Research Methods 09 Observational Research (Exploratory studies)
Dr. Syed Muhammad Sajjad
Department of Cyber-Security and Data Science Riphah Institute of Systems Engineering (RISE), Riphah International University, Islamabad, Pakistan. Observational Research • Observational research, of which exploratory study is a subset, is useful • when you are trying to understand a real cyber system (techno‐social behavior) • This type of research is suited to answer open‐ended or comparatively broad research questions • In general, observational research methods include sensing of real world environments and data mining for discovery of interesting artifacts. Observational Research • Sagarin et al. provide an insightful discussion on the limits and strengths of observational research in their book 1. Observation‐based studies find patterns, but patterns can’t be used to infer process 2. Observation‐based studies rely on the flawed approach of induction, rather than the more precise deductive approach, to reach conclusions 3. They are just a collection of un‐replicated anecdotes 4. They rely too much on correlations between variables Exploratory Study • Exploratory studies are either datasets that are not in the immediate control of the investigator, • Often the scope of the study and the dataset is much larger • Descriptive Studies, are focused on preconceived behavior and a goal of describing the phenomena • Descriptive studies typically have more interaction and control of the subject at hand Exploratory Study • Exploratory studies consist of collecting, analyzing, and interpreting observations about known designs, systems, or models, or about abstract theories or subjects • Studies are largely an inductive process to gain understanding • Where the experimental process goes from a general theory to an understanding in specific, • Studies look at specific phenomena to search for patterns and arrive at a general theory of behavior Exploratory Study • The emphasis is on evaluation or analysis of data, not on creating new designs or models • This is often called a qualitative study in the social and health sciences • When you have no expectations or beliefs about how a system operates • But you want to study and understand it, exploratory study is a helpful method Exploratory Study • By observing the world around us we can start to discover patterns and sequences of events • That can lead to understanding the underlying principles and laws of behavior • Exploratory studies enable researchers to start developing models of system operation Exploratory Study • As some aspects of cyber security is practically uncontrollable, such as threats, • When exploring topics and research questions around uncontrollable variables • Exploratory studies are a useful approach to discover information about a system Types of Exploratory Study 1. Case Control Study 2. Ecological Study 3. Cross‐Sectional Study 4. Longitudinal/Cohort Study Types of Exploratory Study 1. Case Control Study • Type of observational study where a set of measurements can be divided into two groups • The first group is called the case group exhibits the signs or symptoms of the event or phenomena under study • The second group, the control group, does not exhibit these systems • For example, a researcher may want to look at the rare cancer. The researcher would find a group of individuals with cancer(the cases) and compare them to a group of patients who are similar to the cases in most ways but do not have cancer(controls) Types of Exploratory Study 2. Ecological Study • Explore the entire population in either geographic or temporal space (in a region or at a particular time) • The goal of an ecological study is to assess the risk factors confronting a cyber population by evaluating risk mitigations and the subsequent adverse outcomes that affect the population Types of Exploratory Study 3. Cross‐sectional Study • Sometimes called a census • Cross‐sectional studies are related to, and sometimes a branch of, ecological studies in that a time‐slice sample is collected from a population • A sample is collected from the entire population at a specific period of time • Cross‐sectional studies are often used to describe the prevalence of certain • Characteristics of the overall population or enable deductive inferences to be made about the information at hand Types of Exploratory Study 4. Longitudinal/Cohort Study • Is a sequential observation of a cyber system or cyber behavior over time • The goal is to collect sufficient information over time to ensure the entire lifecycle or complete context of the system under evaluation is considered • The evolution of viruses over time • A cohort study is a type of longitudinal study that focuses on a specific group, over time, called a cohort GATHERING DATA • At the point of data collection, a researcher should have a sufficient handle on the research questions and approach they will be undertaking • If you have insufficient clarity to start with, then you run the risk of gathering data that does not end up being relevant to your line of inquiry • From questions about collection methodologies, to formats and syntax, the task of collecting sufficient data, or even knowing when to stop, can dwarf the labor involved in the subsequent analysis. GATHERING DATA To avoid this trap, as much as possible, a well‐crafted research plan will alleviate several of these pitfalls, start by exploring a few questions to help identify the scope, objectives, and likely challenges in data gathering • Have you reviewed your idea with your REC (Research Ethics Committee) informally to seek guidance or feedback? • What types of questions are you asking (qualitative vs quantitative, relationship, volume or routine occurrences)? Clearly the questions must be observable, but the dataset may not exist. • How much data will you need? Do you know what amount will be statistically significant for the conclusions or answers you hope to address? GATHERING DATA • Who owns the dataset? Privacy, access, control, dissemination, and other issues will need to be addressed beforehand • Can you easily recollect or recreate the dataset? Or are you stuck with the dataset from a single point in time? Are you able to come back and revisit your initial assumptions if they do not pan out? • Are you sampling? Will you be able to collect the entire population or dataset? This is very common in network traffic. If you have to sample, what is the method? What artifacts or challenges might it introduce? • Have you reviewed the full study plan with your REC (Research Ethics Committee) ? GATHERING DATA • Broadly there are two types of data: qualitative and quantitative. • Qualitative research includes collection and analysis of descriptive data. • Research involving humans often includes information about their emotional state and social characteristics. • It can be categorized and sometimes ordered but does not provide the ability to mathematically quantify the data. • Quantitative research involves the collection and analysis of numerical data. • It enables the quantification or statistical exploration and explanation of data. • Quantitative provides the most flexibility in analysis and should be sought above qualitative when possible.