Consideration 2: Stick to your research plan, but be prepared to be flexible with collecting data

One of the major benefits of quantitative research is that the dissertation process tends to be very formulaic; you can set out your research hypotheses, decide on a suitable research strategy to collect data, using research methods that are highly standardized with pre-determined measures and fixed-options, and then analyse that data using tried and tested statistical tests, with the results from these tests directly answering your research hypotheses. As an undergraduate or master's students, this is a major advantage compared to dissertations that involve qualitative research.

During the data collection phase, you need to take advantage of this by making sure that you're well prepared. There are a number of areas to make sure that you are well prepared before you start collecting data, especially (a) the measurement procedure and (b) the sampling strategy you are using:

• The measurement procedure

  Remember that a measurement procedure is a lot more than just the questions and fixed-choice options in a survey or structured interview, or the observation schedule in a structured observation. It involves the statement that explains to the participant what the research is about, how the data about them will be stored, and what their rights are. In the case of face-to-face surveys and structured interviews, it tells you how to respond if a participant does not respond according to the fixed-choice options you have given them. In the case of structured observations, it may include cue cards to help you identify behaviours that can be written down (i.e., coded as observation points) according to the observation schedule that you have created. Before starting any data collection, you need to make sure that all of these aspects of your measurement procedure have been prepared so that the way that each participant engages with your research is the same. Any inconsistency in the way that participants are treated because you are not well prepared threatens the internal validity of your results.

  Another important consideration is to be flexible depending on what happens when you start to collect data. Sometimes a measurement procedure, even when shown to be reliable in previous research, is not the most practical in the field. For example, there have been many journal articles published that do nothing more than take a well-established measurement procedure (e.g., a 42 question survey on depression), and try and create a new measurement procedure by reducing the number of measures used (e.g., condensing the 42 questions to 19, for example). In such cases, the idea is to create a shorter measurement procedure that will take less time for participants to complete, which can mean that more participants are willing to complete your measurement procedure, increasing your sample size and statistical power (i.e., your ability to find statistical significance if it exists, something that we explain more about in the Data Analysis part of Lærd Dissertation). Therefore, if you find that too few people are willing to complete your measurement procedure, or that a large proportion of those that take part in your research early on are dropping out early, you need to be flexible to the idea of changing some parts of your measurement procedure. Of course, you have to be careful when doing this, and ideally, only do it early on in the data collection process because any of the data you have collected will be unusable (i.e., because the measurement procedure that those early participants received is not the same as the one future participants are using, which means you cannot compare the scores from these two groups of participants in a like-for-like manner).

• Your sampling strategy

  Like with your measurement procedure, you need to be prepared to be flexible with your sampling strategy. For example, you may have planned to use a probability sampling technique, believing that you can get access to a list of the population you are studying, but this proves too difficult when you try and build such a list in the field (i.e., it is often a lot easier to plan to get such a list than actually getting hold of one). This could mean that you have to use a different probability sampling technique (e.g., switching from a stratified random sample to a simple random sample if you are unable to get a list with the stratifications you need, or more likely, using an alternative non-probability sampling technique that involves stratification, such as a quota sample). Making such a change does not necessarily involve a lot of work, unless you?re changing from a non-probability sampling technique to a probability-based one, but this is unlikely. All you need to do is consider the implications of such a change for the internal validity of your findings when you come to write up.

Overall, when it comes to the data collection phase of your research, be well prepared and stick to your research plan, but be ready to make changes if problems arise.