Implementation of the cause-and-effect conclusion...

Implementing a cause-and-effect conclusion

As we already know from the chapter "General Characteristics and Principles of Experiment Planning", the experimental conclusion about the presence of a causal relationship between two variables presupposes the fulfillment of three basic conditions:

• changing one variable acting as a reason must precede the change of another variable, which is considered as a consequence;

• a relationship must be established between the two variables in such a way that the variance of one variable is partially or completely described by the variance of the other variable;

• the influence of the third variable, which is in a certain relationship to the variables under investigation, should be excluded, but is a side variable.

Realization of the first condition in true experimental plans assumes the use of controlled independent variables. In quasi-experimental plans, instead of controllable independent variables, only measured variables are used, which are only a functional analog of independent variables. Such variables include, in particular, the individual characteristics of the subjects themselves, their physical and personal characteristics, sex, age, education, type of character, habits, social and physical environment of individuals - all those factors that the researcher can manage by virtue of natural or ethical constraints, but which he can observe and measure. That is why the question of the precedence of one factor acting as an analogue of an independent variable to another factor that is considered by the researcher as a dependent variable should be solved first of all in the theoretical plan.

For example, the researcher can assume that the success of teaching in an elementary school is determined by such a personal trait as extraversion or introversion. Since this variable is considered by most researchers as an innate, constitutional property, it is logical to assume that in this reasoning of the researcher the first condition of causal inference will be fulfilled.

On the other hand, this requirement can be fulfilled directly. Suppose, for example, that the researcher is interested in the hypothesis of the connection between the success of the inspector of machine parts and the passage of special tests. One of the variants of the study aimed at testing this hypothesis could consist in the selection of subjects with a low and high score in the test of abilities and further training of subjects. In this case, changing the independent variable, or rather, its analog, would immediately precede the change of the dependent variable.

The second condition for causal inference in quasi-experimental plans in most cases involves the use of the same statistical methods as in conventional experiments. Let us recall that in this capacity, as a rule, various variants of the variance analysis or Student's test are used in the case of single-level plans or their nonparametric analogues. In the correlation studies, various varieties of multivariate regression analysis are also widely used. The relationship between the variables in this case is established on the basis of the estimation of correlation coefficients and regression. Quasi-experimental studies are also characterized by the use of the ANCOVA covariance analysis , which is a combination of dispersion and regression analysis methods.

But the most important, in fact the key difference between the quasi-experimental and correlation plans from the true experimental plans is the fulfillment of the third condition concerning control of the side variables.

The main threats to the valid conclusion in the correlation (quasi-experimental) studies are:

- non-fulfillment of the condition of randomization of the selection of the subjects into groups, the consequence of which is their nonequivalence;

- consideration as an analogue of an independent variable of the difference between groups, which was introduced as the basis for nonequivalence of groups.

Since in such experimental plans the researcher intentionally limits direct experimental control, preference is given to indirect, indirect control. In contrast to direct control, which is already in use at the planning stage of the pilot study, ensuring its approach to the ideal experiment, indirect control is carried out after the experiment is carried out and the data collected during the experiment are subjected to statistical analysis. That's why this kind of control is called statistical.

Thus, instead of stabilizing possible side-effects by means of primary control schemes, the researcher using the quasi-experimental method, on the contrary, collects an excessive amount of data, usually reflecting the different individual characteristics of the subjects and/or the features of their habitat. These data can be little related directly to the experimental hypotheses under investigation, but at the same time can largely reflect the effect of various side variables, the elimination or stabilization of which may be impossible or undesirable for various reasons. In this case, the more side variables will be investigated in the course of statistical analysis, the more accurate and reliable the causal conclusion will be.

As a means of indirect control, it is common to distinguish three complementary methodological procedures:

- control based on the application of mathematical methods of covariance analysis ANCOVA,

- step by step control post factum ;

- linear structural modeling based on multivariate regression analysis and confirmatory factor analysis.

thematic pictures

Also We Can Offer!

Other services that we offer

If you don’t see the necessary subject, paper type, or topic in our list of available services and examples, don’t worry! We have a number of other academic disciplines to suit the needs of anyone who visits this website looking for help.

How to ...

We made your life easier with putting together a big number of articles and guidelines on how to plan and write different types of assignments (Essay, Research Paper, Dissertation etc)