Research Design on staff turnover and employee performance in nonprofit healthcare providers
Cooper and Schindler (2014) assert that research design determines the approach or the
strategy that the researcher uses. Based on the research questions, it’s used as a guide to
collect and analyze data (Saunders, Lewis & Thornhill, 2016). Descriptive research design
was used in this study for its flexibility and appropriateness based on type of data required
data collection methods and the research timing. Zikmund and Babin (2012) define
descriptive research as the study concerned with the frequency or relationship between two
variables. Descriptive survey entails collection of information by administering a
questionnaire to sample of selected population (Orodho, 2004). The design adequately helped
in answering research questions apart from it being fast and cheap. It also showed the causal-
effect relationship of the given variables such that the researcher was able to tell how much
one variable explained another one or explained by another. This design has an advantage of
allowing flexibility in data collection by use of closed and open ended questions. This
provides chance for any additional information by the respondents. It’s argued that a great
deal of accurate information is obtained when descriptive surveys are used in fact finding.
The Target Population
O’Gorman and Macintosh, (2014) asserts that a population as a complete group of elements
amongst which the researcher wants to study. Mugenda and Mugenda (2012) indicate that a
target population is complete set of individuals, cases or objectives with similar
characteristics. The study population composed of all the staffs working with AMPATH
headquarters, WFP and World Vision in Uasin Gishu County, Eldoret. Those interviewed
were senior managers, middle level management staffs and general staff not in management
positions. This group of management has direct responsibilities in daily managements of the
selected programs towards achieving their main goals. The study targeted all the 125 staffs
working with AMPATH, 32 WITH WV and 38 with WFP as its target population. The target
sample size ensured every respondent had a chance of inclusion.
Table 3.1 Target Population
NO. NAME OF HEALTH PROVIDER NO. OF EMPLOYEES IN
THE COUNTY
1 AMPATH 125
2 WFP 38
3 WV 32
TOTAL 195
Sampling Design
Mugenda and Mugenda (2012) assert that sampling is the process of selecting respondents
that ensures selected respondents represents the entire population under the study. The study
conducted a census of all respondents within the three providers chosen through purposive
sampling because the population was small and manageable. This was appropriate as it
included all the respondents.
Data Collection Instruments
Data collection for the study was done using self-administered questionnaire with
questionnaires being the instruments used to collect data on staff turnover in relation to
employee performance within the organizations. The questionnaire had structured and
unstructured questions and the questionnaire divided into three parts as follows: Section A
detailed respondents’ demographic characteristics; Section B explored on causes of employee
turnover in those programs under study; and Section C focused on the negative effect
employee turnover on the performance of the respondents. Structured questions are checklist,
ratings, multiple choice, dichotomous, raking response strategies (Cooper & Schindler,
2010). The questionnaire was formulated in a likert scale that sought to find extreme of
agreement or ranges in given parameters. The questionnaire sought establish the influence of
voluntary turnover of staff on employees performance, how involuntary turnover of staff
affect the employees performance, how functional turnover of staff relates to performance of
employee and how dysfunctional turnover of staff relates to performance of employee in
nonprofit healthcare providers in Uasin Gishu County, Kenya.
Data Collection Procedure
This involved the distribution of questionnaires to respondents by hand and any clarification
done to easy understanding in a timely manner. Questionnaire being a useful data collection
technique was preferred over other methods of collecting data. Primary and secondary data
were used in the study with questionnaires being used to collect primary data. Mugenda and
Mugenda (2012) assert that to obtain crucial information about a population, questionnaires
are mostly used. In this study questionnaires were useful because what to be asked was
known and that standard questions for all the respondents were asked. Extensive information
from many respondents was easily obtained using the questionnaires. Questionnaire used was
accurately structured and effectively administered to bring out the crucial information by
which inferred statements or conclusions were made on the entire population.
Validity and Reliability of Research Instruments
Validity of Research Instruments
This refers to degree by which the empirical measure or a number of measures of a concept
accurately measures the concept (Orodho, 2004). Validity mainly focuses on establishing
whether the questionnaire items are measuring the intended measures. Content validity was
used in the research. The degree of the content for a certain test in relation to the traits
designed to be measured is done through content validity (Best & Kahn, 2004). Items in the
questionnaire were discussed with the supervisor to ensure content validity. The outcome of
the discussion and guidance by the supervisor established the validity of the research
questionnaire. Pretesting of the questionnaire was done in one of the organization using
random sampling of at least 10% of the actual sample to be used.
Reliability of Research Instruments
Mugenda and Mugenda (2003) define reliability to be the degree to which the instruments of
research yields consistent results when repeated. The study conducted test retest as a measure
of increasing reliability of the data collected. In this arrangement the research instrument or
rather questionnaire was administered twice to the same respondent but at different times to
establish whether the same responses would be obtained. Reliability testing was done by
piloting the instrument on 10% of the total sample size to be used in the three service
providers (Kothari, 2004).
Cronbach’s alpha was worked using SPSS to establish the coefficient of reliability. The value
of the alpha coefficient ranges from 0 to 1 and was used to describe the reliability of factors
extracted from dichotomous in cases where a question had two possible answers of those
with high values indicative of reliable generated scale. Coefficients greater than 0.7 leads to a
conclusion that the instruments used were well designed for the study. Any fine tuning of the
questionnaire can be done based on the results of pilot study.
Data Analysis and Presentation
The process of cleaning, analyzing, transforming and modeling data collected is referred to
as data analysis (Wagner, Halley & Zaino 2011). Data obtained was analyzed using
descriptive statistics. In this process the information was coded on certain specific
characteristics and attributes using descriptive statistics of frequency of distribution and
using measures of central tendency. Findings were presented using illustrative tables while
the quantitative data was analyzed to establish differences and consistency; such was
presented in a narrative form as discussions. Qualitative and quantitative data was collected
during the study. Quantitative data was analyzed using descriptive statistics by use of
percentages, frequencies and measures of central tendency such as mean. Qualitative data
was analyzed using content analysis. Data obtained was edited for consistency and
completeness the coded and interpreted in line with the study objectives. Bar charts, pie
charts, frequency tables and cross tabulations were used in data presentation. The statistical
significance of factors affecting performance of staff in relation to turnover was established
using descriptive statistics of frequency distribution and percentages. This showed the rates
and patterns on the distribution of the demographic traits in terms of age, sex, marital status,
and level of education of the respondents on the identifiable variable.
Percentages were used to determine the performance of staffs in relation to the four types of
staff turnover in the study. Influence of staff turnover on performance was established by use
of cross tabulation. Extent by which every independent variable predicts staff performance
was determined using the model of multivariate regression.
Y = β0+β1X1+β2X2+β3X3 + α
Where
β0, β1, β2, β3, β4 = coefficients of constant, voluntary turnover, involuntary turnover,
functional and dysfunctional turnover.
Y is the coefficient of performance of
staffs X1 = Voluntary turnover
X2= Involuntary turnover
X3= Functional
X4 = Dysfunctional turnover
ε= the error term which is normally distributed about a mean of zero. For purpose of
computation, this error will be assumed to be zero
Ethical Considerations
Ethics refers to acceptable standards of behavior or norms intended to moderate
moral choices of an individual behavior and relation with others. Legal constraints
are different with ethics because in legal constraints, the acceptable behavior
standards have specific penalties generally accepted. Ethical treatment of
respondents, research ethics and researcher ethics are some notable observed in this
study. Respondents were requested for the informed consent, benefits and rights to
confidentiality and privacy of their participation. Research ethics includes quality
research by choosing appropriate research design to unfold the research questions
and appropriate data analysis for the obtained data. Researcher ethics ensured all
the self-administered questionnaires were all well filled by the respondents before
collection to avoid incompleteness of the data.
