Designing Quantitative Studies Dr. Belal Hijji, RN, PhD December 2 & 9, 2010
Learning Outcomes • Identify aspects of quantitative design • Have an overview of the quantitative research designs Read Polit & Beck chapter 8
Aspects of Quantitative Research Design • When doing a research study, important decisions need to be made about the study’s design. These decisions will affect the overall believability of the findings. • Intervention: Nurse researchers may want to test the effects of a specific intervention (an innovative programme to promote breast self-examination). • Comparisons: In most studies, researchers develop comparisons to provide a context for interpreting results. The most common types of comparisons are: • Comparisons between two groups or more: For example, if we want to study the emotional consequences of having an abortion. To do this, the researcher may compare the emotional status of women who had abortion with that of women with an unintended pregnancy who delivered a baby. Without this comparison, it would be difficult to know whether the women’s emotional status was of concern without comparing it with that of others
Comparison of one group’s status at two or more points in time: For example, we might want to assess patients’ levels of stress before and after introducing a new procedure to reduce preoperative stress. • Comparison of one group’s status under different circumstances. An example is to compare patients’ heart rates during two different types of exercise. • Controls for extraneous variables: Variables that confound the relationship between the independent and dependent variables and that need to be controlled in the research design. For example, In a drug trial, patients might take other medicines that could affect the medical condition under study. Such patients must either refrain from taking these medicines or be excluded from the study.
Timing of data collection: In most studies, data are collected from participants at a single point in time. For example, nurses may be asked once to fill in a questionnaire about their knowledge and practice of pressure ulcer care. Some designs call for multiple contacts with participants, usually to determine how things changed over time. The researcher, therefore, must decide on the number of data collection points to address the research question properly. For example, we may want to test the effect of an educational intervention on nurses’ knowledge and practice of blood transfusion at 1 week and 1 month after exposure.
Research sites and settings: Sites are the overall locations for the research, and settings are the more specific places where data collection will occur. • Communication with participants: In designing a study, the researcher must decide how much information to provide to study participants. For example, if we want to conduct an observational study on nurses’ hand decontamination practices, full disclosure to participants before obtaining their consent is ethically correct, but this can undermine the value of the research when they are exposed to the elements of the observation schedule.
Overview of Research Design Types • Between-subjects and within-subject designs: As far as we know, quantitative studies involve making comparisons between separate groups of people. For example, the hypothesis that tamoxifen reduces the rate of breast cancer in high-risk women could be tested by comparing women who received the drug and those who did not. Both groups of women are different. Sometimes, we want to make comparisons for the same study participants. For example, a researcher may want to study patients’ heart rate before and after a nursing intervention. This example calls for a within-subjects design.
The time dimension: While most studies involve data collection at a point in time, Sometimes, it is appropriate to collect data at multiple points. First, when studying time-related processes such as healing, learning, and physical growth. Second, when determining time sequences of phenomena. If it is hypothesise that infertility results in depression, then it would be important to determine that depression did not precede infertility. Third, developing comparisons over time to determine whether changes have occurred, such as when a study is concerned with documenting trends in the smoking behaviour over 10-year period. Studies are often categorised in terms of how they deal with time. The major distinction is between cross-sectional and longitudinal designs. Both are described next.
Cross-Sectional Designs • In these designs, data collection occurs at a single point in time when it is appropriate for a study to describe the status of phenomena or for describing relationships between phenomena. For example, we may want to determine whether psychological symptoms in menopausal women are correlated contemporaneously [at the same time] with physiologic symptoms.
Longitudinal Designs • In these designs, data collection occurs at multiple points in time over an extended periods. There are several types of longitudinal designs. • Trend studies are investigations in which different samples from a population are studied over time with respect to some phenomenon. Trend studies permit researchers to examine patterns and rates of change over time and to predict future developments. For example, we may study trends in alcohol consumption in a country over 15-year period to see whether heavy drinking had fallen or remained unchanged. • Cohort studies are a particular type of a trend study in which specific subpopulations are examined over time. The samples are usually drawn from specific age-related subgroups, for example men born between 1960 and 1965 may be studied over time with respect to health care utilization.
In panel studies, the same people are used to supply data at two or more points in time to determine which individuals who changed and those who did not and then examine the characteristics of both groups. For example, we may explore over time the antecedent characteristics of smokers who were later able to quit.
Experimental designs • Basic experimental designs • Simple posttest-only design: An example is a study that tests the effect of gentle massage on the pain level of nursing home residents. Another example is to study the effect of an educational intervention on urinary incontinence on the subsequent help-seeking behaviour of older adults. • Pretest – posttest design: Suppose we want to find out whether convective airflow blankest are more effective than conductive water-flow blanket in cooling critically ill patients with fever. We measure the dependent variable twice; before and after the intervention. Based on the results, we can say whether one blanket type is more effective than the other in reducing fever.
Solomon Four-Group Design: When a pretest- posttest design is used, the posttest measure of the dependent variable may be affected by the treatment as well as pretesting. To avoid this problem, Solomon four-group is called for, which involves to experimental and two control groups. 13
Factorial Design: In this design, the researcher concurrently manipulates two or more variables. For example, if we are interested in comparing two therapeutic strategies for premature infants: tactile [tangible] stimulation versus auditory stimulation. Additionally, we want to know if the daily amount of stimulation (15, 30, or 45 minutes) affects infants’ progress. The dependent variables are measures of infant development (weight gain and cardiac responsiveness) 14