Module 4 Nonexperimental Quantitative and Mixed Designs
Summary (with chapter references for both 4th/5th editions)
Nonexperimental quantitative designs
Chapter 13/14 In non experimental quantitative research, a variety of data collection methods (eg tests, questionnaires, interviews or observation) are used for the purpose of describing, predicting or explaining some phenomenon. Of these purposes, only the latter two involves the investigation of relationships between cause (ie independent variable(s)) and effect (dependent variable(s)).
It is called nonexperimental because in these designs the investigator cannot directly manipulate the causes (NB as we will see in the next module, the direct manipulation of causal factors is one hallmark of an experimental design). Two typical designs are the causal-comparative and the correlational approach.
In order to establish causation, investigators must ensure three conditions are met, the cause and the effect must be related, the cause must precede the effect, and the effect must not result from some other cause. Researchers must therefore be on the lookout for spurious relationships and control for other (ie extraneous) causes. There are several available techniques of control including matching, holding extraneous variables constant, and statistical methods.
Nonexperimental research can also involve studying changes over time, and there are several designs that are appropriate (ie cross-sectional, longitudinal and retrospective).
Many texts on research methods include sections devoted to sample survey designs. However, in this course sample survey designs, whether descriptive or correlational in nature, are subsumed under nonexperimental quantitative designs. A sample survey is just a nonexperimental approach with a particular sampling procedure (ie usually some form of probability sampling) and a particular method of data collection (ie usually questionnaires or interviews).
Analysis of quantitative
In recent years, there has been a growing trend to combine elements of both qualitative and quantitative approaches within the one study. This approach results in a mixed design. In general terms there are two types of mixed designs, mixed model and mixed method.
Mixing quantitative and qualitative elements within a single study is advantageous because while the weaknesses of the two component designs tend to cancel each other out, the strengths of both are retained. However, mixed designs tend to be quite complex, expensive, time consuming and require a detailed understanding of both qualitative and quantitative paradigms.
Analysis of quantitative data, whether derived from a pure or mixed design, requires an understanding of both descriptive and inferential data statistics. Whilst these are covered in the text, they are not being explicitly dealt with in this course.
Make sure you understand these key concepts
Nonexperimental Quantitative Research
- Cause and Effect Relationships
- Independent Variables
- Dependent Variables
Techniques of Control in Nonexperimental Research
- Holding Extraneous Variables Constant - Statistical Control
Time Dimension in Research
- Cross-Sectional Research
- Longitudinal Research
- Retrospective Research
Objectives of Nonexperimental Research
- Descriptive Research - Predictive Research
- Explanatory Research
- Mixed Research