HEALTH/CARE SEEKING BEHAVIOUR

Get Complete Project Material File(s) Now! »

CHAPTER 3 RESEARCH DESIGN AND METHODS

INTRODUCTION

Prior to applying the empirical phase of a study, the researcher is required to make a decision about the research design and the research methods including the sample and sampling techniques to be applied in the selection of the study population. In this study, the researcher chose the research design and methods that would be appropriate to answer the research objectives and questions. The research method included discussions on the research population, sampling and sampling technique, and data collection methods and data analysis. Aspects of validity and reliability of the instrument and ethical issues were also described.

RESEARCH CONTEXT

Addis Ababa is located at 90° 2′ N latitude and 38° 45′ E longitude. Established in 1886 and with 2.7 million population size (female=52%), Addis Ababa is one of the oldest and largest cities in Africa. It constitutes four percent of the national and nearly a quarter of the urban population (23%) in Ethiopia which is eight times larger than the second largest city, Dire Dawa. Addis Ababa is much larger and more diverse than any other city in the country and the city is growing more rapidly (nearly by 4%) doubling every decade (Office of the Population Census Commission 2007:7; UN-Habitat 2008:7). It is also located at one of the highest at an average altitude of 2400 meters. Being the capital of a non-colonised country in Africa, it has been playing a historic role in hosting regional organisations such as the African Union, and the Economic Commission for Africa, which contributed to the decolonisation of African countries, and later bringing Africa together (UN-Habitat 2008:4). Politically, Addis Ababa is organised through smaller units called sub-cities. The sub-cities are further divided into Kebeles, the smallest formal administrative units in the political structure.
Addis Ababa hosts peoples of diversified ethnic background (more than 80) of which the majority are Amhara, Oromo, Guragie and Tigraway. Amharic constitutes the biggest share of mother tongue followed by Oromipha. In terms of religion, about three-fourth are Orthodox Christians followed by Muslims and Protestants (Office of the Population Census Commission 2007:91, 116, 143).
Located in the centre of Ethiopia and given the lack of development policies in other urban centres, Addis Ababa is privileged with the majority of social and economic infrastructure in the country. As a result, the very high rate (40%) of rural-urban migration from all the corners combined with rapid natural population growth poses critical challenges on the city. Like any other major city of Africa, it is presently suffering from widening income disparity, deepening poverty, rising unemployment, severe housing shortage, poorly developed physical and social infrastructure and the proliferation of slum and squatter settlements (UN-Habitat 2007:1). Soil, air and water (river) pollution as a result of industrial wastes are real growing concerns in Addis Ababa (UN-Habitat 2008:4). The capital is plagued with slum areas or Kebele houses built since the 1970’s with no significant change to date at the lowest standards in terms of density, sanitation and availability of potable water (USAID 2012:2).
The current health coverage in the capital has increased, the distance travelled to find a health facility is less than 2 kilometres, vaccination coverage is high, and the rate of HIV/AIDS infections have started to decline yet the second highest (5.2%) in the country. In Addis Ababa, the total fertility rate (TFR) (1.5%) is below replacement level. However, because of its primacy, the city has a disproportionate share of growth accompanied with health problems including congestion, pollution and streams of rural-urban migrants. Hence, the nature of health problems in Addis is different from other cities in the country (USAID 2012:2). The under-five mortality and neonatal mortality rates for the capital were 53 and 21 per 1,000 live births respectively (CSA [Ethiopia] and ICF International 2012:64, 71, 79, 80, 97, 101, 113, 120, 126, 129, 182, 235). The rise of metabolic syndrome is another emerging problem of the capital city (Misganaw, Mariam, Araya & Ayele 2012:3-4). The private sector has been playing a considerable role in improving the physical coverage with access remaining as a challenge due to unaffordable prices for the majority of the population. More than 83% of the Hospitals are owned by private owners as compared to only five public hospitals. There are also more than 500 private clinics in the capital (FMOH 2011:55).
Nearly 11% of all the physicians, both general practitioners and specialists, and 7% of midwives in Ethiopia are found in the capital, Addis Ababa, which constitutes only slightly less than four percent of the national population. Though this may connote the inequitable human resource for health distribution in the country, these health professionals are still overstretched. In Addis Ababa, one physician, either a general practitioner or a specialist, serves 17,607 people and one midwife serves 18,598 people (FMOH 2011:63). This ratio is far smaller than the 1:5,000 ratio for midwives and 1:1,000 ratio for physicians recommended by the WHO. In 2011, the share of the health budget for the City Administration was only five percent, the lowest of all the Regional States and City Administrations in Ethiopia (FMOH 2011:58; FMOH 2010a:1).

THE RESEARCH DESIGN

Research design, as defined in the Dictionary of Epidemiology, is “the ‘architecture’ of the study whereby its structure, the details of the study population, time frame, methods, and procedures are explicitly stated in the research protocol”(Porta 2008:216). According to Blaikie (2010:15):
“A research design is an integrated statement of and justification for the technical decisions involved in planning a research project. …designing social research is the process of making all decisions related to the research project before they are carried out … Designing a research project is the way in which control is achieved [I italicized].”
Research design, also sometimes called as a research strategy, can also be defined as an approach in research determined by stated research questions (Gravetter & Forzano 2010:159).

Design chosen

Choosing the right research design avoids invalid inferences and the research design ensures that the evidences obtained from the data analyses answer the stated research questions as explicitly as possible (Blaikie 2010:24-25). A research design tells us the level of control that the investigator imposes on the study. Compared to experimental designs, in descriptive and correlational designs the researcher has less or no control over the independent variables though random selection can provide control. The research design determines the sampling, measurement and analysis techniques to be applied in the research process (Gravetter & Forzano 2010:216-20).
A correlational design is a type of descriptive research design used to indicate the strength of association among variables of interest (Daughtery 2011:176; Shaffer 2009:22). According to Stangor (2010:160), correlational research design is ‘used to search for and describe relationships among measured variables’. The term correlation expresses the degree to which two or more variables change together or are related (Porta 2008:53). In this study, correlational design was employed because the researcher was interested to determine the relationship between the outcome variables and independent or predictor variables based on existing theories. Quantitative descriptions were made on the extent of the problem and other measurable attributes.
Correlational designs can be cross-sectional or longitudinal; retrospective or prospective based on the nature of the research questions, samples and measures (Gravetter & Forzano 2010:213). A correlational design uses descriptive design to describe the relationship between variables and also applies both descriptive and inferential statistics to identify relationships (Deming & Swaffield 2011:90, 93). It can be used for model testing based on existing theories or models (Gravetter & Forzano 2010:213). The inferential statistical part depends on mathematical models and there are three important dimensions to be considered in employing this design:
• The relationship between the variables
• The strength of the relationship
• The generalisation of the knowledge from the relationships of the variables obtained from the sample to the real world (Deming & Swaffield 2011:94)
Strengths of correlational research design
• Provides information about the strength of the relationship between variables
• It allows to study behaviour in everyday life
• It can be used when experimental research is impossible when predictor variables can’t be manipulated (Stangor 2010:177)
Limitation of correlational research design
• It is not possible to conclude that causal relationship is established among the measured variables (Stangor 2010:170-171)

RESEARCH METHOD

Research methods refer to all the methods, including data collection, statistical techniques and evaluation of the accuracy of the results, applied in the whole course of the research operation (Crowther & Lancaster 2008:68).

Sampling and population

Population universum

The group of people on which a researcher is interested in and on which the results of the study are applied refers to population (Bruce, Daniel & Stanistreet 2008:133). Population is the set of all measurements of interest to the sample collection (Ott & Longnecker 2010:5). Polit and Beck (2008:67) defined population as ‘all the individuals or objects with common, defining characteristics. The population universum for this study was all women of 15-49 years old residing in Addis Ababa.

Target population

Target population is a complete collection of objects or peoples whose description is the major goal of the study (Ott & Longnecker 2010:24). According to White (2011:248), a target population is a population which meets criteria for sampling. The target population for this study was all women of 15-49 years of age who have experienced at least one birth in the last 1-3 years before the date of data collection.

Accessible population

An accessible population refers to the population out of which a sample is taken or a group which can be accessed by a researcher reasonably (White 2011:251; Boswell & Cannon 2011:146). The accessible population for the study included those women (15-
49) with a history of at least one birth in the last 1-3 years preceding the date of data collection. The study population was selected based on the following inclusion and exclusion criteria. Lists of inclusion and exclusion criteria were provided for data collectors before they went for the house-to-house survey.

Inclusion criteria

Inclusion criteria also termed eligibility criteria refer to the characteristics that the study participants will have in common and that must be considered in selecting the study sample (Boswell & Cannon 2011:149). The inclusion criteria for this study include:
• Women of reproductive age groups (15-49 years)
• Women with history of childbirth in the last 1-3 years before the time of data collection or those who have under-three children
• Women who have been residents of Addis Ababa. In this case residents are those women who had been living in Addis Ababa for at least one year (one gestational period)

Exclusion criteria

Exclusion criteria refer to the characteristics of individuals that will make them ineligible to be in the sample (Boswell & Cannon 2011:149). The exclusion criteria in this research include:
• Children less than 15 years of age and women greater than 49 years of age regardless of their childbirth status
• Women without history of childbirth or those who have never given birth in the last 1-3 years before data collection
• Women who are defacto residents of Addis Ababa or visitors

Sampling technique

Researchers have a wide alternative of choosing either probabilistic or non-probabilistic sampling techniques. The non-probability sampling technique is used to select research participants non-randomly. Hence, non-probability sampling method doesn’t guarantee the representativeness of the sample for the larger population.
The most important sampling for generalisation to the population on the basis of samples is the random selection of study units through probability sampling (Babbie 2013:72). However, Babbie (2011:243) states that no probability sample is perfectly representative in all respects though errors can be minimised through controlled selection. Probability sampling can be simple or extremely difficult, time consuming or expensive. According to Babbie (2011:241), this method remains the most effective method for the selection of samples for two major reasons: it avoids the researcher’s unconscious or conscious bias in sample selection; and probability sampling allows statistical estimation of sampling errors.
The heterogeneity of the population under study forces a researcher to use probability sampling so as to reflect the variations that exist in the population (Babbie 2010:194). The most common types of probability sampling are simple random sampling, systematic sampling, stratified sampling and cluster sampling (Babbie 2011:242; 2013:89; Gravetter & Forzano 2010:153, 154). We can avoid incorrect conclusion of our study from selection bias, if we apply the right sampling technique (Babbie 2010:195; Bruce et al 2008:134).
The most preferred sampling technique used in large population that is spread in wider geographic areas is cluster sampling (Babbie 2010:215). Cluster sampling is a multistage sampling technique in which existing groups called clusters are selected or sampled initially and the members of each cluster are selected randomly afterwards (Babbie 2011:234). When the sample selection needs different levels of clusters, the sampling technique is referred to as multistage cluster sampling (Brase & Brase 2010:16). Cluster sampling is an easy way of getting representative samples through relatively random selection (Gravetter & Forzano 2010:154).
Strengths of cluster sampling:
• The sampling doesn’t require the complete listing of the entire population of the study area except those to be studied (Babbie 2011:234).
• No need to know all of the elements of the population.
• It is efficient strategy when the population is large or spread in wider geographic areas (Babbie 2011:235; Boswell & Cannon 2011:152).
Limitations of cluster sampling:
• The listing of households in the selected strata or groups (clusters) is both labour intensive and expensive.
• A multistage cluster sampling is subject to multiple sampling errors resulting in less accurate sample (Babbie 2011:235).
Addis Ababa, the study area, is divided into 10 sub Cities and each sub City is further divided into several small administrative units called Kebeles. Because of the different political or administrative structures and wider geographic areas, cluster sampling technique was employed for this study. In the 2007 Ethiopia Housing and Population Census, Kebeles were further subdivided into enumeration areas (EAs). An EA is a geographic area consisting of a convenient number of dwelling units which was used as a counting unit for the census. The average number of households (HHs) per EA in urban Ethiopia is 169. The number of clusters (EAs) in Addis Ababa was about 3865 (CSA [Ethiopia] and ICF International 2012:275-6).
The study employed a stratified, two-stage cluster design. Since Addis Ababa is entirely urban, stratification was achieved by using the sub Cities (10 strata). In the first stage, 30 sample points (EAs) were selected independently from all the strata with Probability Proportional to (EA) Size (PPS) of households using the 2007 Population and Housing Census data (Annexure A). A new household listing was not conducted for this study but the random sample selection was done using the number of households identified for EDHS 2011. In the second stage, 906 households were selected with PPS of households in each EA (Table 3.1). PPS is a special and efficient method in multistage cluster sampling (Babbie 2011:243).

Sample frame

A sampling frame is a list of individuals from which we select samples (Brase & Brase 2010:17). According to Babbie (2011:242), a sampling frame refers to ‘a list or quasi list of the members of a population’. For this particular research, the sampling frame for the first stage was the lists of clusters (enumeration areas) per strata and for the second stage were the lists of households in each EA. Households were the sampling units for this study. A sample unit refers to the object that is actually measured (Ott & Longnecker 2010:24).

Ethical issues related to sampling

This study employed an optimal sample size that can ensure external validity. Research participants were selected fairly or randomly among the target population. Participants were not selected systematically for reasons not related to the research focus. Participants were not paid incentive in return for their participation for this study.

Sample

A sample is any subset of the measurements selected from the population (Ott & Longnecker 2010:5); and a random sample of individuals should possess variations which are observed in the source population (Babbie 2013:76).
The term sample is also defined by Polit and Beck (2008:765) as a subset of the population that is selected for a particular study, and the members of a sample are the respondents. There are various types of samples with different degree of suitability for a particular study. A randomly selected sample is cost-effective and more efficient to produce good quality evidence as it is difficult to reach the whole target population (Boswell & Cannon 2011:147). Polit and Beck (2008:289) say quantitative researcher seeks to select samples that will make generalisation to the wider community or group possible.
In quantitative approach, we use a sample since it is unnecessary or impractical to study the whole population due to time, financial and other constraints. However, for generalisation to the general population to be realistic, the sample should be representative of the population. A sample is considered as representative if it approximately shares the “aggregate characteristics” of the general population and if all members of the population have equal chance of being selected in the sample (Babbie 2013:77; Boswell & Cannon 2011:147). Representativeness is a criterion to assess the adequacy of a sample. Ideally, a sample is representative of the accessible population and the accessible population is representative of the target population (Polit & Beck 2008:67, 353).
As there are multiple variables to be treated in this study, the sample size estimation was based on the maximum sample size for estimating single proportion approach (Bui & Taira 2010:414).
n = α²p(1-p)
ɛ²
Where:
n=required sample size
α=critical value for the chosen confidence level at 95% (standard value of 1.96)
p=estimated prevalence of the problem (variable being assessed)
ɛ=margin of error of 5% (standard value of 0.05).
For this particular calculation, p=84% (percentage of skilled attendance at birth for Addis Ababa) and hence the required initial sample size was 206. A design effect of 2 obtained in EDHS 2011 report was considered in determining the sample size (CSA [Ethiopia] and [ICF] International, 2012:127 & 303). In cluster sampling, a design effect of 2 or 3 is recommended (Dattalo 2008:36). Then n=2*206=412 was the minimum sample size for each of the two sampling stages namely the Enumeration Areas (EAs) and the households and hence, n=2*412= 824. The sample size was further increased by 10% to account for contingencies such as non-response or recording errors. Therefore, the minimum sample size required to conduct this research was 906. Based on rough calculation from the 2007 Census data, at least one mother of under-three children is expected in every five household (Office of the Population Census Commission 2007:405).

Table of contents
CHAPTER 1 ORIENTATION TO THE STUDY
1.1 INTRODUCTION
1.2 BACKGROUND TO THE RESEARCH PROBLEM
1.3 STATEMENT OF THE RESEARCH PROBLEM
1.4 AIM OF THE STUDY
1.5 RESEARCH QUESTIONS
1.6 SIGNIFICANCE OF THE STUDY
1.7 CONCEPTUAL DEFINITIONS OF TERMS
1.8 OPERATIONAL DEFINITIONS OF TERMS
1.9 FOUNDATIONS OF THE STUDY
1.10 SCOPE OF THE STUDY
1.11 STRUCTURE OF THE THESIS
1.12 CONCLUSION
CHAPTER 2 LITERATURE REVIEW
2.1 INTRODUCTION
2.2 HEALTH/CARE SEEKING BEHAVIOUR
2.3 THE ROLE OF MATERNAL HEALTH CARE
2.4 GLOBAL OVERVIEW OF MATERNAL HEALTH CARE
2.5 MATERNAL HEALTH CARE IN SUB SAHARAN AFRICA
2.6 MATERNAL HEALTH CARE IN ETHIOPIA
2.7 FACTORS AFFECTING MATERNAL HEALTH CARE UTILISATION
2.8 CONCLUSION
CHAPTER 3 RESEARCH DESIGN AND METHODS
3.1 INTRODUCTION
3.2 RESEARCH CONTEXT
3.3 THE RESEARCH DESIGN
3.4 RESEARCH METHOD
3.5 DATA AND DESIGN QUALITY: VALIDITY AND RELIABILITY
3.6 CONCLUSION
CHAPTER 4 ANALYSIS, PRESENTATION AND DESCRIPTION OF FINDINGS
4.1 INTRODUCTION
4.2 DATA MANAGEMENT AND ANALYSIS
4.3 EMPRICAL RESULTS
4.4 OVERVIEW OF RESEARCH FINDINGS
4.5 CONCLUSION
CHAPTER 5 DISCUSSION
5.1 INTRODUCTION
5.2 ADEQUACY OF ANTENATAL CARE IN ADDIS ABABA
5.3 USE OF ULTRASOUND SCREENING AND CAESAREAN-SECTION DELIVERY
5.4 PREFERENCES FOR PLACES TO GIVE BIRTH
5.5 FACTORS INFLUENCING ADEQUACY OF ANTENATAL AND DELIVERY CARE SERVICES
5.6 CONCLUSION
CHAPTER 6 PROPOSED FRAMEWORK FOR PROVIDING OPTIMAL ANTENATAL AND DELIVERY CARE SERVICES IN ADDIS ABABA
6.1 INTRODUCTION
6.2 PURPOSE OF THE FRAMEWORK
6.3 BASIS FOR THE DEVELOPMENT OF THIS FRAMEWORK
6.4 POLICY CONTEXT
6.5 RESEARCH CONTEXT
6.6 RECOMMENDED STANDARDS OF ANTENATAL AND DELIVERY CARE
6.7 INDICATORS EXAMINED IN THE CURRENT RESEARCH
6.8 ELEMENTS FOR IMPROVEMENT
6.9 PRIORITIES FOR ACTION
6.10 IMPLEMENTATION OF THE FRAMEWORK
6.11 RESOURCES
6.12 CONCLUSION
CHAPTER 7 CONCLUSION AND RECOMMENDATION
7.1 INTRODUCTION
7.2 RESEARCH DESIGN AND METHOD
7.3 SUMMARY AND INTERPRETATION OF THE RESEARCH FINDINGS
7.4 CONCLUSIONS
7.5 RECOMMENDATIONS
7.6 CONTRIBUTIONS OF THE STUDY
7.7 LIMITATIONS OF THE STUDY
7.8 CONCLUDING REMARKS
REFERENCES
GET THE COMPLETE PROJECT