Group and the Nature of the Study

¶ … group and the nature of the study population. Both may present limitations to the researcher that will be reflective in the study design. In a quasi-experimental study, for instance, the study design lacks a fundamental component of a customary experimental design, namely randomization of the participants into study groups. Geographic limitations or the specificity of the participation qualifications may hinder the researcher from randomizing the subjects. An ex post facto study investigates possible cause-and-effect relationships by observing an existing condition and looking back into the past for valid causal factors. A certain study bias, however, is inherent in this type of study design because the variables are separated by time. Meanwhile, a correlational study compares two or more variables concurrently in detailed bivariate regression analyses. A common objective of this type of study is to determine the correlation between certain defining characteristics of the subjects and the effectiveness of some kind of treatment. Each study, regardless of type, must be examined for validity, reliability and generalizability of its results. Looking closely at the study's methodology including the sampling and data collection techniques, levels of variable measurement, types of statistics used and inferences that can be made based on the results can ascertain the quality and value of a quantitative research study.

In a quasi-experimental study, Hielkema, de Winter, de Meer and Reijneveld develop a study design for evaluating the effectiveness of a new family-centered method for the early identification of social-emotional and behavioral problems in young children (2011). The authors chose a convenience sample from geographic regions that served the purpose of their study. The experimental group was chosen from a region where the family-centered approach had already been implemented and the control from a region where it had not. Regions were comparable in socio-demographic conditions. The sampling was not randomized and was limited to those families visiting Preventative Child Healthcare (PCH) center in the Netherlands. The authors calculated an anticipated sample size of 1400 subjects for each study group. Data collection was done through questionnaires and standardized subject evaluation forms filled out by parents and qualified child health professionals (CHP) who visit the subjects' homes at 2, 3, 4, 6, 9, 14, and 18 months of age.

Psychosocial development will be evaluated by parents through the "Ages and Stages Questionnaire Social Emotional" (ASQ-SE) and the "Child Behavioral Checklist" (CBCL). The CHP will use a number of detailed evaluations to measure variables such as competence of the parent, role of the partner, social support, as well as the psychosocial healthcare of the child. "Chi-squared tests for categorical variables and t-tests for continuous variables will be used" (Hielkema, de Winter, de Meer & Reijneveld, 2011, p.6) to compare the baseline characteristics of the participants in the two groups. Logistical regression analyses will be conducted to assess the proportion of and risk for psychosocial problems identified by the CHP in both study groups. The study does not present results but merely a design for a study. However, inferences can be drawn about the expected feasibility of conducting such a study. The primary weakness lies in the sampling technique, which is conducted without randomization. The parents, whose evaluations play a large role in the study, know which group they are in and therefore are biased towards presenting positive outcomes. This further reduces the reliability of the data. The objectives are valid and statistical analyses detailed. However, the study design cannot be generalized to other studies and the reproducibility of the study appears difficult.

In an example of an ex post facto study, Watson, Clark and Stasik examined positive and negative type affects as predictors for emotional disorders (2011). Two study samples of outpatients were recruited from several clinics and mental health centers in Iowa and subjects were individually approached. Patients were randomly assigned to one of two study groups. The first completed the Expanded Form of the Positive and Negative Affect Schedule (PANAS-X) to rate specifically their trait affectivity, while those in the second group described their emotional experiences over the previous week. Data collection was done exclusively through the PANAS-X evaluation measure. The questionnaire asks the patient to evaluate on a scale of 1 (very slightly) to 5 (extremely) to what extent he/she felt a certain emotion. (Watson, Clark & Stasik, 2011, p.433). The PANAS-X is composed of scales that measure negative emotions including fear and irritability and general positive emotions including activeness, alertness and interest. Additionally, it measures sadness, guilt, hostility, joviality, and self-assurance. Each subject was diagnosed in a structured clinical interview with current DMS-IV diagnoses.

A number of bivariate analyses were conducted to determine any correlations between the DSM-IV diagnoses and responses from the PANAS-X. An inference that can be made from the results of this study is that different emotions interact in a complex way to construct a certain DMS-IV emotional disorder diagnoses. While the study design is very reliable and the authors ensure the inter-reliability of the professionals who make the diagnoses, the ex post facto nature of the study design nonetheless leaves doubt to the validity of the outcomes. While sadness, guilt and joviality showed clear specificity to depression, these results cannot be generalized due to the specific nature of each depression and the variable influence of these emotions on it. Further, the study relied on a single, self-evaluation measure, which reduces the reliability of the results.