Experimental Designs

Experimentation is one of the common methods used in quantitative research. Premised on the positivist philosophy, an experiment is essentially conducted to investigate causal relationships between variables (Bryman, 2008). Indeed, this is one of the major strengths of experimental research compared to other types of studies -- it not only describes association between variables, but also explains causation between variables (Kothari, 2004). This essay describes the various components of an experimental method plan. The paper also explains threats to validity as well as nuances involved in interpreting results from an empirical study.

An experimental design has four major components: participants, materials, procedures, and measures (Creswell, 2014). Participants denote the subjects from which the required data will be obtained. The participants section should describe the process of selecting and assigning the participants. This involves explaining whether random or non-random procedures will be used to select participants, whether the participants will be randomly assigned to groups (true experimental design) or not randomly assigned (quasi-experimental design), and how systematic bias will be controlled (Creswell, 2014).

Though random sampling is important for avoiding or minimizing bias, at times the researcher may be compelled to use non-random techniques such as convenience sampling due to the nature of the target population (Martin & Bridgmon, 2012). The participants section should also describe the number of participants selected and the number assigned to each group (treatment and control groups). An appropriate sample size should be used to ensure the sample is sufficiently representative of the target population (Kothari, 2004). Measures denote the variables or outcomes that will be measured (Thomas, 2009).

The researcher should clearly identify the independent variables (variables not influenced by other variables) and dependent variables (variables influenced by the independent variables). The treatment variable must be one of the independent variables (Creswell, 2014). Once variables are defined, the researcher must describe the instruments used to measure the variables. This involves explaining the development of the instruments as well as their items, scales, validity, and reliability (Creswell, 2014). This is a particularly important step as lack of instrument validity and reliability may undermine the eventual findings (Thomas, 2009).

The researcher must also describe the materials administered for the experimental treatment. The component of procedures involves describing the actual experimentation process. First, the researcher identifies the nature of experiment to be conducted -- true experiment, quasi experiment, pre-experimental design, or single-subject design (Creswell, 2014). As mentioned previously, a true experiment involves randomization of subjects into groups, while a quasi-experiment involves non-randomization. In a pre-experimental design, only a single group is studied -- there is no control group. A single-subject design entails studying a single subject over time.

The chosen design must be properly justified. Once the design is justified, the researcher describes what will be compared (in terms of how many groups and how many variables). A diagram can be used to depict the procedures. This gives the reader a better grasp of the research design. In addition to the four components, the researcher must explain measures to address validity. Validity essentially means guaranteeing that the outcome(s) is influenced by the intervention and not other factors. Validity may be threatened by internal and external factors (Thomas, 2009).

Threats to internal validity emanate from experimental procedures, treatments, and participant experiences (Creswell, 2014). Major threats include history (time), maturation (changes in participants), regression (extreme scores reverting towards the average over time), selection (choosing subjects with attributes that predispose them to certain outcomes), mortality (subjects may withdraw during the course of the experiment), testing (subjects may become aware of the outcome measures), and communication between subjects in the treatment and control groups (Creswell, 2014). Measures should be undertaken to prevent these threats.

These include exposing both groups to the same external occurrences, recruiting a large sample, keeping both groups separate, as well as choosing subjects randomly, with similar maturation, and without extreme scores. Threats to external validity stem from inaccurate generalization of the findings to subjects, settings, and situations beyond the selected sample (Creswell, 2014). For instance, if the sample included a certain racial group, it may be inappropriate to generalize the findings to another racial group. Empirical results are interpreted with the aim of generalizing the findings to the larger population.

Interpretation entails a number of steps (Bryman, 2008). First, the researcher describes whether the formulated hypotheses and questions were confirmed or refuted (Creswell, 2014). Next, the researcher explains whether the intervention administered had an impact on the subjects. Then, with reference to previous literature and the identified theoretical framework, the researcher highlights the significance or insignificance of the findings. Finally, the researcher.