independent and dependant variables in any given population. One type of research design is that of the quantitative design. In quantitative research, the goal is to determine specific relationships, and as such, all research is either considered descriptive, where subjects are measured once, or experimental, where subjects are measured before and after a specific treatment or event (Hoover, Donovan, 2004). In descriptive studies, only observation is used, whereas in experimental designs, actual manipulation of variables occurs.
This paper will focus on two types of quantitative design, those of the cross-sectional design and the longitudinal design. Both types of design are considered to be descriptive in nature, in that no manipulation of variables is done (Woolf, 1998). However, each type has its own advantages and disadvantages, each of which will be discussed.
A cross sectional design study is, as noted, a type of quantitative research design known as a descriptive design. In cross sectional studies, data is collected a single time from the subject pool on the relevant variable. All data is collected within a short period of time, generally through the use of surveys (Saint-Germain, 2004). While cross-sectional designs are extremely useful in determining variables across populations, there cannot be an analysis of cause and effect, since the variables are not manipulated, nor is the data collected more than once. Thus, it is impossible to infer causality (Woolf, 1998).
Cross-sectional designs have many advantages over other forms of research designs. First, and perhaps most importantly, cross sectional designs are more cost effective than other forms, such as the cohort study, which examines data over time (Larkin, 1985). Since data collection occurs only once, the cost of continued data collection and follow up costs are avoided.
Further, since data is generally collected with surveys, data can be collected from a large number of subjects simultaneously, thereby increasing the validity of any conclusions from the study. This effect is furthered by the ability to study a large variety of subjects simultaneously, enabling researchers to examine numerous variables across a wide variety of subjects (Saint-Germain, 2004).
Still another advantage of the cross-sectional study is that this method can collect data on attitudes and behaviors, which other observational or descriptive research methods cannot. A correctly modeled cross sectional research design can answer questions on exploratory subjects, without the expense of more in-depth forms of research (Hopkins, 2001). This allows the researchers to generate hypotheses for future research in a cost effective way (Saint-Germain, 2004).
However, there are disadvantages to the cross-sectional design. First, as mentioned, this method cannot be used to establish cause and effect, since data is collected only once during the study. With only a single data point, any relationship noted could have been caused from any number of untested variables. Furthermore, change cannot be measured with this method. In order to measure change, more than one data point would be necessary (Saint-Germain, 2004).
Another disadvantage is the cost increases involved in adding subjects or locations to the study. For each subject added, and each location tested, higher costs are incurred. While it is still true that cross sectional methods are cost effective, any alterations needed in study locations involves increases in funding which can be a problem (Larkin, 1985). For example, if the original study design was based on a collection of surveys from a specific college from 200 participants, and another location is later added for a more variable subject pool, the costs of the study increase dramatically.
There are also disadvantages relating to the variables and conclusions generated by cross sectional research designs. First, since this method is descriptive only, there is no control over the independent variable and further, since no causality or change can be measured, there is no possibility of refuting alternative hypotheses. Finally, since the data is collected only a single time, the study is static, in that it is bound in time to whenever the data is collected (Saint-Germain, 2004).
Longitudinal designs, on the other hand, involve the collection of data over a period of time. Rather than collecting a single data set, the longitudinal design allows researchers to study an independent variable over time, in order to measure changes in that variable within the subject population (Gliner, 2000). Measurements are taken for each variable to be studied on more than one occasion in order to measure the changes in those variables.
It is important to note that there are two forms of longitudinal design, that of the time series, and the panel. With a time series design, data is collected on a single variable at regular time periods, such as each week, and are combined to form a collective measurement. For example, unemployment rates are considered a time series longitudinal design. The second form, the panel design, involves the collection of data from a specific group of subjects over time, revealing a more individualized pattern of change (Saint-Germain, 2004).
As with the cross sectional design, there are advantages to using the longitudinal method. In particular, this design can show how relationships emerge over the course of time, which a cross sectional method cannot. Furthermore, with this method the researcher is able to show the time order of variables, such as whether one variable or another occurred first. This can help to establish hypotheses of cause and effect (Saint-Germain, 2004).
Other advantages of this method include the ability to report information on an individualized level. Many other types of research methods only allow for generalization, but with longitudinal designs, it is possible to predict, at least on a short-term basis, trends for small groups of individuals (Woolf, 1998). Additionally, the data results tend to be easy to present in graph formats, and easy for the average individual to comprehend (Saint-Germain, 2004).
However, there are disadvantages. Any fluctuation in data would require a qualitative research design to explain, and since data in this method is collected over a longer period of time, there is a higher risk of data fluctuations. Additionally, any forecast of trends must first make assumptions that the data will not change over time, which is often an untrue assumption. Also, with the long time period involved, subject participation is likely to reduce over the course of the research, thus possibly rendering the data inaccurate (Larkin, 1985). Since it is difficult to retain the initial subjects, additional cost may be incurred while locating additional subjects. Finally, any study that measures data more than once has a higher probability of bias, in that repeated measurements tend to influence the behavior of the subject (Saint-Germain, 2004).
It is perhaps easiest to see the fundamental differences between these two types of methods by examining possible outcomes of a study conducted on the same topic by each. In this example, the researchers will be studying whether or not smoking causes harmful effects on the lungs. Harmful affects, according to this hypothetical study, may include chronic bronchitis, frequent lung infections, lung cancer, or emphysema.
With a cross sectional design method, a survey would likely be conducted on a given population during a single data collection, since this type of method is generally used (Community Foundation, 2003). The researchers would choose a population, such as patients in a long-term care facility such as a nursing home, and survey a given number of subjects. In this case, the researchers will hypothetically survey 200 patients in a nursing home. The survey would consist of questions regarding the subject's overall health condition, lung conditions, race, gender, age, and smoking habits.
From those results, researchers could develop a hypothesis of whether or not smoking causes long-term lung conditions in specific populations. While cause and effect cannot be established, the results could suggest a relationship between length of smoking, lung disease, and a specific race. This would enable researchers to develop…