Cross sectional studies are mainly based on interviews with selected groups during focused period of data collection. Here, information is collected at a single point in time. Whereas longitudinal studies are based on repeated measures as sample is taken at different time points to observe changes in trends. It can also be a cohort or panel study where certain groups are followed up. This type of study provides a time order of events and demonstrates causality but requires more time and money.
A comparative study comprehends phenomena by drawing a contrast between them in different contexts. It is based on complexity and comparability. It can either be used to compare two or more cases across different cultures or highlight differences and similarities between them.
Case studies are centred on case selection and involve intensive study of a single case. They can also be used to compare different cases. Cases can be of five types. Critical cases are grounded on a well-developed theory and hypotheses and subject is chosen to test it. Extreme or unique cases, as the name suggests, involve selection on the basis of being atypical. Typical cases, on the other hand, are selected for being representative of a particular category. In revelatory cases, the researcher observes and analyses a new phenomenon. Lastly, when cases are studied over time, they are called longitudinal cases.
A research design constructed more on a qualitative basis is ethnography. This type of study is closely linked to case studies but is differentiated due to the time and personal involvement of researchers. It primarily entails participant observation through interviews, attending events, visual recordings, documents and material. As the recording of reflections play a significant role in this study, field diaries are essential. Another research design particular to qualitative research is grounded theory. This type of study is built upon an inductive or bottom-up approach as it provides a means of generate theories from data.
The entire set of people or entities that we want to describe is called the population of the research. This has to be clearly defined. All the members of the population are referred to as the census. This gives us an accurate overview of the population and is essential for policy making but it is also expensive and laborious. A list of all the units in the population is the sampling frame and the subset of the population that we target is known as the sample.
Sampling can be of two types, probability and non-probability. In probability sampling, every unit of population has a known probability of being selected. We can generalise from our sample to the population and it is possible to make inferences and draw conclusions from our sample about the population. Probability sampling an also be of three kinds, simple, stratified and multistage or cluster. In simple sampling, each unit of population has some inclusion probability and respondents are randomly selected. This removes any human bias but it is possible to get a biased sample by fluke. Stratified sampling identifies strata within population and then randomly selects units within each stratum. It requires sampling frame including information about the stratifying variable. Division of strata can be based on age, gender, ethnic group, etc. based on the topic of research. This method ensures reflection of population distribution in selected criteria and also allows purposeful oversampling in certain groups depending on the research question.
Probability sampling often leads to sampling error, which is an inevitable part of the sampling process and is manifested when a statistic is based on the sample not the population. Nevertheless, the error is systematic and quantifiable, thus, it can be accounted for.
In non-probability sampling, there is no generalisation or inference and the probability for each sample in unknown. The characteristics of the population are used as the basis of selection and cases are deliberately selected to reflect particular features or groups within the sampled population. Non probability sampling can be of three kinds. The first types if opportunistic or convenience sampling which lacks any clear strategy and is based on ease of availability or accessibility. This method is easy and fast but he sample is usually not representative as researchers are likely to sample more approachable people and there is volunteer bias as well. Snowball sampling is used to target hard to reach populations where a frame is missing. In this method, researchers contact ‘seeds’ who recruit additional participants from their network. It is useful for targeting hidden or deviant people within well networked groups but the sample is not representative as it is based on convenience and willingness of ‘seeds’. This type of sampling may also lead to homophily bias as people tend to know people similar to themselves. The third type of non-probability sampling is called quota or targeted sampling. This aims to include certain types of participants or get variability across sample. It is considered a widely employed response to the deficiencies in chain referral sampling and is commonly used by market researchers.
Purposive sampling selects units or people based on particular features or characteristics that enable detailed exploration of central issues and themes that the researcher wants to study. Participants are selected with a purpose, in direct reference to the research question and is also called criterion sampling. It facilitates the study and helps answer the research question but may lead to a research error if subjects that are recruited are not representative of group. Purposive sampling can be theoretical and generic. In theoretical sampling, cases are sampled on the basis of their potential contribution to the development and testing of theoretical constructs. This selection based on the relevance to the research question continues until theoretical saturation is reached. The criteria for the sampling process of generic sampling is determined by the research question. These criteria are decided prior to the sampling process and remains fixed throughout. A sample is chose from the identified appropriate cases.
...(download the rest of the essay above)