Time

The lesson takes around 20 minutes to complete.

About the lesson ‘Good data collection practices’

What do you need to think about when you collect data for assignments and projects?

Before you start collecting data, there are certain things you need to think carefully about. If you collect incorrect data, you won’t be able to carry out the analysis you’re interested in. If you collect data you don’t need, you waste time that could have been spent on something more useful.

In this lesson, we’ll look at certain factors you need to pay attention to while collecting data: your purpose in collecting data, delineation of the research problem, and neutrality in question formulation.

The lesson ends with an introduction to data collection and organisation in practice.

Please note: This lesson combines practical advice and recommendations for epistemological considerations. Obviously, theory and practice will vary from discipline to discipline, so you should treat the following as a general introduction.

Learning outcomes

When you have completed the lesson, we expect that you:

are familiar with some of the factors affecting data collection, including the methodological purpose of data collection, delineation of the research question and the quantity of data.

can recall how to structure data (tidy data), including metadata/documentation allowing data to be reused.

Source

The lesson was produced by the University of Copenhagen as part of its learning resources for digital literacy, 2023.

Course directors:

Christian B. Knudsen, specialist consultant, Copenhagen University Library (KUB)
Daniel H. Pryn, information specialist, Copenhagen University Library (KUB)

__________________________________________________________

Purpose of data collection

The collection of data for a project may serve different purposes depending on the specific study or survey, and the questions to be answered. Some important steps in an academic study involving data collection are presented below. Not all projects involve all the steps, and some projects will have many more steps.

Nor will the process always be linear – in fact it will usually be iterative, with earlier points being revisited and adjusted – but if we try to illustrate the process in steps, it will usually look like this:

Graphic showing the process in a project involving data, divided into eight phases: 1 Problem formulation – 2 Study design – 3 Validation – 4 Data collection – 5 Data cleaning – 6 Data structuring – 7 Analysis – 8 Communication/visualisation.

One purpose of data collection, for example, could be to demonstrate generalisable results, in other words to study observations for a large group of individuals or phenomena. In this case, the purpose of data collection is to provide an overview of a particular population and identify possible data patterns or trends.

When writing assignments, you’ll often find that the purpose of data collection is to demonstrate that you know how to collect data, rather than actually collecting sufficient data to draw conclusions from. In this situation, the purpose of data collection is not so much to provide a basis for answering a question in an academically valid way, but to show that you are able to collect data accurately, reliably and ethically in a particular context. This not only teaches you how to collect data but also trains you to document your process and your methods.

Whether or not the purpose is to demonstrate generalisable results, it is important for university students like you to focus on correct data collection practices every time you collect data. This is crucial in order to ensure that your results are valid and reliable, and that they can be used to generate knowledge about a particular topic.

__________________________________________________________

Delineation of research question

The delineation of the research question is a process of formulating specific and clear questions concerning the topic being studied, as well as delineating the study itself (for example, in terms of a specific period of time, geographical location or population group) to ensure that the data collected are reliable and relevant to the study. This is important because it helps you focus on a specific topic and decide what data are useful and relevant for the study.

It can be difficult to formulate clear and specific questions as this requires a thorough understanding of the topic being studied, as well as requiring you to consider what data will be relevant and useful in order to answer the question you are formulating. It can also be a challenge to delineate the study in terms of a specific time period, geographical location or population group, as this requires you to consider what data will be relevant and reliable for the study. Even so, it is an important process, as it helps to ensure that the collected data are relevant and useful for the study, and that the questions you formulate can be answered by the collected data.

Examples of poorly delineated research questions

What can be done to increase the growth of tomato plants?
This question is poorly delineated because it is not specific enough. For example, it does not specify what kind of growth is to be increased – is the aim to increase the size of the plants, the number of leaves or the quantity of fruit? This will make it difficult to collect relevant and useful data that can answer the question.
What are the expectations of AI in Danish public digitalisation strategies 2017-2022?
This question is poorly delineated because it does not specify the theoretical meaning behind “expectations”. Unless this is specified, it will be impossible to identify and categorise the texts being analysed. The selection of data is partially delineated in terms of years, in particular, but is poorly delineated because it is unclear whether the question means digitalisation strategies in public-sector organisations or, more broadly, published digitalisation strategies in general.
How does gardening affect people’s quality of life?
This question is poorly delineated because it is not specific enough. It does not specify what kind of gardening or how much. Nor is there any definition of what is meant by quality of life. This will make it difficult to collect relevant and useful data that can answer the question.

Examples of well-delineated research questions

How do different irrigation amounts affect the weight of tomatoes?
This is a well-delineated research question because it focuses on a specific factor (the amount of water), which is controlled, and a specific type of plant (tomatoes) and asks how they relate to each other. This allows data to be collected that directly answer the question and help to explore how this particular factor affects the weight.
On the basis of Joseph Berger’s expectation states theory and the concept of performance expectations, what are the expectations of the use of AI in digitalisation strategies in Jutland municipalities in the period 2017-2022?
This is a well-delineated research question because it specifies the theoretical meaning behind the concept of “expectations”. In addition, the period and the digitalisation strategies are clearly delineated. This allows data to be collected that directly answer the question and which help to bring about an understanding of which factors to focus on.
How do different amounts (measured by time) of different types of gardening at different frequencies affect self-reported stress levels and the concentration of cortisol in the participants’ saliva?
This is a well-delineated research question because it is highly specific in terms of the scope, type and frequency of gardening, and specifies the ways in which stress levels are measured. This will allow data to be collected that can directly answer the question and help to bring about an understanding of how the specific activity affects both the participants’ subjectively perceived and objectively measurable stress levels.

__________________________________________________________

What data to collect?

As mentioned above, it’s important to consider what the data will be used for, as this will affect what data you need to collect and how you should collect it. Once you know what you are going to use the data for, you can more easily select the most appropriate methods of measurement and ensure that you collect the right amount of data for this purpose.

For example, if you want to study the relationship between inflation and unemployment in different countries, it will be important to collect data about employment and about price changes. However, it will also be important to collect data on other factors that may affect these parameters, such as interest rates and levels of industrialisation in the various countries.

Once you’ve decided what data to collect, it is also important to choose an appropriate data collection method. For example, you could use questionnaires, databases, interviews or laboratory measurements.

See also the section about causation and correlation in the lesson ‘Data analysis’.

__________________________________________________________

Neutrality

When collecting data, it is important to consider the concept of neutrality, whatever method you end up using. We’re talking about neutrality in terms of the selected population and also in terms of how you collect and analyse your data. It may be a good idea to consider how you can design your study in a way that minimises the possibility of bias, for example, by choosing a representative population or by using appropriate data collection methods. It may also be a good idea to cast a critical eye over any prejudices you might have, and take time to reflect on them as the study progresses.

See the video about the “IF I APPLY” method in the lesson ‘New methods for digital source criticism’ – more specifically, the first ‘I’ which stands for “Identify your own emotions”.

No matter what method you end up using, you should make sure that your research questions or your study protocols do not contain biases or prejudices, and that they are formulated clearly and intelligibly. It may be a good idea to review the questions or the protocols with a critical eye to ensure that they cannot be misconstrued or interpreted in more than one way.

It may also be a good idea to test the questions or the protocols on a small group before going ahead with data collection, to ensure that they are formulated correctly and intelligibly. This can give you an indication of whether the questions or protocol are formulated appropriately, and whether there is a need to adjust them before you start collecting data for real.

When you collect data from measurements such as chemical analyses or biological observations, it’s important to ensure that the method is suitable for the purpose of the study. For example, consider whether there are other methods that could be used to obtain the same data. You should also make sure that the measuring equipment and/or methods are calibrated and valid, and that any measurement uncertainty is accounted for.

__________________________________________________________

How much data to collect?

If you are collecting data to demonstrate general trends, a large amount of data is often required in order to ensure that the results are representative of the population you are investigating. Normally, the larger the amount of data, the more certain you can be that the results can be generalised to the whole population. However, it is important to be aware that a large amount of data does not always mean reliable results. It may also be necessary to consider how the data were collected in order to ensure that they are representative and reliable.

Generally speaking, the amount of data (the number of observations) collected will be greater in quantitative studies than in qualitative studies. It is appropriate to collect a large amount of data in quantitative studies to guarantee that the results are representative and reliable, whereas in qualitative studies a smaller amount of data will often be collected, which will then be examined in depth.

It’s important to note that this is only a general tendency which does not apply to all quantitative and qualitative studies. There may be examples of quantitative studies that collect a small amount of data, such as a questionnaire with a limited number of participants, and of qualitative studies that collect a large amount of data, such as in-depth interviews with a large number of participants, or collecting a large number of observation notes. What is the right amount of data to collect always depends on the purpose of the study and the questions to be answered.

For you as a student, the purpose can sometimes be more about documenting the methodical process than demonstrating generalisable results. In this case, the amount of data is not necessarily so important. For example, you may want to focus more on a specific part of the methodological process and investigate this process in depth, or show that you are aware of the general data collection methodology within a particular field. These methodologies are often different from one discipline to another and will be covered by one or more of the courses on methodology in your degree programme.

This is why it’s important right from the outset to consider the purpose of the study and the questions to be answered – that will determine the right amount of data you need to collect.

__________________________________________________________

Data collection in practice

For the practical process of data collection you can use manual as well as automatic methods.

Manual data collection involves gathering data by hand, in other words by conducting interviews, completing questionnaires or manually entering data into a computer. The advantage of manual data collection is that it can be relatively easy to implement. The disadvantage is that it can be time-consuming and prone to errors, for example, deficiencies or errors during data entry. However, some input errors can be prevented, for example by using automatic data entry validation features in Excel which will warn you, for example, if you enter text in a numbers field.

Automatic data collection involves using technology to collect data, for example, by means of sensors or software. The advantage of automatic data collection is that it can be more precise and efficient than manual data collection. One possible drawback is that it may be more complicated to implement – both technically and in terms of the ways qualitative data can be collected.

__________________________________________________________

‘Tidy data’

‘Tidy data’ is a concept that refers to a specific type of structured data that makes data easier to work with and analyse. Data can be structured without necessarily being tidy. Find out more about structured data in the lesson ‘What are data, and how are they different from information and knowledge?’

Tidy data have the following characteristics:

Each variable has its own column.
Each observation has its own row.
Each type of observational unit forms its own table.

In other words, the data must be structured in such a way that each column contains data for one particular variable, whereas each row contains data for one particular observation. In addition, the data must be divided into separate tables for each observational unit.

Qualitative data cannot always be arranged in a ‘tidy format’, as they are not always organised in rows and columns. They can still be organised appropriately to make them easier to work with, so the principles of tidy data can also be useful when it comes to organising qualitative data.

When the data are structured in this way, they become easier to find, process and analyse because there is no unnecessary information or redundant data. It is also easier to visualise the data and to present the results in a meaningful way.

Example of tidy data – plant growth

In the table below, each column is dedicated to one particular variable (date, vegetable, precipitation, sunshine, pH and growth), and each row contains data for one particular observation (growth of one vegetable on one particular day). There is no unnecessary information or redundant data, and the table is easy to process and visualise.

Date	Vegetable	Irrigation (ml)	Sunshine (hours)	pH of soil	Growth (cm)
2023-06-20	Tomatoes	500	6	7.0	10
2023-06-20	Cucumbers	500	6	7.0	8
2023-06-21	Tomatoes	400	8	6.5	12
2023-06-21	Cucumbers	400	8	6.5	10
2023-06-22	Tomatoes	300	4	7.0	14
2023-06-22	Cucumbers	300	4	7.0	12
2023-06-23	Tomatoes	450	7	6.8	16
2023-06-23	Cucumbers	350	5	7.2	14

__________________________________________________________

Documentation and metadata

Metadata are information about data, for example, what the data measure, how they were collected and when they were collected. Metadata are important because they enable us to understand and use data appropriately.

Documentation is a detailed description of how data have been collected and analysed. The documentation should contain information about methods, analyses and results. Documentation is important as it allows others to reproduce the study and understand how the results were obtained.

Metadata and documentation are therefore both important in order to ensure that the data are reliable and (re)usable. They help you to ensure that the data can be understood and used correctly, that the study can be reproduced and that the results can be verified. For this reason, metadata and documentation should always form part of an academic study.

Read more in the lesson ‘Describe your data so they can be understood – how to reuse data’ in the ‘Data management’ module.

__________________________________________________________

If you want to know more

Svend Brinkmann and Lene Tanggaard (Eds.): Kvalitative metoder: https://soeg.kb.dk/permalink/45KBDK_KGL/1pioq0f/alma99123305429905763

Mie Femø and Svend Skriver: Metodekogebogen: https://soeg.kb.dk/permalink/45KBDK_KGL/1pioq0f/alma99122686898605763

Lotte Rienecker and Peter Stray Jørgensen: Den gode opgave: https://soeg.kb.dk/permalink/45KBDK_KGL/1pioq0f/alma99124030748405763

Video: “Principles of tidy data”

__________________________________________________________