12.3: Longitudinal and Similar Designs

Learning Objectives

1. Explain why longitudinal designs are not quasi-experimental designs.
2. Provide an example of longitudinal designs compared to cross-sectional designs.

Studies of Time

When researchers wish to study change over time (for example, when developmental psychologists wish to study aging) but they do not (or cannot) manipulate an IVs), they usually take one of three non-experimental approaches: cross-sectional, longitudinal, or cross-sequential.

Longitudinal Design

To understand the other types of designs to understand changes though time, it is easier to start with longitudinal designs. Longitudinal designs follow one group of people over time, taking measurements regularly. These type of designs are often use to follow children through development in psychology, or follow the progression of diseases in medicine. As professor in a Psychology department, Dr. MO is interested in the progression of statistical literacy (the ability to understand and interpret numbers and statistics) for Psychology majors from their first course to their graduation. She hopes that taking Psychology courses will improve statistical literacy as students progress through the Psychology curriculum. To test this, Dr. MO could send a statistical literacy assessment to 100 new students who identified their major as Psychology a week before one fall semester, then continue to send this same assessment to the same participants until they graduate (or for 10 years, whichever happens first).

One benefit of this type of design is that the development of changes in thoughts, behaviors, and feelings can discovered. There are a couple of disadvantages, however. One disadvantage is that there is usually no random assignment of IVs or interventions in these types of designs. Instead, once the sample is chosen, the researcher measures the variables of interest but does not manipulate anything. While there may end up being comparisons between groups, these would be natural groups or quasi-experimental, and cannot be guaranteed. Another disadvantage is that longitudinal studies are, by definition, time consuming. It is tough to argue to a granting agency to fund a 20-year research study that may not begin providing any useful data for up to 10 years. In addition to the time and cost investment of researchers, longitudinal designs also require a large investment on the part of the participants. As participants stop participating in the study, researchers have to consider if the current sample represents the desired population.

Cross-Sectional Design

To try to avoid these disadvantages, researchers sometimes choose instead to conduct cross-sectional studies. Cross-sectional studies are a type of research design that involves measurement of the same variable for individuals who are at varying points of progress or development. This definition doesn't really make sense until you have an example, so let's go back to those Psychology majors' statistical literacy. Rather than waiting 10 years for all of the data, Dr. MO could instead send the statistical literacy assessment to 100 new students who are Psychology majors, 100 Psychology majors who are sophomores, 100 Psychology majors who are juniors, 100 Psychology majors who are seniors, and 100 Psychology majors who graduated the last semester. She could do this a week before this fall semester, and would receive the data from those who choose to participate by the start of the fall semester, rather than waiting up to 10 years! Just like with longitudinal designs, this is non-experimental because there is no manipulation of an independent variable and there's no initial equivalence or ongoing equivalence.

A benefit of this design is that the data from the participants is received quickly. A disadvantage of using this type of design is that differences between the groups other than progression or development may account for differences in the dependent variable. For instance, in our study of statistical literacy, we might find that those who graduate do have higher scores on the assessment than those sophomores, which could suggest that the curriculum improves statistical literacy. However, because we tested everyone at the same time, it could be that the students who had already graduated had taken their first behavioral statistics course from a specific instructor that helped them understand, but that instructor was assigned to teach research methods so the sophomore students took behavioral statistics from another instructor. Was it the curriculum that improved statistical literacy, or was it something about what the graduates had experienced but none of the other cohorts would experience? Or, because cross-sectional designs are nonexperimental, could there be other variables related to the groups that affect the outcome that's not time, progress, or development? For studies looking at (physical, emotional, or social) maturation through time, differences between the groups may reflect the generation that people come from (another example of a cohort effect) rather than a direct effect of age.

Cross-Sequential Design

A third approach, known as cross-sequential studies, combines elements of both cross-sectional and longitudinal studies. Rather than measuring differences between people in different age groups or following the same people over a long period of time, researchers adopting this approach choose a smaller period of time during which they follow people in different age groups. For example, they might measure changes over a ten year period among participants who at the start of the study fall into the following age groups: 20 years old, 30 years old, 40 years old, 50 years old, and 60 years old. This design is advantageous because the researcher reaps the immediate benefits of being able to compare the age groups after the first assessment. Further, by following the different age groups over time they can subsequently determine whether the original differences they found across the age groups are due to true age effects or cohort effects.

Quasi-Experimental or Nonexperimental?

As noted throughout this section, these types of designs (longitudinal, cross-sectional, and cross-sequential) are typically nonexperimental. There is usually no manipulation of an IV because time or development is considered the "intervention". This means that there's no initial equivalence nor ongoing equivalence between the groups; in fact, the different ongoing experiences through out time is often what's being studied! This places these types of designs squarely into the "nonexperimental design" approach.

However, it would be possible to start with (non-equivalent) groups, which could make specific versions of these designs a quasi-experimental design. For example, In our example of statistical literacy in Psychology majors, we could compare the statistical literacy of students throughout their 4-10 years in college but compare students who took a behavioral statistics course with students who took a statistics course from the Math department. This would be more like an interrupted time series design now, which is quasi-experimental.

In the end, you as the researcher and you as the reader of research must decide who comfortable you are in concluding that nothing other than the IV could have affected the DV. Even if a design "looks" like nonexperimental, it might be quasi-experimental. And with comparing pretests with posttests, quasi-experimental designs may have more control than some poorly-ran experimental designs.