What is research - Essay Example

What is Research Research is "something that people undertake in order to find out things in a systematic way, thereby increasing their knowledge."1 An important component of this definition is 'systematic', meaning research is approached logically. Research describes, explains, understands, criticizes, and analyzes a subject area. Research is important in business and management. Research must produce practical conclusions leading to better decision-making and/or profit.2 Managers draw on knowledge developed by other disciplines to aid their work routine, and tend to prefer to conduct research with personal or commercial advantage. In research, combining two knowledge bases sometimes creates a single large knowledge base plus a new one. 'Basic' research generates new knowledge and options, whereas 'applied' research generates new knowledge with a practical aim.3 New insights can help change the way business-related activities are conducted. Research Design Five research models will be examined: experimental design, cross-sectional surveys, longitudinal design, case studies, and comparative design. Experimental Design Good experimental design is important in producing reliable results. Independent variables should be manipulated to determine how they influence dependent variables. There are several forms of experimental research. Classic design is based on manipulation of independent variables, observations made before and after testing an experimental group, and relating observations to the dependent variable. The field experiment is used in experimental fields such as social psychology. The main advantage is that the researcher has no influence over experimental arrangements. Milgram's (1974) electric shock experiment or Zimbardo's prison experiments, for example, provide knowledge of how individuals and groups perform under certain conditions. Quasi-experiments are studies that have certain characteristics of experimental design but do not fulfil all internal requirements (e.g. control groups may be absent). Quasi-experimental design is useful in comparison and evaluation studies. Cross-sectional Design Also known as social survey design, this normally involves questionnaires and structured interviewing. Observation is a key feature of such research. Qualitative or quantitative data is collected at a single point in time, and is examined for patterns in relation to two or more variables. Longitudinal Design Longitudinal design provides more than one time-point for analysis, and is often used to plan changes in business and management. There are two types of design. A panel study is often a randomly selected national study which collects data from various types of cases at two or more time points. A cohort study is a randomly-selected sample of people who share certain pre-determined characteristics such as birth-date or employment status. Both types are concerned with revealing trends and improving the understanding of causal influences. They also share similar problems. Good planning is essential to provide useful data. Employee and company turn-over may cause sample attrition, and the remaining people may not form a representative group. Continued participation in such studies may influence how respondents behave. Case Studies A case can be an organization, location, person, or event. Stake (1995) observes that case studies are concerned with the complexity and nature of the case in question. There are three types of case study. The critical case is carried out with a clear hypothesis in mind, the unique case is used for clinical studies with a common focus, and the revelatory case is used "when an investigator has an opportunity to observe and analyse a phenomenon previously inaccessible to scientific investigation" (Yin 1984:44) Case studies can be conducted both deductively and inductively and may be linked with qualitative research using tools such as participant observation and unstructured interviews. Comparative Design Comparative design applies similar research methods to multiple contrasting cases, and can be applied in both qualitative and quantitative research to provide multiple levels of analysis. Such research takes the form of multiple-case studies which compare two or more cases with certain common characteristics. Literature Reviews Literature reviews help generate and refine research ideas, and provide a knowledge base with which to build a research topic. For most research projects, literature reviews occur in the initial phase where parameters for research questions and objectives are defined. Key ideas generated from initial searches are evaluated to record ideas and draft a review of research questions and objectives. It is important to review critically and use the full range of literature available in the reviewing process. Research Approach Deductive and Inductive Research Both deductive and inductive research have strengths and weaknesses which make them appropriate for use in certain situations, and may be used in different phases of the same study, depending on its requirements. The deductive approach starts with a general hypothesis which is refined as information is gathered, narrowing the study's focus onto areas of interest as defined by the data. Deductive research is often applied to questionnaires and other quantitative data. Deductive research is a structured approach based on scientific principles. It is easy to standardise, so the researcher need not be highly knowledgeable about the topic. The time-consuming phases are set-up and data collection, so the amount of time needed for such research can be scheduled and predicted easily. Deductive research is relatively low risk, however there are some concerns. Causal relationships between all variables in the research must be defined. Sample sizes must be large enough to generate conclusions that are reliable in all applications of the data. Abstract concepts must be clearly defined to reduce ambiguity for respondents, and increase reliability of data. One of the main disadvantages of deductive research is that it relies on subjects to complete and return questionnaires, which may become problematic as sample size increases. The methodology of such research tends to be inflexible, meaning that if data does not support the hypothesis, then a new hypothesis must be formed, and new research carried out. The inductive approach requires the researcher to have a good understanding of the context of the research. Gaining qualitative data is the primary focus, making personal interviews essential. Inductive research is flexible, allowing change in the focus of research. The researcher is part of the research process, enabling data-gathering to be dynamic and interactive. Because the focus is on qualitative rather than quantitative data, there is less concern with producing generalised conclusions. The research has a wider focus, creating an overall picture of relationships between variables. The main disadvantage of inductive research is that data collection and analysis is carried out over an extended period. The big picture often emerges gradually and it can be some time before useful data patterns emerge. Research Strategies Quantitative vs. Qualitative Strategies In the collection and analysis of data, quantitative research focuses on numbers, and qualitative research on words. Quantitative research collects quantifiable data that are used to form general conclusions about a larger population. Replication in different settings increases the validity of conclusions. It is important to consider differences between the social and natural worlds, and to make sure that theoretical data is relevant in the real world. Qualitative research is beneficial when good understanding of the context of research is required. However, it can be impressionistic and subjective and therefore difficult to replicate. Limiting generalisations is important in conducting such research. Fundamental differences between quantitative and qualitative research strategies: Quantitative Qualitative Deductive; testing of theory Inductive; generation of theory Positivism Interpretive Objectivism Constructivism Numbers Words Point of view of researcher Points of view of participants Researcher distant Research close Static Process Structured Unstructured Generalisation Contextual understanding Hard, reliable data Rich, deep data Macro Micro Behaviour Meaning Artificial Setting Natural setting Research strategy is also about developing an approach to answer specific research questions. Valid reasons must be clearly demonstrated for the research strategy that is chosen, for the questions that are asked, and for the subjects that are chosen for study. A good research strategy may use more than one type of research method. Surveys are usually associated with deductive research. Surveys are popular in business and management research because they can be done independently, and allow large amounts of data to be collected economically. Survey data is normally obtained through use of a questionnaire, and is standardised and easy to compare and understand. However, surveys can be time-consuming to organise and analyse, and the number of questions must be limited so the goodwill of participants is not taken advantage of, which limits the scope of the survey. Action Research focuses on gathering information and enacting change within an organisation. As Coghlan and Brannick (2001) note: "the purpose of research and discourse is not just to describe, understand and explain the world but also to change it". The researcher is involved in applying knowledge gained to implement change within the company. Employee involvement throughout the process is important, as they are more likely to apply changes they have helped create. Practitioner-Researcher: when the researcher works within the company, they have a significant advantage in their knowledge of the organisation and how it functions. However this can lead to assumptions and preconceptions when analysing data and forming theories. Care must also be taken to ensure that data-gathering is not inhibited when working with senior colleagues. Combining two different roles at work can also be very demanding. Questionnaires 'Questionnaire' is a general term describing data collection techniques in which subjects respond to a set of questions in a predetermined order. (deVaus, 2002) It includes structured interview and telephone questionnaires as well as written surveys. Questionnaires contain standardised questions that can be interpreted confidently in the same way by all respondents. (Robson, 2002) Questionnaires can be classified into two categories: structured questionnaires that involve fixed questions with pre-coded response choices (such as online, postal, delivery and collection, telephone survey), and semi-questionnaires that are defined by a more flexible interaction with respondents (such as an interview schedule). Both types must link back to research questions and objectives. Type of Questionnaire Structured Questionnaires have an unambiguous approach, enabling easy collection and analysis of quantitative data. This allows large numbers of respondents to be surveyed at once, making data collection economical and simple. However, the pre-coded responses in structured questionnaires means that some respondents may choose responses that do not entirely reflect their opinions, leading to errors in data collection. There is also scope for bias as respondents may sometimes desire to present a particular image in the answers they choose. Computer-assisted Questionnaires allow respondents to type in their answers on a computer screen. Inputting each answer prompts the display of the next question in the series. Interviews Interviews allow for a better understanding of the context of data and more flexible interaction with subjects. More complicated questions can be asked, and more information and in-depth understanding can be obtained. Inconsistencies and misinterpretations can be avoided, as clarification can be asked for by either party if needed. Response rate is generally higher than with other questionnaire methods (such as postal or telephone). However, such methods can be costly and time-consuming, and a well-trained data analyst is essential to reduce potential bias. If interviews are well-designed, highly accurate data can be obtained. Data collection Three types of data can be generated by participant observation. Primary observations are made by the researcher, who notes events as they occur. Secondary Observations are statements made by others and recorded by the researcher. Interviews may be used as a tool of informal discussion here. Questions may be asked of respondents to clarify observations and of the researcher to clarify what is required of the respondents. Experimental data are gained from evaluating research processes and data. Data can be classified in two ways. Descriptive observation concerns physical setting, key participants and their activities, attendant processes and their emotions. A narrative account is how a researcher develops a framework of theory to enable understanding and explanation of the research setting under observation. Figure 1: the literature review process. Sampling Sampling techniques enable reduction of the amount of data required by considering only data from subgroups of interest rather than all possible cases or elements. Sampling techniques can be divided into two types. Probability (representative) sampling: Participants are randomly selected from a larger population, allowing estimation of characteristics of the population the sample represents. Non-probability (judgemental) sampling: Participants are not selected entirely at random; therefore the larger population from which the sample is drawn may not be accurately represented. Figure 2: sampling techniques Statistical analysis Raw data must be collected and analyzed before it can be interpreted. Bowling (2002) defines evaluation as "the use of scientific method, and the rigorous and systematic collection of research data to assess the effectiveness of organizations, services and programs in achieving predefined objectives."4 Saunders (2003) suggests a four-step approach. 1) Preparing data for analysis Some data are more easily measured than others. Values of categorical data are not measurable numerically but can be classified or ranked according to the researcher's interests. They include nominal5, ordinal6, interval7 and ratios8.9 Quantifiable data can be measured numerically and are separated into continuous10 and discrete11. Sometimes categorical data cannot be input into a computer program. This can be overcome by data coding, which also provides quicker input and fewer errors. Inductive or deductive12 coding may be used. The former is when codes are developed after the study is taken, the latter when codes are set before the study is conducted13. Obtaining an accurate overall picture requires taking into account differences in collection processes, known as weighting the data,14 and checking for errors in data entry15. 2) Choosing appropriate figures to explore and present data Types of data Presentation Data with specific values Tables When highest and lowest values are important Multiple bar charts, column charts, histograms Trends over time Line graphs Proportions Pie charts, percentage component bar charts Distributions Histogram for continuous, bar chart for discrete - look for positively or negatively skewed or symmetrically distributed, box plot Conjunctions Point where 2 or more variables intersect Totals Stacked bar chart Interdependence Contingency table or cross-tabulation, chi square Relationships Scatter graph, dependent and independent variables 3) Choosing appropriate statistics to describe data Statistics describe central tendency16 and dispersion17. Measures of dispersion are suitable for quantifiable data but can be used in categorical data if there are numerical codes. 4) Choosing appropriate statistics to examine relationships and trends in data Statistical significance is a systematic way of ruling out the possibility of inaccurate results due to random variation in the sample. It tests the likelihood of relationships occurring by chance alone and comparing it to sample data. Incorrect interpretation of statistical significance can lead to error18. Probability (p-value) reflects the likelihood of extreme variations occurring by chance. If the p-value is 0.05 or lower the relationship is statistically significant. If the probability is higher than 0.05, the relationship is not statistically significant. The Chi squared test19 is used to determine if two variables are related. Once a relationship between two variables is demonstrated, Cramer's Phi20 is used to test the strength of the relationship. The correlation between variables which are strongly linked can be estimated using Tendall's Tau-b21. Finally, if the sample size is small, probability of possibilities and associations between variables should be tested using Fisher's Exact.22,23 Index numbers24 are calculated to compare short-term trends. For trends over time, a moving average for a time series of data values is used. Research Ethics It is important for researchers to receive informed consent from respondents before a study is conducted. Respondents must be informed of procedures and possible outcomes of the research25, privacy protection, and the right to withdraw from participation. Respondents should watch for unethical practices such as fraud and seek legal protection if abnormality is detected. Evaluation of research Evaluation is used to determine the value of data. There are two evaluation types: formative and summative26. Formative evaluation seeks to strengthen or improve the object of examination, while summative evaluation examines effects or outcomes related to it. 27,28 Ovretveit (1998) recognizes three steps to evaluations29, its advantages, and limitations. Advantages Limitations Inform others to carry out changes (evaluation for action) Cannot carry out changes Give information that can attribute values for others Cannot attribute value Clear criteria to judge value of the evaluated Cannot include all the criteria people need to consider how to act Not expensive and long randomized controlled trial, simply description of intervention and implementation Results may not be as accurate as RCTs Effects of the evaluated looked at Not all look at effectiveness Managers can improve their performance by learning applicable evaluation techniques and critically assessing evaluations conducted by others. Evaluation can help allocate resources rationally, protect businesses by valuing efficiency over politics, and make better forecasts of impacts of interventions30. Recommendation Evaluating employee morale will require use of questionnaires and interviews. Longitudinal studies may be useful to examine changes over time. Case-study comparisons of multiple companies may also be useful. Morale is a personally-defined concept; therefore concepts and response scales must be clearly defined to ensure accurate responses. For example, a satisfaction-level scale must be established, and reasons for levels of satisfaction must be thoroughly examined to include all possible responses in the questionnaire. It may be beneficial to carry out both quantitative and qualitative research within a larger study on employee morale in a multi-site firm. A general survey could gather initial data, followed by interviews to focus on key issues raised by survey results. Practitioner-research would be of benefit to permit deeper understanding of the firm, its employees, and the possible consequences of solution strategies. The use of multiple methods helps ensure accurate interpretation of data. Each method has strengths and weaknesses (Smith 1975), and using multiple methods cancels out the 'method effect', allowing more accurate conclusions to be reached. For example, studying employee morale in a multi-site firm data may indicate relationships between morale, workload and salary. Initial data can be used to refine the hypothesis and new data can be collected for analysis. This may include input from multiple departments, to increase sample size and note the effects of varying workloads or salaries, thus increasing accuracy of results. A larger sample size allows further refinement of the hypothesis - it may narrow to focus on young employees, and allow the researcher to pinpoint effects of low or high morale, such as absenteeism. An initial hypothesis concerning morale can therefore be refined to pinpoint causal relationships that become evident as the focus of the research narrows. Chi square, Cramer's Phi, and Kendall's Tau-b should be used to analyze the roles of each variable across multiple sites. A large multi-site firm can provide a rich mix of data, making it unnecessary to use Fisher's Exact. Appendix I: Research Functions Basic Research Applied Research Generates new knowledge and options Generates new knowledge with a practical aim Understanding theory Developing tools and simulations Developing instrumentation and measurement techniques Tracking and absorbing external knowledge Appendix II: Types of Evaluation (Trochim 2004) "Formative evaluations strengthen or improve the object being evaluated -- they help form it by examining the delivery of the program or technology, the quality of its implementation, and the assessment of the organizational context, personnel, procedures, inputs, and so on. Summative evaluations, in contrast, examine the effects or outcomes of some object -- they summarize it by describing what happens subsequent to delivery of the program or technology; assessing whether the object can be said to have caused the outcome; determining the overall impact of the causal factor beyond only the immediate target outcomes; and, estimating the relative costs associated with the object." (Trochim, 2004) Formative Evaluation needs assessment determines who needs the program, how great the need is, and what might work to meet the need evaluability assessment determines whether an evaluation is feasible and how stakeholders can help shape its usefulness structured conceptualization helps stakeholders define the program or technology, the target population, and the possible outcomes implementation evaluation monitors the fidelity of the program or technology delivery process evaluation investigates the process of delivering the program or technology, including alternative delivery procedures Summative Evaluation outcome evaluations investigate whether the program or technology caused demonstrable effects on specifically defined target outcomes impact evaluation is broader and assesses the overall effects - intended or unintended - of the program or technology as a whole cost-effectiveness and cost-benefit analysis address questions of efficiency by standardizing outcomes in terms of their dollar costs and values secondary analysis re-examines existing data to address new questions or use methods not previously employed meta-analysis integrates outcome estimates from multiple studies to arrive at an overall or summary judgment on an evaluation question Appendix III Coding Categorical Data 1. Existing coding schemes such as industrial classification, occupation, and socioeconomic group should be used as much as possible because they save time, are normally well tested, and allow comparison of results with other surveys. 2. Coding at data collection. 3. Coding after data collection is necessary when the likely responses are unclear. 4. Coding missing data may be necessary due to a number of factors. For example, respondents may not answer certain questions due to lack of knowledge or ambiguity in the question. Appendix IV Appendix V Saunders (2003) suggest that we follow a 4-step approach. 1. Calculate the percentage of the population responding for each stratum 2. Establish which stratum had the highest percentage of the population responding 3. Calculate the weight for each stratum using the formula (weight = highest proportion of pop responding for any stratum/ proportion of pop responding in stratum for which calculating weight) 4. Apply the appropriate weight to each case. Bibliography Bowling, A. 2002, Research methods in health - Investigating health and health services, 2nd edn, Open University Press, Great Britain. Bryman, A. & Bell, E. 2003, Business research methods, first edn, Oxford University Press, Great Britain. Ovretveit, J. 1998, Evaluating health interventions, first edn, Open University Press, Great Britain. Trochim, William M. The Research Methods Knowledge Base, 2nd Edition. Internet WWW page, at URL: (version current as of Aug 16, 2004). Saunders, M., Lewis, P. & Thornhill, A. 2003, Research methods for business students, 3rd edn, Prentice hall, Financial Times, Italy. Read More