Evaluation Of The Choice Of Design And Threats To Validity In A Quasi-Experimental Design

In criminology, research is a crucial aspect in assessing outcomes of a particular scope of information and study. Controlled research experiments allow a criminologist researcher to come up with unambiguous pivotal relationships between two variables. However, many phenomena that may be of interest to a researcher in the field of criminology are not open to clear-cut applications of experimental designs.

To begin with, a quasi-experimental design together with a cross-sectional one present themselves, by and large, as weaker designs based on core validity. Compared to experimental designs which are perceived to be stronger, quasi-experimental designs are only subject to data analysis methods as a technique of control and therefore do not need randomization. They however, on the basis of using more samples over a long period of time, depict themselves superior than cross-sectional designs.

Read also 4 Steps Of Evaluating An Argument – Evaluating Truth and Validity

In contrasted-group design, a dependent variable gets observed in a couple of groups that are different, grounded on independent variables. Internal validity however appears to be a setback because is weak in this design. However, it can be compensated by expanding both the variety and the number of contrasted groups. Additionally, it can be compensated again by using many observations rather than a single observation of the dependent variable(Chava & Nachmias, 1996).

On another platform, planned variations alienate a number of studies on the dependent variable before the independent variable was introduced and thereafter. The design allows a criminology researcher to do away with testing and maturation as probable accounts of changes after the independent variable is introduced(Chava & Nachmias, 1996). Regression and history effects are much harder to do away with, specifically if the researcher does not make observations over an extended period of time.

Read also Levels of Measurement and Concept of Validity

In the control series designs, the time series are generally developedfor both the number of “control” groups and the “experimental group.” The “control” groups considered here are those that may appear to be fundamentally equivalent to the “experimental” group. Since randomization was not used to form the groups, they are not particularly equivalent. However, they offer protection from the threats posed to internal validity of history, testing and maturation(Chava & Nachmias,1996).

Quasi-experimental designs, if scrutinized carefully, expose certain weaknesses. A researcher, therefore, can at times combine two or more of these sub-designs in one study so as to allow strengths of each one of them to compensate for the flows of the other. The researcher may be required to construct and develop smaller experiments within the larger quasi-experimental framework.

Read also Evaluating Design Choice and Threats to Validity in An Experimental Design

Two of the most common and basic problems in quasi-experimental research are generalizing the findings and inferring causation. Sometimes, a researcher can sacrifice generalizability so as to secure unmistakable evidence about causation. It is simply because the two problems create a dilemma in the particular research. The arguments further lead us to the simplified version of this problem; the relationship between external and internal validity. When external validity becomes strong, then internal validity correspondingly becomes weaker and vice versa.

As opposed to experimental designs, quasi-experimental designs are less internally valid most of the times but are very easy to generalize from. The difference hinges mostly on if representative sampling or randomization is used on the artificiality of the study background. The tight spot between internal and external validity can be partly dealt with using demonstrative samples of satisfactory described populations in experiments and by also seeking supplementary information to discard certain opposing hypotheses in investigations.

Scroll to Top