Memoire Online - Service quality at a military hospital

The research is conducted at the military hospital situated in Libreville in Gabon.

The study is carried out at the military hospital located in Libreville, Gabon. The research is selected in this city because Libreville is perceived as the capital, the largest and the most populated city of Gabon.

The population refers to the broader group from which sampling elements are taken and to which results can be summarised. The population includes all the people which characterize the unit of evaluation. A target population should be defined in very particular terms. This will make the selection of respondents from the population for sampling, simpler (Terre Blanche et al., 2006:133).

Self-administered questionnaires were distributed to all existing patients of the military hospital in Libreville, Gabon, 18 years or older, males and females, who had experienced medical services and stayed over at the military hospital for at least one night. Questionnaires were only given to them once they have been discharged from hospital.

However, individuals who did not experience medical services at the military hospital were excluded from the study. Figure 4.2 provides a sum up of the target population, sample units, sample elements and actual sample size of the study.

Figure 4.2: Target population, sample units, sample elements and actual sample size

Target population


	Sample unit		Military hospital in Libreville, Gabon
		Sampling elements		Patients experiencing services at the military hospital
				(Period of March 2013)

Actual sampling size				200 patients

The sampling technique used to choose a representative sample for the study was crucial for the research and will be described next.

The sampling procedure includes any process using a small number of constituents from the entire population to draw conclusions related to the whole population. A sample is an extract of the broader population. The aim of sampling is to allow researchers to assess some unknown population's traits. There are two major sampling techniques namely probability and non-probability sampling. Non-probability sampling is based upon the researcher's own judgement to choose the sample where he or she chooses what elements to incorporate. Probability sampling takes place when sampling constituents are chosen by chance. All units may not necessarily have the same chance

of being chosen, but the probability of choosing each unit can be specified. Non-probability sampling involves convenience sampling, judgement sampling, quota sampling and snowball sampling. Probability sampling techniques encompass simple random, systematic, stratified and cluster sampling (Rahman & Miazee, 2010:27-28).

In this research, the probability sampling technique was used to choose respondents in the study, since it constitutes the root for all survey research (Parasuraman et al., 2007:340).

This sampling technique was used in this research. Probability sampling is usually appropriate in survey-based research where one is required to make interferences from the sample about a population to resolve research questions. Probability sampling can be divided into four phases:

? Verify that the sample is a good representation of the population (Holder, 2008:73).

The probability sampling technique was selected for this research since in this technique, each unit of the population, namely all patients 18 years or older, males and females, experiencing services at the military hospital in Libreville, Gabon had a known, non-zero chance of being incorporated in the sample. Sampling was not conducted at the discretion of the researcher.

In this technique, the probability of each unit being chosen from the population is known and is often equivalent to all cases. This indicates that there is a possibility to resolve the research questions by statistically estimating the population traits from the sample (Parasuraman et al., 2007:340).

There are five major methods which can be utilised throughout probability sampling. These are (Zikmund & Babin, 2007:273):

? Simple random sampling: In this sampling method, every unit in the population has a known and same chance of being chosen in the sample. Each unit is chosen independently. Simple random sampling will comprise of putting all the units of the population in a container, and extracting the sample from this.

? Systematic sampling: In this technique, the units of the population are counted from one to the number of units that constitute the sample, Prior to completing systematic sampling, the population size should be divided by the volume of the sample to establish an interval i. The response is rounded off to the closest integer. If the population is 100 000 units and a sample of 1 000 is chosen, then one will divide 100 000 per 1 000 which is 100, to find the interval.

? Stratified sampling: It refers to a two-stage procedure where the population is primarily divided into strata or subgroups. A population stratum is a fragment inside that population which has one or more similar features. These strata must be communally exclusive and jointly complementary. This implies that every unit must be incorporated into only one subgroup. In the next stage, units are chosen from every strata or subgroup through simple random sampling.

? Cluster sampling: With this technique, the population is divided into communally exclusive and jointly complementary clusters, after which some clusters are chosen in the sample. Cluster sampling is opposed to stratified sampling since a variety of clusters must be as similar as possible. The units of all the clusters will thus have the same traits. The supposition is thus made that any of the chosen clusters in the sample will correspond to the clusters which are not chosen in the sample.

? Two and multistage sampling: This method is often utilised to solve issues related to a geographically dispersed population when face-to-face contact is required, but will be too costly. Through that method, a sample is primarily extracted from the population, such as in the metropolitan regions in Gabon. From it, a second sample will be made, as in particular residential zone in a metropolitan region and finally, another sample will be made from that to concentrate only on a particular street in the residential zone.

Both, stratified sampling and simple random sampling were conducted in this research. The motive for choosing that sampling method was that the sampling framework of the research was divided into strata, and the sampling procedure was conducted

independently of each stratum. Stratified samples are perceived to be very efficient, and they enable investigating the interests of particular subgroups inside the population. Stratified random sampling provides better representativeness of the whole population, and also leads to fewer sampling errors, providing more accuracy in estimation (Du Plessis, 2010:140). In stratified sampling, strata should be mutually exclusive and jointly exhaustive in that each population element should be assigned to one and only stratum and no element should be excluded (Malhotra, 2007:327). The Department of Internal Medicine of the military hospital in Libreville in Gabon is divided into four main units, each unit represented an independent stratum.

As there is only one reception in each unit, no further random selection was required. As all the clinical units were not equal in size and did not serve an equal number of patients, a proportionate number of patients who received medical services for at least one night were selected at each unit (stratum). Self-administered questionnaires were distributed to identify patients at each unit. Permission to conduct the study was obtained from the nurse manager of the Department of Internal Medicine of the military hospital in Libreville in Gabon. The patients interviewed at each clinical unit were randomly selected. The study made use of a simple random technique where each population element had not only a known, but an equal chance of being selected (Munyaradzi, 2010:209). If a patient did not want to be involved in the research, the next willing patient was selected, and thereafter, the second patient after each willing one.

The volume of the sample implies the statistical accuracy of the findings. The size of the sample is a result of alteration in the population parameters and the assumption of quality which is needed by the researcher. In general, bigger samples reduce the likely error in generalising the population. In other words, larger samples are more representative of the population and result in more accurate findings. The volume of the sample can also be decided on the basis of personal judgement and statistical evaluations (Terre Blanche et al., 2006:236).

In the Gabonese health care industry, though some hospitals keep records of their patients, this is expected to be a problem for some medical institutions. The core reason is the durability of the service product sought by patients from hospitals from time to

time. It is difficult to tell when a patient will re-visit and purchase the service at the hospital. Against this background, the following formula will be used to estimate the response rate and the actual sample size needed (Saunders et al., 2007:214):

This calculation is based on three major aspects namely the level of confidence of the accuracy of the estimate, the margin of error which can be accepted, and the proportion of answers that the researcher expects to have some particular attribute.

Assuming that the researcher knows the level of confidence and the margin of error, it will be easier to have an estimation of the proportion of answers that the researcher expects to receive some particular attribute (Saunders et al., 2007). In general, researchers use a 95% level of confidence, which means that if one selects a sample 100 times, at least 95 of these samples will reflect the true characteristics of the population. The margin of error relates to the precision of the researcher's estimates of the population. The standard deviation, also known as error margin usually used in business and management researches is 5% (Munyaradzi, 2010:211). This means that if 40% of the researcher's sample lies in a certain category, then the estimate for the total population within this same category will be 40% plus or minus 5%. For the purpose of the current study, a 5% margin of error and a 95% confidence level will be used.

Table 4.2 shows the minimum sample sizes for different sizes of the population at a 95% confidence level to provide a good decision model (Saunders et al., 2007:212).

Population	Five per cent (margin of error)
100	44
200	132
300	168
400	196
500	217

Thus, according to this study, a minimum population frame of 100 patients for the military hospital in Libreville, in Gabon was estimated. This entails that referring to the Table 4.2, the minimum sample which can be expected is 44 respondents.

According to Saunders et al. (2007:215), a 50% response rate was suitable for surveys done through questionnaires. Thus, since the current study uses a questionnaire instrument to gather data, the researcher estimated a 50% response rate.

According to the formula provided above, the expected sample size for this research will be:

Service quality at a military hospital

4.2.6 Step 6: Determine the research frame