The effect of HRM Practices on the Relationship between Employee Age and
Employee Performance
The second specific objective was to determine the effect of HRM Practices on the relationship
between employee age and employee performance. The hypothesis formulated for this objective
was as follows:
Hypothesis II:
The effect of Employee Age on Employee Performance is moderated by Human Resource
Management Practices in Kenyan State Corporations
The effect of Human Resource Management Practices on the relationship between Employee Age
and Employee Performance was tested using stepwise linear regression, a method advanced by
Baron and Kenny (1986). This is a regression technique in which the choice of predictive
variables takes the form of sequence of F-tests or T- tests. Baron and Kenny’s (1986) model
involves 3 step procedures in which several regression analyses are conducted and the
significance of the beta coefficient in each step determined. This method involved testing the
effects of the independent variable (employee age) and the moderating variable, (Human
Resource Management Practices) on the dependent variable (employee performance) and the
interaction between employee age and Human Resource Management Practices.
The independent variable and the moderating variable are entered and an interaction term is
created by multiplying the independent variable and the moderating variable. To find out if the
moderating variable alters the strength of the existing relationship, the interaction term is entered
in the regression equation in stage 3. Both R2 change and the interaction term should be
significant at (p ˂ 0.05) to indicate moderation. Regression results for the tests specified above are
presented in Table 4.34. a, b and c. The first stage, Model 1 tests the single relationship
between employee age and employee performance. This model produced R2 = 0.001, F (1,286) =
0.345, p ˃.05. The model reveals a statistically non-significant relationship between employee age
(independent variable) and employee performance (dependent variable) as the p-value (0.557) ˃
0.05. The findings show 0.1 % of the variation in employee performance is due to employee age.
The influence of employee age on employee performance was also non-significant (β = 0.026, t =
0.587, p ˃ 0.05). The findings confirmed the first step in testing for variance.
In stage 2 both employee age and HRM Practices were entered into the regression equation
simultaneously as presented in Model 2, Table 4.34. This model produced R2 = 0.069, F (2, 285)
= 10.535, p ˂ 0.05. As shown in the table, 6.9% of the variation in employee performance is
explained by (R2 = 0.069, F (2, 285) =10.535, p ˂ 0.05). The value of R2 (6.9 %) implies that
93.1% in employee performance is due to other factors not included in the study. The influence of
HRMP on employee performance was weak but significant (β = 0.181, t = 4.550, p ˂ 0.05). For
every unit change in HRMP, there is a corresponding 18.1% of change in employee performance
(β = 0.181, t = 4.550, p ˂ 0.05).
In stage three, the interaction between employee age (independent variable) and human resource
management practices (HRMP) was created. To create the interaction term, employee age (EA)
and Human Resources Management Practices (HRMP) were multiplied and entered in the
regression model to get a single indicator representing the product of the two variables. This
model produced R2 = 0.073, F (3,284) = 20.972, p ˂ 0.05) which is statistically significant. The
change of variance in age is explained by R2 = 7.3 % and a beta coefficient of β = 0.172 (17.2%)
and t = 6.248. The findings from Table 4.33 indicate that the overall model was significant (R2 =
0.073 (7.3 %), F (3, 284) = 20.972, p ˂ .05. This confirmed hypothesis two, that HRM practices
moderated the relationship between employee age and employee performance.
Regression results for the moderation effect of HRM Practices on the
relationship between Employee Age and Employee Performance.
a. MODEL SUMMARY
Model R R
Square
Adjusted
R Square
Std. Error
of the
Estimate
Change Statistics
R
Square
Change
F
Change
df1 df2 Sig F
Chang
e
1
.035
a
0.001 -.002 0.58557 0.001 .345 1 286 0.557
2
.262
b
0.069 0.062 0.56639 0.068 20.701 2 285 0.000
3.270
c
0.073 0.063 0.56621 0.004 1.186 3 284 0.277
b. ANOVAa
Model Sum of Squares Df Mean Square
F
Sig.
Regression
1Residual
Total
Regression
0.118 1 0.118 0.345 0.557b
98.069 286 0.343
98.187 287
6.759 2 3.380
.321
10.535 0.000c
91.428 285
2Residual
Total
Regression
3Residual
Total
20.972
98.187
17.807
287
35.936 0.000d
80.380 284 0.283
98.187 287
c. COEFFICIENTS
Model Unstandardized
Coefficients
Standardize
d
Coefficients
t
B Std. Error Beta Sig.
(Constant)
1
Age
4.279 0.081 53.070 0.000
0.026 0.045 0.035 0.587 0.557
(Constant)
2 Age
3.676
0.043
0.154
0.043 0.057
23.905
0.994
0.000
0.321
HRM practices 0.181 0.040 0.261 4.550 0.000
(Constant)
Age
3
HRM practices
EA* HRMP
3.861 0.147 26.187 0.000
-.646 0.118 -.857 -5.497 0.000
0.125 0.038 0.181 3.264 0.001
0.172 .027 0.971 6.248 0.000
a. Dependent Variable: Employee Performance
b. Predictors: (Constant), Employee Age
c. Predictors: (Constant), Employee Age, HRM management practices
d. Predictors: (Constant), Employee Age, HRM management practices, Employee Age * HRM
Practices.
,
Substitutin
g for eac
h
coefficie
nt at each
step
of t
h
e
R
egre
ss
io
n
p
roce
ss
The following regression equations have been generated by substituting actual beta coefficient.
STEP 1. EP = β0 + β1 EA + ε
EP = 4.279 + 0.026EA +
0.081
STEP 2. EP = β0 +β1EA+HRM P +ε
EP = 3.676 + 0.043EA + 0.181 HRM P + 0.154.
STEP 3 EP = β0 +β1EA+β2HRM P + β2 EA* HRM P +ε
EP = 3.861+ -.646EA + 0.125 HRM P + -.646 EA * 0.125 HRM P + 0.147
Discussion
The second objective of the study sought to determine the influence of HRM Practices on the
relationship between employee performance and employee age in State Corporations of Kenya.
Hypothesis Two, drawn from this objective, stated that the effect of employee age on employee
performance was moderated by HRM Practices. Stepwise regression was used to test this
hypothesis. In stage one, the findings revealed a relationship that was statistically non-significant
between employee performance and employee age (R2 = .001, F (1, 286) = 0.345, p ˃.05). The
influence of employee age on employee performance was also non-significant (β = 0.026, t =
0.587, p ˃ 0.05). This finding contrasts with that of Grand (2005) which explored the relationship
between corporate age structures and performance of employees, and found that that the mean
employee age and age dispersion were inversely (u – shaped) related to performance. A study by
Waang et al., (2016) found that the average company age was positively related to company
performance. The difference between the current study and company average age study could be
attributed to the focus of the previous study on organizational performance rather than employee
performance.
When employee age and HRM Practices were entered into the regression model simultaneously in
the second step, the model produced R2 = 0.069, F (2, 285) = 10.535, p ˂ 0.05). 6.9 % of the
variation in employee performance is explained by R2 = 0.069, F (2, 285) = 10.535, p ˂ 0.05. This
confirmed that there was a weak but positive significant relationship between the two study
variables. The relationship was significant as p (0.000) ˂ 0.05. The influence of HRMP on
employee performance was also significant (β = 0.181 t = 4.550, p ˂ 0.05). For every unit change
in HRMP, there is a corresponding 18.1 % of change in employee performance (β = 0.181, t =
4.550, p ˂ 0.05). The results of the current study do not support a direct relationship between the
variables employee age and employee performance in state Corporations of Kenya. The findings at
this stage are consistent with porter (2008), who found that HRM Practices differently influenced
age cohorts to better their performance.
In step 3 the interaction term (EA*HRMP) was entered in the regression model. The effect of the
interaction term on the relationship between EA and EP indicated a statistically significant
relationship (R2 = 0.073, F (3, 284) = 20.872, p ˂ 0.05). The analysis yielded a positive significant
beta coefficient (β = 0.172, t = 6.248, p ˂ 0.05). The prediction of expectancy theory that
performance does not depend only on the magnitude of the effort exerted is indirectly supported
by the findings of this study. That is the strength of the effect of the relationship between
employee age on employee performance is made stronger by Human Resource Management
Practices. Upon entering the interaction term in the regression model, the effect of employee age
on employee performance was enhanced by 7.3 %. The results of the current study do not support
the direct relationship between employee age and employee performance in Kenyan State
Corporations. These findings are in line with expectancy theory. The findings support the
expectancy theory of motivation (1964) which states that employee performance depends not only
on the magnitude of efforts but also on other factors such as individual abilities, traits and
motivation. The current study builds on expectancy theory to establish the fact that age alone is
not an effective determinant of employee performance in the absence of other factors such as
perception, reaction to the organization and motivation. In this case HRM practices are one of the
moderating variables and therefore the results strengthen expectancy theory.
The current study results support those of May et al. (2014) who tested the moderating effect of
HR practices on the relationship between employee characteristics (age) and employee work
engagement. Results indicated that HR practices impacted the relationship positively. In her
study, Omari (2012) found that HR practices contributed significantly to the moderation of the
relationship between age and employee outcomes as demonstrated by F values. This essentially
implies that good and favorable HR practices are appreciated by all regardless of age. The
difference between the current study and Omari’s study is that the latter focused on organization
performance. Distribution of age in Kenyan State Corporations may be largely similar across the
board, meaning that the effect of age on organization performance may not show clearly when
State Corporations are compared. The current study directly focuses on employee performance.
When examining age differences and core work performance, most studies have looked at one
single characteristic, organization performance which tends to give rise to generalized results. The
current study corrects the situation by looking at employee performance which gives specific and
more clear results on individual workers facilitating easy identification of weaknesses or strengths
of the workers.
A hierarchical (Stepwise) linear model by Collins, et al. (2009), sought to determine the
relationships between job performance and employee affective commitment and the moderating
effect of HR strength on this relationship. Findings indicated that HR practices moderated this
relationship positively. A related study by Muindi (2014), tested the effect of personality as a
moderator on the relationship between QWL and job satisfaction. The hypothesis of this study
stated that the relationship between QWL and job satisfaction is moderated by personality. When
the interaction between QWL and Personality was introduced into the regression model, results
indicated a statistically positive significant relationship. The interaction term (QWL*JS) created
regression weights which moderated the effect on the relationship between QWL and job
satisfaction and the hypothesis was supported.
A study by Njenga (2017) examined the moderating effect of leadership style on the relationship
between psychological contract and organizational performance using stepwise regression
analysis. The result of the moderation was not confirmed. In step one, the regression results
between psychological contract and organization performance were significant. R2 = 0.238, F (1,
37) = 11.548, p ˂ 0.05). In step 2, psychological contract and organizational performance in the
presence of leadership style, the overall model was significant at R2 = 0.261, F (2, 36) = 5.11, p ˂
0.05). In step 3, the interaction term (PC*LS) was introduced in the regression model. The results
became insignificant R2 = 0.266, F (3, 35) = 3.377, p 0. ˃ 05). This implied that leadership style
as a moderator had no significant effect on this relationship.
