a value of 7.40 or less on the underlying latent variable that gave rise to the STAY variable would be classified as having said that they “strongly disagree” with the STAY statement given that they rated all eight HR management practices zero. The respondents that have a value of 8.98 or less would be classified as having said that they “strongly disagree” or “disagree” with the STAY statement given that they rated all eight human resource management practices zero. The respondents that have a value of 12.48 or less would be classified as having said that they “strongly disagree” or “disagree” or having stayed “neutral” with the STAY statement given that they rated all eight human resource management practices zero. The respondents that have a value of 16.41 or less would be classified as having said that they “strongly disagree” or “disagree,” or having stayed “neutral,” or having said that they “agree” with the STAY statement given they rated all eight human resource management practices zero. All coefficients are statistically significant.

Say, Stay, Strive

24

Interpretation of the Odds Ratios

Table 4 Odds Ratios for Model 2: STAY Variable

Odds Ratio

Communication

3.249347

Training and Development

1.728301

Compensation

1.652762

Work/Life Balance

1.629788

The variable with the highest proportional odds ratio is Communication (see Table 4). For a one unit increase in a respondent’s evaluation of Communication practices in a firm, the odds of strongly agreeing to the STAY statement versus the combined four other response categories are 3.25 times greater, given the other variables are held constant in the model. Likewise, for a one unit increase in a respondent’s evaluation of Communication practices in a firm, the odds of the combined “strongly agree” and “agree” responses to the STAY statement versus the combined three other response categories are 3.25 times greater, given the other variables are held constant. The variable with the next stronger influence is Training and Development with an odds-ratio of 1.73.

Say, Stay, Strive

25

Model 3: STRIVE

When asked whether they exert extra effort and engage in behaviors that contribute to business success, a quarter (24 percent) of the 1,003 respondents said they “strongly agree,” about three-fifths (58 percent) said they “agree,” a sixth (16 percent) were “neutral,” while the rest (two percent) said they either “disagree” or “strongly disagree.” Using this data and the employees’ ratings of the eight human resource management practices, maximum likelihood estimation yielded that following estimation output:

Table 5 Estimation Output for the STRIVE Statement Dependent Variable: STRIVE Method: ML - Ordered Logit (Quadratic hill climbing) Included observations: 1003 Number of ordered indicator values: 5 Convergence achieved after 5 iterations Covariance matrix computed using second derivatives Coefficient Std. Error Communication 1.367634 0.251999 Recruitment 0.259901 0.203988 Training and Development 0.499707 0.147543 Performance Management 0.166287 0.168139 Compensation 0.763926 0.160437 Benefits -0.239974 0.202415 Work/Life Balance 0.571278 0.148005 Labor Relations -0.095914 0.199322 Limit Points LIMIT_2:C(9) 5.123155 0.920191 LIMIT_3:C(10) 7.692394 0.641342 LIMIT_4:C(11) 10.74181 0.675071 LIMIT_5:C(12) 14.49075 0.752247 Akaike info criterion 1.647199 Schwarz criterion Log likelihood -814.0702 Hannan-Quinn criter. Restr. log likelihood -1047.317 Avg. log likelihood LR statistic (8 df) 466.4944 LR index (Pseudo-R2) Probability(LR stat) 0.000000

z-Statistic 5.427148 1.274100 3.386856 0.988987 4.761521 -1.185553 3.859865 -0.481204

Prob. 0.0000*** 0.2026 0.0007*** 0.3227 0.0000*** 0.2358 0.0001*** 0.6304

5.567493 11.99421 15.91213 19.26328

0.0000 0.0000 0.0000 0.0000 1.705952 1.669526 -0.811635 0.222709

Say, Stay, Strive

26

The Log likelihood of Model 3 (STRIVE) is -814.0702. The LR statistic obtained (with eight degrees of freedom) is 466.494. The p-value of this LR stat is very low, which leads to the conclusion that at least one of the predictors' regression coefficient is not equal to zero in the model. The McFadden pseudo-R2 of the model is 0.2227.

Interpretation of the Logit Parameter Estimates

Of the eight human resource management practices, each of six have a positive effect on the ordered log-odds of being in a higher STRIVE category while the other variables in the model are held constant. Meanwhile, Benefits and Labor Relations have a negative effect, which contradicts a priori expectations. Taking the absolute value, the variable with the highest ordered log-odds regression coefficient is Communication (1.3676), followed by Compensation (0.7639) and Work/Life Balance (0.5713). Labor Relations, on the other hand, has the lowest ordered log-odds regression coefficient (0.0959). Using the Z-statistic, the coefficients of Recruitment, Performance Management, Benefits, and Labor Relations are not significant, while the rest are significant at the 0.005 level of significance.

Interpretation of the Ancillary Parameter Estimates

The respondents that have a value of 5.12 or less on the underlying latent variable that gave rise to the STRIVE variable would be classified as having said that

Say, Stay, Strive

27

they “strongly disagree” with the STRIVE statement given that they rated all eight human resource management practices zero. The respondents that have a value of 7.69 or less would be classified as having said that they “strongly disagree” or “disagree” with the STRIVE statement given that they rated all eight human resource management practices zero. The respondents that have a value of 10.74 or less would be classified as having said that they “strongly disagree” or “disagreed” or having stayed “neutral” with the STRIVE statement given that they rated all eight human resource management practices zero. The respondents that have a value of 14.49 or less would be classified as having said that they “strongly disagree” or “disagree,” or having stayed “neutral,” or having said that they “agree” with the STRIVE statement given they rated all eight human resource management practices zero. All coefficients are significant.

Interpretation of the Odds Ratios

Table 6 Odds Ratios for Model 3: STRIVE Variable

Odds Ratio

Communication

3.926051

Compensation

2.146688

Work/Life Balance

1.770528

Training and Development

1.648238

Say, Stay, Strive

28

The variable with the highest proportional odds ratio is Communication (see Table 6). For a one unit increase in a respondent’s evaluation of Communication practices in a firm, the odds of strongly agreeing to the STRIVE statement versus the combined four other response categories are 3.93 times greater, given the other variables are held constant in the model. Likewise, for a one unit increase in a respondent’s evaluation of Communication practices in a firm, the odds of the combined “strongly agree” and “agree” responses to the STRIVE statement versus the combined three other response categories are 3.93 times greater, given the other variables are held constant. The variable with the next stronger influence is Compensation (2.15).

Table 7 Summary of Results

Variable

SIGN:

SIGNIFICANCE:

MAGNITUDE:

Regression Coefficient

Regression Coefficient

Odds Ratio (rank)

SAY

STAY

STRIVE

SAY

STAY

STRIVE

SAY

STAY

STRIVE

Comm

+

+

+

***

***

***

1

1

1

Re

+

+

+

***

TD

+

+

+

***

2

4

PM

+

+

+

*

Co

+

+

+

***

3

2

Be

–

+

–

WL

–

+

+

4

3

LR

–

+

–

2 ***

***

4 5

***

***

***

***

3

Say, Stay, Strive

29

Conclusion and Managerial Implications

Human resource management performs an integral and strategic function in managing firms. It is a key management tool since human resources are considered essential in the continued existence and success of any business (Edralin, 2007). The empirical results presented by the models provide interesting managerial implications in this regard.

Communication of Business Strategies

The variable Communication has a positive coefficient, is statistically significant, and has the highest odds ratio across all three models. This would imply that more than implementing the different human resource management practices, communication between management and employees is paramount for employee engagement. This also confirms Proposition 2 of Joo and McLean (2006), i.e. “for employees to be engaged in organization and work, the business strategy should be communicated and shared with individual employees.” The leaders of the organization should utilize every available communication channel to help their people understand the direction in which the company is heading.

Say, Stay, Strive

30

Compensation and Benefits

Next to Communication, Compensation is the variable with the biggest impact to employee engagement, having a positive coefficient and being statistically significant across all three models. However, Benefits seems to not have a significant impact on employee engagement at all. This might seem counterintuitive at first glance, but the theories about rewards have evolved along with the models of organizational behavior. According to Newstrom (2007), in the 19th and early 20th centuries, employees were presumed to want primarily money; therefore, money was believed to produce direct motivation. While money is still a social medium of exchange and most employees do respond to money as a reward, the custodial belief that money should be the managerial orientation and benefits should be the employee orientation has been successfully buried by the idea that economic rewards operated through the attitudes of workers in the social system to produce an indirect incentive. Interestingly, other rewards may even decrease worker incentive. Vroom’s expectancy theory (see Figure 10) might hold the explanation. First, employees might be indifferent or – theoretically – have a strong avoidance for certain rewards such that their motivation might decrease (valence). Second, they might not believe that their efforts will be successful in producing desired performance (expectancy). And third, they might not trust that rewards will follow better performance (instrumentality). This will be an interesting theory to test empirically.

Say, Stay, Strive

31

Figure 10. Vroom’s expectancy theory.

The key, therefore, may be offering a wide range of non-economic programs to supplement the organization’s pay program. Newstrom (2007) presented the “reward pyramid” (see Figure 11) to describe the makeup of a complete pay program.

Say, Stay, Strive

32

Figure 11. The reward pyramid: The makeup of a complete pay program.

Training and Development

Training and Development is a significant predictor across all three models, confirming the engagement model proposed by Robinson et al. (2004) where training and development was the most important factor. This could also help confirm the importance of self-efficacy in employee engagement. Newstrom (2007) defines selfefficacy as “an internal belief that one has the necessary capabilities and competencies to perform a task, fulfill role expectations, or meet a challenging situation successfully” (p. 463). The infusion of these capabilities and competencies are the primary objectives of the training and development practices of organizations.

Say, Stay, Strive

33

Aldag and Kuzuhara (2002) proposed four sources of self-efficacy and linked them to training and development. First, people may gain self-efficacy by actually mastering a task (enactive attainment). Second, seeing that others who are similar can master a task may enhance self-efficacy (vicarious experience). Third, employees may simply be convinced that they can master tasks through words of encouragement (verbal persuasion). And fourth, techniques that create emotional support or foster a supportive and trusting group atmosphere may reduce that states that lower selfefficacy (emotional arousal). Managers of firms would do well to integrate these sources of self-efficacy in their training and development program, both formal and informal.

Implications on the Other Human Resource Management Functions

While Work/Life Balance is not a significant predictor in the SAY model, it is highly significant in the STAY model and in the STRIVE model. On the other hand, Recruitment and Performance Management were significant predictors only in the SAY model. Finally, Labor Relations is not a significant predictor across all three models. These results might suggest that employee engagement is more linked to job itself and the employee himself/herself, rather than his/her colleagues or the organization. Nevertheless, all human resource management functions reinforce one another to build an engaged workforce, and none should be neglected due to the results of this study.

Say, Stay, Strive

34

References

Aldag, R. J., & Kuzuhara, L. W. (2002). Organizational behavior and management: An integrated skills approach. Cincinnati, OH: South-Western. DeCenzo, D. A., & Robbins, S. P. (2005). Fundamentals of human resource management (8th ed.). New York, NY: John Wiley & Sons. Edralin, D. M. (2007). Model employers in the Philippines. LCCM Research Journal, 17, 2, 58-77. Endres, G. M., & Mancheno-Smoak, L. (2008). The human resource craze: Human performance

improvement

and

employee

engagement.

Organization

Development Journal, 26, 1, 69-78. Hewitt Associates. (2003). Best employers in Asia 2003: Overview and findings. Retrieved February 1, 2008, from http://www.asria.org/events/hongkong/june03/index_html/lib/BestEmployersInAsi a2003.pdf Joo, B. K., & McLean, G. N. (2006). Best employer studies: A conceptual model from a literature review and a case study. Human Resource Development Review, 5, 228-257. Looi, P. W., Marusarz, T., & Baumruk, R. (2004). What makes a best employer? Insights and findings from Hewitt’s global best employers study. Retrieved February 1, 2008, from http://was7.hewitt.com/bestemployers/pdfs/BestEmployer.pdf

Say, Stay, Strive

35

Newstrom, J. W. (2007). Organizational behavior: Human behavior at work (12th ed.). New York, NY: McGraw-Hill/Irwin. Noe, R. A., Hollenbeck, J. R., Gerhart, B., & Wright, P. M. (2004). Fundamentals of human resource management. New York, NY: McGraw-Hill/Irwin. Robinson, D., Perryman, S., & Hayday, S. (2004). The drivers of employee engagement. Retrieved February 1, 2008, from http://www.employmentstudies.co.uk/summary/summary.php?id=408 Schermerhorn, J. R. (2005). Management (8th ed.). New York, NY: John Wiley & Sons. UCLA Academic Technology Services (n.d.) Stata annotated output: Ordered logistic regression. Retrieved February 1, 2008, from http://www.ats.ucla.edu/stat/stata/output/stata_ologit_output.htm