1 / 12100%
DEMOGRAPHIC INFORMATION OF THE RESPONDENTS ON EFFECTS OF
HUMAN RESOURCE PRACTICES ON EMPLOYEE TURNOVER IN THE
FLOWER INDUSTRY
A total of 321 out of 357 respondents fully filled and returned the questionnaires.
Therefore the return rate for questionnaires used for data analysis was 89.9% which was
considered adequate to provide sufficient information on the influence of the human
resource practices on employee turnover in flower industry. The demographic information
gathered from participants included; age, gender, marital status, education level and work
experience.
Age of the Respondents
The respondents were asked to indicate their age in the questionnaire. The results are
presented in Figure
Figure Age of the Respondents
Source: Survey Data 2022
The results on employees’ age are presented in Figure 4.1. Showed that majority of the
(139, 43.3%) respondents were aged 36-45 years, (122, 38.0%) respondents were aged 26-
35 years, (33,10.3%) respondents were aged 18-25 years and (19, 5.9%) respondents were
aged 46-55 years while (8,2.5%) respondents were aged over 55years. The study findings
imply that most of the employees in the flower firms are within the productive age and can
enhance high productivity of the firms. However, Grady et al.,(2008) argued that the Irish
employee force was ageing and could lead to higher demands for more flexible working
arrangements for employees with eldercare and childcare responsibilities.
Gender of the Respondents
In addition, the respondents were asked to indicate their gender in the questionnaire
provided. The results are presented in Figure
Figure Gender of the Respondents
Source: Survey Data 2022
The study results revealed on gender in Figure 4.2 shows that majority (200,62.3%) of the
respondents were female while (121,37.7%) employees were male. From the responses, it
emerged that majority (62.3%) of the employees working in flower firms in north Rift
Region were females and compared to their male counterparts. This shows that workforce
in flower firms is female dominated. It has been shown that gender inequity in an
organization could have an effect on employee turnover as argued by Dolan et al.(2001).
Table Gender of the Respondents
Gender Frequency Percentage (%)
Male 121 37.7
Female 200 62.3
Total 321 100
Marital Status
Figure Marital Status of the Employees
Source: Survey Data 2022
Further, the employees were asked to indicate their marital status, the results are presented
in Figure 4.3. As shown in Figure 4.3 majority (246,76.6%) respondents were married,
(65,20.2%) were single, (9,2.8%) were divorced while (1,.3%) was widowed. The
responses showed that majority (76.6%) of the respondents were married. The study
results on employee marital status show that most of the employees had family issues
which could have an effect on their work schedules. It has been argued that among the
reasons for employee turnover in most organizations is marriage (Zlotnic et al., 2005).
This therefore implies that the employees in flower farms in north Rift Region could leave
their organizations due to their family ties.
Employees’ Education Level
Similarly, the employees were asked to indicate their highest level of education. Their
responses were tabulated and the results are presented in Figure 4.4.
Figure Education Level of the Respondents
Source: Survey Data 2022
Study results on employees educational level as presented in Figure 4.4 showed that
majority (103,32.1%) of the employees were secondary certificate holders, (98,30.5%)
were primary certificate holders, (48,15.0%) were primary school dropouts, (42,13.1%)
had technical institutes level of education, (14,4.4%) were diploma holders and (11,3.4%)
were first degree holders while (5,1.6%) employees had masters’ level of education. The
study findings suggested that majority (62.6%) of the respondents had either primary or
secondary school education level. This could imply that flower farms attract employees of
low education level. This concurred with what Ayieko (2011) found in Timau Flower farm
workers which indicated that Flower farms in Timau mainly engaged the services of
young men and women mostly with no education or those of primary school level. In
addition, the employees in the flower firm industry may want to leave but cannot find
other jobs since they lack skills as Harris, (2007) puts that talented candidates in the global
job skills market have the luxury of choice.
Work Experience
In addition, the employees were requested to indicate their work experience. The results of
data analysis are presented in Figure
Figure Employees’ Work Experience
Source: Survey Data 2022
The results as presented on Figure 4.5 show that majority (115,35.8%) employees had
work, (86,26.8%) employees had work experience of less than 1 year and (8,2.5%)
employees had work experience of 11-15 years while (7,2.2%) employees had work
experience of 16-20 years. The study findings showed that the bulk of the employees in
flower firms had work experience of 2 -10 years implying that the respondence had
adequate knowledge to be able to give reliable information. Greenberg and Baron (1995)
contend that employees with many years of service perceived higher job satisfaction than
their colleagues with less job experience. This shows that those employees who have
stayed for more than 10 years could be satisfied with the job and could not want to leave
the profession.
Normality Test Results of Dependent Variable
A normality test is a statistical process used to determine if a sample or any group of data
fits a standard normal distribution (Shapiro & Wilk, 1965). When carrying out statistical
analysis using parametric methods, the assumption of normality is a fundamental concern
for the analyst. Statisticians, conclude that the data are normal or not normal based on
some test of normality results (Hogg & Tanis, 2006). Normal distribution is an underlying
assumption for many statistical procedures. It is also the most frequently used distribution
in statistical theory and applications. Though it is important for certain statistical
procedures to assume that data should come from a normal distribution, in real life it is
indeed impossible for the data to be perfectly normal (Geary 1947). There never was and
will never be a normal distribution (Hart and Hart 2002).
Table Normality Test Results for Dependent Variable
Factors Kolmogorov-Smirnov
a
Statistics df Sig
Shapiro-Wilk
Statistics
df P-value
Significance
Employee turnover .041 321 .200 .990 321 .485
a. Lilliefors Significance Correction
Source: Survey Data 2022
To test for normality of the dependent variable (Employee turnover), Kolmogorov-
Smirnova and Shapiro-Wilk tests were conducted. This was fundamental to determine
appropriate tests to be conducted and make sure that assumptions of a normal distribution
were not violated (Math-Statistics-Tutor, 2010). Kolmogorov-Sminov and Shapiro-Wilk.
Shapiro-Wilk test for normality were used to detect all departures from normality (Math-
Statistics-Tutor, 2010). The tests reject the hypothesis of normality when the p-value is
less than or equal to 0.05 (Sharpiro and Wilk, 1965). Table 4.2 shows that the
Kolmogorov-Smirnova and Shapiro-Wilk statistics were .041 and .990 respectively. The
associated p-value was .200 and .485 for Kolmogorov-Smirnova and Shapiro-Wilk
statistics respectively. Since the p-values were greater than the significance level (0.05)
(not significant at p<.05), this implies that the variables were normally distributed.
Figure Normal Q-Q Plot for Employee Turnover in flower farms
Source: Survey Data 2022
The visualized distribution of random variables of different between expected distribution
and the observed distribution of employee turnover are presented in Figure 4.6. A quartile
by quartile or Q-Q plot forms a 45-degree line when observed values are in conformity
with the hypothetical distribution. Q-Q plots the quartiles of a variable’s distribution
against the quartiles of the test distribution. The table above shows normality plots of the
data. Normal Q-Q Plot provides a graphical way to determine the level of normality. The
black line indicates the values the sample should adhere to if the distribution was normal.
The dots are actual data. If the dots fall exactly on the black line, then the data are normal.
If they deviate from the black line, then the data are non-normal. From the above table the
dots fall along the line which implies that the data distribution was normal.
Source: Survey Data
Figure 4The Figure shows minimal deviation from normality. Thus overall, the
distribution appeared normally distributed. On the basis of the computed significant test
statistics, for Kolmogrov-Smirnov and Shapiro Wilk tests, normality of dependent variable
was maintained. This means that the significance test conducted on the data were fairly
accurate (Shlin and Miles, 2010).
Table Homoscedasticity of the Residuals of Dependent Variable
Levene Statistic df1 df2 P-value
4.642 11 310 .000
Figure Histogram for Employee Turnover in ower farms
Assessment of homoscedasticity of the residuals of employee turnover was conducted.
OLS makes the assumption that the variance of the error term is constant
(Homoscedastic). If the error terms do not have constant variance (have differing
variance), they are said to be heteroscedastic. Violation of this assumption leads to bias in
test statistics and confidence intervals. Levene Statistic was used to test the hypothesis for
the homogeneity of variance that is, the error variances are all equal or homoscedastic.
Table 4.2 shows Levene Statistic of 4.642 with an associated p-value of.000. Since the
probability associated with the Levene Statistic is 0.000 which is less than 0.05 level of
significance, we fail to reject the hypothesis and conclude that the variance of the
dependent variable were homogeneous.
Table Breusch-Pagan and Koenker Test for Heteroskedasticity
SS df MS
F
Sig
Model 12.757 4.000 3.189 1.088 .000
Residual 416.364 142.000 2.932 -999.000 -999.000
Similarly Breusch-Pagan and Koenker test statistics was also used to test the null
hypothesis that heteroskedasticity was not present (homoskedasticity) if significant-value
is less than 0.05, reject the null hypothesis. Breusch-Pagan test is a large sample test and
assumes the residuals to be normally distributed. Table 4.4 shows Breusch-Pagan and
Koenker test statistics of 12.757 with an associated p-value of .000. Since the probability
associated with the Breusch-Pagan and Koenker test was 0.000 which is less than 0.05
Students also viewed