STEP TWO

Justify taking on a replication-based dissertation

When considering the justifications for replication-based dissertations, you should consider (a) the spirit of replication, (b) general justifications for replication-based dissertations, and (c) specific justifications for replication-based dissertations.

CONSIDERATION ONE

The spirit of replication

There are a number of reasons why we choose to replicate research, whether we are referring to replication in general, or a specific type of replication (i.e., duplication, generalisation and extension). In a way, you can view these justifications as those that involve checking research in some way, and others that aim to build on previous research:

• Checking previous research

  There are countless journal articles that have been published, building a vast body of knowledge that we rely on in all walks of life. However, once a piece of research is produced and published, it is rarely replicated. Seldom do researchers check the quality of a previous piece of research. The findings from a study are too often taken-for-granted. Imagine all the knowledge that has accumulated in the world, and all those studies that have built on previous studies that have not even been checked. For this reason, reproducibility is one of the cornerstones of good science (whether in the social or physical sciences). It helps to stop the spread of false knowledge. This is the focus of Route A: Duplication, but is also a part of Route B: Generalisation and Route C: Extension, since these also involve the testing of the findings from the original study that is being replicated.

• Building on previous research

  Stopping the spread of false knowledge is important, but so is the creation of new knowledge. In this respect, replication is not just about reproducibility and checking, but also using previous studies as a platform to launch the pursuit of new knowledge. This is the focus of Route B: Generalisation and Route C: Extension, which aim to not only test previous findings, but also either see how far they can be generalised (i.e., Route B: Generalisation), or modified and extended in some way (i.e., Route C: Extension). Generalisation is one of the main goals of quantitative research; that is, testing to see whether the findings from the study you are interested in replicating hold across a range of populations, settings/contexts, treatments and time. Extension allow you to test previous studies whilst trying to improve aspects of the original study, as well as test new, but related knowledge that may add greater understanding to the original study.

CONSIDERATION TWO

General justifications for replication-based dissertations

Three of the general justifications for replication-based dissertations that are worth considering are the desire to (a) improve the internal validity of a study, (b) test the generalizability of a study, and (c) build the construct validity and reliability of a measurement procedure.

Improving the internal validity of a study

Internal validity is important because we want to be able to say that the conclusions made in a piece of research accurately reflect what was being studied. For example, if a study concludes that exercise reduces heart disease, the authors want to make sure that they can say this with as much confidence as possible, believing that what was studied, and not other factors, especially factors that were not included in the research, explain the results. Whilst a number of factors play a role in improving the internal validity of quantitative research, the way that the research is designed is crucial. However, even good research designs cannot account for all of the many threats to internal validity that researchers face. These include history effects, maturation, testing effects, instrumentation, selection biases, experimenter and subject effects, amongst other threats we discuss in detail in the Research Quality section of Lærd Dissertation.

Since there are so many potential threats to internal validity, it is possible that at least one of these could have affected the original study in a way that questions the findings that were presented. Some of these will be unknown, whilst others are more obvious. Unknown threats to internal validity include factors such as experimenter effects, which occur because of (a) the personal characteristics of the researcher that influences the choices made during a study, and (b) non-verbal cues that the researcher gives out that may influence the behaviour and responses of participants. However, you won't know whether such factors threatened the results of the study you are interested in because you were not there when the research took place, and the potential for such problems are rarely reported in journal articles. Alternately, there are those threats that are known, such as sampling bias, which is often either easier to spot in journal articles, and sometimes even reported by the authors.

Irrespective of whether these threats to internal validity are obvious or reported by the authors, they present a good justification for carrying out a replication-based dissertation. After all, you may be able to improve some of the more obvious threats to internal validity when carrying out your dissertation, such as reducing sampling bias. If your findings are the same (or similar) to those in the original study, improving the internal validity of your study will give greater confidence not only to your findings, but also those of the original study. On the other hand, if your findings are different to those in the original study, you may be able to show that it was the flaws in the original study (i.e., the threats to its internal validity) that explain the results that were found, whilst highlighting that more studies should be conducted to replicated the findings in your dissertation.