3 Types of study designs
You have learnt how to ask a RQ. In this chapter, you will study how to design a method for collecting the data needed to answer the RQ. You will learn to:
- design scientifically sound studies to answer simple quantitative research questions.
- describe the various types of quantitative research studies.
- compare and contrast experimental and observational studies.
- describe and identify the directionality in observational studies.
- describe and identify true experimental and quasi-experimental studies.
- explain external and internal validity.
3.1 Three types of study designs
The RQ implies what data must be collected from the individuals in the study (the response and explanatory variables)... but how are the data obtained? After all, data are important: they are the means by which the RQ is answered.
Different types of studies are used for different types of RQs:
- Descriptive studies (Sect. 3.2) to answer descriptive RQs;
- Observational studies (Sect. 3.3) to answer relational RQs; or
- Experimental studies (Sect. 3.4) to answer interventional RQs.
Observational and experimental studies are sometimes called analytical studies.
Suppose we wish to compare the effects of echinacea on the symptoms of the common cold (based on Barrett et al. (2010)). What decisions would need to be made to design a study to collect the data?
3.2 Descriptive studies
Descriptive studies answer descriptive RQs (Fig. 3.1), which specify a population and outcome.
Definition 3.1 (Descriptive study) Descriptive studies answer descriptive research questions, and do not study relationships between variables.
FIGURE 3.1: A descriptive study, used to answer a descriptive RQ
Example 3.1 (Descriptive study) Consider this RQ:
For obese men over 60, what is the average increase in heart rate after walking 400 metres?
The outcome is the average increase in heart rate. The response variable is the increase in heart rate for each individual man, found by measuring each man's heart rate before and after the walk (measured within-individuals).
The increase in heart rate for each man would be computed as the after heart rate minus the before heart rate. Some differences might be positive numbers (heart rate increased), and some may be negative numbers (heart rate decreased).
No between-individuals comparison is being made: every man in the study is treated in the same way. This is a descriptive RQ, which can be answered by a descriptive study.
We do not explicitly discuss descriptive studies further, as they can be considered as a special case of observational studies.
3.3 Observational studies
Observational studies (Fig. 3.2) answer relational RQs to study relationships. They are commonly-used, and sometimes are the only study design possible.
Definition 3.2 (Observational study) Observational studies answer relational research questions.
FIGURE 3.2: An observational study, used to answer a relational RQ
Definition 3.3 (Condition) Conditions: The conditions are the values of the comparison or connection that those in the observational study experience, but are not imposed by the researchers.
Example 3.2 (Observational study) Consider again this RQ (Barrett et al. 2010):
Among Australian teens with a common cold, is the average duration of cold symptoms shorter for teens taking a daily dose of echinacea compared to teens taking no medication?
This would be a relational RQ if the researchers do not impose the taking of echinacea (that is, the individuals make this decision themselves). The two conditions are 'taking echinacea', and 'not taking echinacea' (Fig. 3.3).
FIGURE 3.3: Observational studies. The dashed lines indicate steps that are not under the control of the researchers
3.4 Experimental studies
Experimental studies (Fig. 3.4), or experiments, are commonly-used to study relationships. Well-designed experimental studies can establish a cause-and-effect relationship between the response and explanatory variables. However, using experimental studies is not always possible. Experiments have an intervention, and so experimental studies answer interventional RQs.
Definition 3.4 (Experiment) Experimental studies (or experiments) answer interventional research questions.
Definition 3.5 (Treatments) The treatments are the values of the comparison or connection that the researchers impose upon the individuals in the experimental study.
In an experimental study, the unit of analysis (Def. 2.7) is the smallest collection of units of observations that can be randomly allocated to separate treatments.
FIGURE 3.4: An experimental study, used to answer interventional RQs
Two types of experimental studies (Table 3.1) are true experiments and quasi-experiments.
| Study type | Do researchers allocate individuals to receive the comparison/connection? | Do researchers allocate individuals to treatments | Reference |
|---|---|---|---|
| Observational | No | No | Sect. 3.3 |
| True experiment | Yes | Yes | Sect. 3.4.1 |
| Quasi-experiment | No | Yes | Sect. 3.4.2 |
3.4.1 True experimental studies
True experiments are commonly used, but are not always possible. An example of a true experiment is a randomised controlled trial, often used in drug trials.
Definition 3.6 (True experiment) In a true experiment, the researchers:
- allocate treatments to groups of individuals (i.e., determine the values of the explanatory for the individuals), and
- determine who or what individuals are in those groups.
While these may not actually happen in these explicit steps, they can happen conceptually.
Example 3.3 (True experiment) The echinacea study (Sect. 2.5) could be designed as a true experiment. The researchers would allocate individuals to one of two groups, and then decide which group took echinacea and which group did not (Fig. 3.5).
These steps may happen implicitly: Researchers may allocate each person at random to one of the two groups (echinacea; no echinacea). This is still a true experiment, since the researchers may decide to switch which group receives echinacea; ultimately, the decision is still made by the researchers.
FIGURE 3.5: True experimental studies
A researcher wants to examine the effect of an alcohol awareness program (M. MacDonald 2008) on the amount of alcohol consumed in a university Orientation Week. She runs the program at University A only, then compares the average amount of alcohol consumed per person at two universities (A and B).
Is this study observational or true experimental?
Neither observational, or a true experiment.
The researcher did not determine the groups: the students (not the researcher) would have chosen University A or University B for many reasons. The researcher did decide how to allocate the program to University A or University B.
3.4.2 Quasi-experimental studies
Quasi-experiments are similar to true experiments, but treatments are allocated to groups that already exist (i.e., may be naturally occurring).
Definition 3.7 (Quasi-experiment) In a quasi-experiment, the researchers:
- allocate treatments to groups of individuals (i.e., allocate the values of the explanatory variable to the individuals), but
- do not determine who or what individuals are in those groups.
Example 3.4 (Quasi-experiments) The echinacea study (Sect. 2.5) could be designed as a quasi-experiment. The researchers could find (not create) two existing groups of people (say, from Suburbs A and B), then decide to allocate people in Suburb A to take echinacea, and people in Suburb B to not take echinacea (Fig. 3.6).
FIGURE 3.6: Quasi-experimental studies. The dashed lines indicate steps that are not under the control of the researchers
3.5 Comparing study types
Different RQs require different study designs (Table 3.2). In experimental studies, researchers create differences in the explanatory variable through allocation, and note the effect this has on the response variable. In observational studies, researchers observe differences in the explanatory variable, and observe the values in the response variable.
Importantly, only well-designed true experiments can show cause-and-effect. Nonetheless, well-designed observational and quasi-experimental studies can provide evidence to support cause-and-effect conclusions, especially when supported by other evidence. Although only experimental studies can show cause-and-effect, experimental studies are often not possible for ethical, financial, practical or logistical reasons.
The advantages and disadvantages of each study type are discussed later (Sect. 9.2), after other issues are examined in the following chapters.
| RQ type | P | O | C | I | Study type |
|---|---|---|---|---|---|
| Descriptive | Yes | Yes | Descriptive | ||
| Relational | Yes | Yes | Yes | Observational | |
| Interventional | Yes | Yes | Yes | Yes | Experimental |
Example 3.5 (Cause and effect) Many studies report that the bacteria in the gut of people on the autism spectrum is different than the bacteria in the gut of people not on the autism spectrum (Kang et al. (2019), Ho et al. (2020)), and suggest the bacteria may contribute whether a person is autistic. These studies were observational, so the suggestion of a cause-and-effect relationship may be inaccurate.
Other studies (Yap et al. 2021) suggest that people on the autism spectrum are more likely to be "picky eaters", which contributes to the differences in gut bacteria.
The animation below compares observational, quasi-experimental and true experimental designs.
FIGURE 3.7: The three main study designs
3.6 Directionality
Analytical research studies (observational; experimental) can be classified by their directionality (Table 3.3):
- Forward direction: The values of the explanatory variable are obtained, and the study determines what values of the response variable occur in the future. All experimental studies have a forward direction.
- Backward direction: The values of the response variable are obtained, then the study determines what values of the explanatory variable occurred in the past.
- No direction: The values of the response and explanatory variables are obtained at the same time.
Directionality is important for understanding cause-and-effect relationships. If the comparison/connection occurs before the outcome is observed, a cause-and-effect relationship may be possible. That is, studies with a forward direction are more likely to provide evidence of causality.
| Type | Explanatory variable | Response variable |
|---|---|---|
| Forward direction | When study begins | Determine in the future |
| Backward direction | Determined from the past | When study begins |
| No direction | When study begins | When study begins |
Example 3.6 (Backwards study) A study (Pamphlett 2012) examined patients with and without sporadic motor neurone disease (SMND), and asked about past exposure to metals.
The response variable (whether or not the respondent had SMND) is assessed when the study begins, and whether or not they had exposure to metals (explanatory variable) is determined from the past. This observational study has a backward direction.
Example 3.7 (Forward study) Chih et al. (2018) studied dogs and cats who had been recommended to receive intermittent nasogastric tube (NGT) aspiration for up to 36 hours. Some pet owners did not give permission for NGT, while some did give permission; thus, whether the animal received NGT was not determined by the researchers (hence this is an observational study). The researchers then observed whether the animals developed hypochloremic metabolic alkalosis (HCMA) in the next 36 hours.
Since the explanatory variable (whether NGT was used or not) was recorded at the start of the study, and the response variable (whether HCMA was observed or not) was determined within the following 36 hours, this study has a forward direction.
Example 3.8 (Non-directional study) A study (J. Russell et al. 2014) asked older Australian their opinions of their own food security, and recorded their living arrangements. Individuals' responses to both both the response variable and explanatory variable were gathered when the study began. This observational study is non-directional.
In South Australia in 1988--1989, 25 cases of legionella infections (an unusually high number) were investigated (O’Connor et al. 2007). All 25 cases were gardeners.
Researchers compared 25 people with legionella infections with 75 similar people without the infection. The use of potting mix in the previous four weeks was associated with an increase in the risk of contracting illness of about 4.7 times.
What direction does this observational study have?
Backward directionality: people were identified with an infection, and then the researchers looked back at past activities.
Research studies are sometimes described as 'prospective' or 'retrospective', but these terms can be misleading (Ranganathan and Aggarwal 2018) and their use not recommended (Vandenbroucke et al. 2014).
Experimental studies always have a forward direction. Observational studies may have any directionality, and are sometimes given different names accordingly.
3.6.1 Forward-directional studies
All experimental studies have a forward direction, and include randomised controlled trials (RCTs) and clinical trials.
Observational studies with a forward direction are often called cohort studies. Both experimental studies and cohort studies can be expensive and tricky: tracking a gfroup of individuals (a cohort) into the future is not always easy, and individuals may be lost to follow-up (drop-outs). Forward-directional observational studies:
- may add support to cause-and-effect conclusions, since the comparison/connection occurs before the outcome (only well-designed experimental studies can establish cause-and-effect).
- can examine many different outcomes in one study, since the outcome(s) occur in the future.
- can be problematic for rare outcomes, as the outcome of interest may not appear (or may appear rarely) in the future.
3.6.2 Backward-directional studies
Observational studies with a backward direction are often called case-control studies. Researchers find individuals with specific values of the response variable (the cases and the controls), and determine values of the explanatory variable from the past. Case-control studies:
- only allow one outcome to be studied, since individuals are chosen to be in the study based on the value of the response variable of interest.
- are useful for rare outcomes: The researchers can purposely select large numbers with the rare outcome of interest.
- do not effectively eliminate other explanations for the relationship between the response and explanatory variables (called confounding; Def. 6.4).
- may suffer from sampling bias, as researchers try to locate individuals with a rare outcome.
- may suffer from recall bias when the individuals are people: accurately recalling the past can be unreliable.
3.6.3 Non-directional studies
Non-directional observational studies are called cross-sectional studies. Cross-sectional studies are:
- good for findings associations between variables (which may or may not be causation).
- generally quicker and cheaper than other types of studies.
- not useful for studying rare outcomes.
- do not effectively eliminate other explanations for the relationship between the response and explanatory variables (called confounding; Def. 6.4).
3.7 Internal validity
Ideally, all studies should be designed to be internally valid (Chaps. 7 and 8).
Internally validity refers to how reasonable and logical it is to conclude that changes in the value of the response variable can be attributed to changes in the value of the explanatory variable; that is, it refers to the strength of the inferences made from those studied. Internally valid studies are generally accurate and repeatable.
Studies with high internal validity show that changes in the response variable can confidently be related to changes in the explanatory variable in the group that was studied; the possibility of other explanations has been minimised.
In contrast, studies with low internal validity leave open other possibilities, apart from changes in value of the explanatory variable, to explain changes in the value of the response variable. Experimental studies usually have higher internal validity than observational studies.
Definition 3.8 (Internal validity) Internally validity refers to how reasonable and logical it is to conclude that changes in the value of the response variable can be attributed to changes in the values of the explanatory variable; that is, the strength of the inferences made from those studied.
A study with high internal validity shows that the changes in the response variable can be attributed to changes in the explanatory variables; other explanations have been ruled out.
Example 3.9 (Low internal validity) In a review of studies that used double-fortified salt to manage iodine and iron deficiencies, one conclusion was:
Internal validity of the efficacy trials was generally weak [...] because of issues around selection bias, unaccounted confounders, and participant withdrawals.
--- L. M. Larson et al. (2021), p. 26S
One of many potential threats to internal validity is that the groups being compared are initially different; for example, if the group receiving echinacea is younger (on average) than the group receiving no medication. This is a form of confounding (Def. 6.4).
To check this, the baseline characteristics of the individuals in the groups can be compared: the groups being compared should be as similar as possible, so that any differences in the outcome cannot be attributed to pre-existing difference in the groups.
Example 3.10 (Baseline characteristics) In a study of treating depression in adults (Danielsson et al. 2014), three treatments were compared: exercise, basic body awareness therapy, or advice.
If any differences among the treatments were found, the researchers need to be confident that the differences were due to the treatment. For this reason, the three groups were compared to ensure the groups were similar in terms of average ages, percentage of women, taking of anti-depressants, and many other aspects.
An internally valid study requires studies to be carefully designed, discussed in Chaps. 7 and 8. In general, well-designed experimental studies are more likely to be internally valid than observational studies.
3.8 External validity
A study is externally valid if the results of the study are likely to generalise to the rest of the population, beyond just those studied in the sample. To be externally valid, a study first needs to be internally valid, since the results must at least be sound for the group under study before being extended to other members of the population.
Using a random sample helps ensure external validity. In addition, the use of inclusion and exclusion criteria (Sect. 2.3.1) helps clarify to whom or what the results may apply outside of the sample being studied.
Definition 3.9 (External validity) External validity refers to the ability to generalise the results to the rest of the population, beyond just those in the sample studied. For a study to be truly externally valid, the sample must be a random sample (Chap. 5) from the population.
External validity does not mean that the results apply more widely than the intended population.
Example 3.11 (External validity) Suppose the population in a study is Californian university students. The sample comprises the Californian university students actually studied. The study is externally valid if the sample is a random sample from the population of all Californian university students.
The results will not necessarily apply to university students outside of Californian (though they may), or all Californian residents. However, this is irrelevant for external validity. External validity concerns how the sample represents the intended population in the RQ, which is Californian university students. The study is not concerned with all Californian residents, or with non-Californian university students.
3.9 The importance of design
Choosing the type of study is only one part of research design. Planning the data collection process, and actually collecting the data, is still required. Sometimes, data may be already available (called secondary data), or may need collecting (called primary data).
Either way, knowing how the data are obtained is important. The design phase is concerned with planning the best approach to obtaining the data, to ensure the study is internally and externally valid, as far as possible.
Internal validity considerations include:
- What else might influence the values of the response variable, apart from the explanatory variable? (Chap. 6)
- How can the study be designed effectively to maximise internal validity? (Chaps. 7 and 8)
External validity considerations include:
- Sampling: Since the whole population cannot be studied, who or what do we study in the population (Chap. 5)? And how many do we need to study? (We need to learn more before we can answer this critical question in Chap. 26.)
The details of how the data will be collected (Chap. 10) and ethical issues (Chap. 4) must also be considered. Furthermore, the limitations of the study must be communicated (Chap. 9).
The following short (humourous) video demonstrates the importance of understanding the design!
3.10 Summary
Three types of research studies are: Descriptive studies (for descriptive RQs), observational studies (for relational RQs), and experimental (for interventional RQs).
Observational studies can usually be classified as having a forward direction (cohort studies), backward direction (case-control studies), or no direction (cross sectional studies). Experimental studies always have a forward direction, and can be classified as true experiments or quasi-experiments. Cause-and-effect conclusions can only be made from well-designed true experiments.
Ideally studies should be designed to be internally and externally valid. In general, experimental studies have better internal validity than observational studies.
FIGURE 3.8: Chapter 3 summary
The following short videos may help explain some of these concepts:
3.11 Quick review questions
- A study (Fraboni et al. 2018) examined the 'red-light running behaviour of cyclists in Italy'.
This study is most likely to be:
- A study of a sample whose results apply to the wider population of interest would be called:
- In a quasi-experiment, the researchers allocate treatments to groups that they have not organised. True or false?
- What is the difference between an true experiment and a quasi-experiment?
-
A research study compared the use of two different education programs to reduce the percentage of patients experiencing ventilator-associated pneumonia (VAP). Paramedics from two cities were chosen to participate. Paramedics in City A were chosen to receive Program 1, and paramedics in the other city to receive Program 2. What type of study is this?
-
Which of the following are true?
- True experiments have a higher internal validity than observational studies.
- Internal validity refers to the strength of the inferences made from the study.
- External validity refers to the ability to generalise the results to other groups apart from those studied.
- Inclusion and exclusion criteria can be used to clarify the internal validity.
- Observational studies have a higher external validity than experimental studies.
- True experiments have a higher internal validity than observational studies.
3.12 Exercises
Selected answers are available in Sect. D.3.
Exercise 3.1 In a study on the shear strength of recycled concrete beams (Gonzalez-Fonteboa and Martinez-Abella 2007), beams were divided into three groups. Different loads were then applied to each group, and the shear strength needed to fracture the beams was measured. Is this a quasi-experiment or a true experiment? Explain.
Exercise 3.2 A nursing study aimed to compare "the effectiveness of alternating pressure air mattresses vs. overlays, to prevent pressure ulcers" (Manzano et al. (2013) , p. 2099). Patients were provided with either alternating pressure air overlays (in 2001) or alternating pressure air mattresses (in 2006). The number of pressure ulcers were recorded.
This study is experimental, because the researchers provided the mattresses. Is this a true experiment or quasi-experiment? Explain.
Exercise 3.3 Consider this initial RQ (based on Friedmann and Thomas (1985)), that clearly requires a lot of refining: "Are people with pets healthier?" To answer this RQ:
- Describe useful and practical definitions for P, O and C.
- Describe an experimental study to answer the RQ.
- Describe an observational study to answer the RQ.
Exercise 3.4 Consider this journal article extract:
We randomly assigned 811 overweight adults to one of four diets [...] The diets consisted of similar foods and met guidelines for cardiovascular health [...] The primary outcome was the change in body weight after 2 years in [...] comparisons of low fat versus high fat and average protein versus high protein and in the comparison of highest and lowest carbohydrate content.
--- Sacks et al. (2009), p. 859
- Define POCI.
- Is this study observational or experimental? Why?
- Is this study a quasi-experiment or a true experiment? Why?
- What are the units of analysis?
- What are the units of observation?
- What is the response variable?
- What is the explanatory variable?