As you will have briefly learnt in the introduction, there are three routes that you can follow when taking on a replication-based dissertation: Route A: Duplication, Route B: Generalisation or Route C: Extension. In the sections that follow, we explain what these three routes are. The purpose of STEP ONE is simply to help you get to know the main characteristics and terms associated within replication-based dissertations. This will help you when you come to choosing the topic for your dissertation later in The Route #1 Process. For now, just read through the routes, and thinking about your own dissertation, consider which of these routes may interest you. Unless you want to, or already have a dissertation topic in mind, you don't need to start thinking about a specific topic for your dissertation at this stage.
At the undergraduate and master's level, duplication is often not considered to be sufficient because there is generally an expectation that you should do something with at least an element of originality. However, there are valid exceptions to this, where duplication studies can become great dissertations.
As briefly discussed in the previous section, duplication means copying the original study in almost every way in order to see whether the same (or similar) results can be found. However, there can be different reasons to duplicate a study, which also affects how your dissertation will differ from the original study (if at all). Imagine the following two scenarios:
Sometimes a replication study involves no more than testing the data that was used by the original authors to see if the same results are found. We may choose to do this for a number of reasons:
Data analysis problems
The study you are interested in replicating has received little criticism in terms of the underlying theoretical arguments or research design used, but another study has questioned its findings, perhaps on theoretical grounds, or by collecting data that led to different findings. This leads you to ask the question: Was the data from the original study analysed properly? Now you may immediately think that you are not equipped to answer this question. After all, if the academic that published the research didn't do it correctly, how are you not going to make the same mistakes? Well the reality is that academics not only can make mistakes, but there can be a lot of judgement involve in the statistical analysis of data. Two researchers could analyse data in two different ways because of the way that the data was interpreted (e.g., whether you should include outliers, whether and how to group data before analysis, whether the data should be treated as normal, etc.). Whilst this may sound complicated, it often isn't. In any cases, we provide lots of articles in the Data Analysis section of Lærd Dissertation to help you understand what you need to think about when analysing data, how to identify mistakes, and how to carry out statistical analysis on your own.
New data analysis techniques and capabilities
Duplication-based studies are often not conducted straight after the original study has been published. It may have been 5 years, 10 years, 15 years, or even longer since the study of interest was published. In such cases, especially longer periods of times, new data analysis techniques may have emerged that can be used on the data (e.g., new statistical tests). After all, computing has come a long way in the last 1-2 decades, with statistical programs like SPSS helping researchers to analyse data easier and in more sophisticated ways. You may even find that the authors of the original study have identified an error in their data, but at the time of publication, had no statistical technique to address this problem.
Dealing with these kinds of problems, as well as using new statistical techniques to analyse the original data from a study can provide a useful addition to the literature.
This brings us to using new data. There is probably a 50% chance, if not greater, that you will not be able to obtain the original data used in the study you want to duplicate. If the study was published a long time ago, the data may have been lost. But more often than not, the reason will simply be that the authors refuse to give you access to their data. However, since this is not always the case, it is best to ask the authors before trying to get your own data.
Ironically, if you cannot get hold of the original data, your supervisor may view a duplication-based dissertation in a better light because you at least have to go out and collect the data on your own. If this is the case, you need to ask yourself two important questions before deciding whether you can do a duplication-based dissertation:
Is the research strategy of the original study clearly set out?
You can't duplicate a study if you cannot work out how the authors carried out their research; in other words, what research strategy they followed. By research strategy, we mean everything from the research design that was used, right through to the sampling strategy that was followed, the research methods, measurement procedures and measures that were adopted, as well as the data analysis techniques that were carried out. We explain all about research strategies and their components at a later date. For now, just think of it in a simple way: If the authors used a questionnaire, and the journal article does not include the questionnaire (i.e., the actual questions that were used and the measures for each of these questions), how can you duplicate the research? In such an instance, if the authors cannot (or will not) give you access to this questionnaire, you will not be able to duplicate their study. If this is the case, it would be worth thinking about either (a) finding another study you are interested in duplicating, but with a clear research strategy, or (b) moving on to Route C: Extension, where you adopt many of the goals of the study you are replicating, but have more freedom in terms of your selected research strategy.
Can I get access to a similar sample and population?
Duplication is not about using new samples and populations, but mirroring the samples and population used in the original study as closely as possible. So what do we mean by new samples and populations?
Again, whilst you can find out a lot more about samples and populations in the section on Sampling Strategy, the basic point is that a population signifies the units that we are interested in studying. These units could be people (e.g., students enrolled at a university, such as Harvard, or studying a particular course, perhaps Statistics 101; United States Senators or Congressman who are Democrats; users of Facebook or Twitter; nurses working at hospitals in the State of Texas, etc.), cases, whether organisations, institutions, or countries (e.g., law firms in Manhattan, New York, United States; the World Trade Organisation (WTO); countries that are members of NATO, etc.), or pieces of data (e.g., university applications in the United States in 2011; customer transactions at Wal-Mart or Tesco between two time points, e.g. 1st April 2009 and 31st March 2010; the breaking distances (in kpm/m) of a particular model of car, etc.). Whatever the population that was the focus on the original study, duplication-based dissertations must investigate the same population. If you cannot investigate the same population, perhaps because you cannot get access to it, or if you are interested in another population, you need to think about moving onto Route B: Generalisation, which specifically focuses on investigating different populations (as well as different contexts/settings). However, if you can investigate the same population, you need to think about the sample that will be used.
When we are interested in a population, it is often impractical and sometimes undesirable to try and study the entire population. For example, if the population we were interested in was frequent, male Facebook users in the United States, this could be millions of users (i.e., millions of units). If we chose to study these Facebook users using structured interviews (i.e., our chosen research method), it could take a lifetime. Therefore, we choose to study just a sample of these Facebook users. The important point to remember here is that a sample consists of only those units (in this case, Facebook users) from our population of interest (i.e., X million frequent, male, Facebook users in the United States) that we actually study (e.g., 500 or 1000 of these Facebook users). When it comes to duplication, you need to try and investigate a similar sample. This means looking to the characteristics of a given sample (e.g., of the 500 male Facebook users in the original sample, what were their ages, educational level, socio-economic background, etc.?). If the sample that you investigate when carrying out a duplication-based dissertation is different from the original study, it is possible that any differences in the findings that you obtain are due to the differences in the characteristics of the sample, and not the phenomenon that you are investigating. This is not always a deal breaker when choosing to do a duplication-based dissertation, but it is something that you will need to consider when it comes to assessing the quality of your findings.
Therefore, if the research strategy of the original study is clearly set out, and you can get access to a similar sample and population, you could choose to carry out a duplication-based dissertation using new data.
Whilst it's worth reading on to see if you would rather pursue a dissertation based on Route B: Generalisation or Route C: Extension, it's worth reiterating that if you want to pursue Route A: Duplication, you really do have to have a strong justification for doing so. If not, it is unlikely that this will be sufficient for an undergraduate or master's level dissertation.