On Quasi-experimental Design and Covariance Analysis

The study of Plan A includes the intention of a series of process elements, the observed changes and their effects, and the statistical analysis of these results to determine the relationship between changes in the process, so as to change the process.

definition

In the experimental design of experimental psychology, the definition of experimental design is:

broad sense

Generalized experimental design refers to the general procedure of scientific research knowledge, including result analysis, hypothesis formation, variable selection, paper writing and so on. It enables researchers to show how to conduct scientific research and the research problems to be solved in the whole process.

narrow sense

Experimental design of stenosis refers to the implementation of planned experimental treatment scheme and statistical analysis scheme.

The narrow sense of experimental design focuses on how to establish the statistical hypothesis conclusion of this period.

& ltBr experimental design activities are as follows:

1。 Establish statistical assumptions and assumptions;

2。 The experimental treatment of measurement (independent variables) is used in experiments, and the excessive conditions (additional variables) must be controlled; 3。 Determining the pilot unit (object) requires a large number of experiments and sampling people.

4。 Assigned to the determined experimental conditions;

5。 Determine the statistical analysis to be recorded (dependent variable) and used in each measurement experiment.

The main steps, methods and strategies of the experimental scheme are studied by the researchers according to the research purpose before the experiment. Its main content is the reasonable arrangement of the experimental process, and puts forward how to use it in the statistical analysis of experimental data. The main steps of psychological experiment design can be summarized as follows: ① According to the research purpose, hypothesis is the process of verifying hypothesis; ③ Select appropriate experimental data processing and statistical analysis methods.

The characteristics of experimental design, its main function of evaluation is to control variables, first of all, effectively manipulate or change the changes of independent variables and dependent variables under controlled conditions (observation of response variables). For example, the study of children's academic performance by two teaching methods and the arrangement of experimental design make other conditions as same as possible, such as family and school environment and similar academic foundation. Two groups of children of the same age should not only control the use of two different teaching methods, but also check the impact on all academic performance. It is mainly manifested in the reasonable arrangement of experimental design and the effective control of irrelevant variables. Independent variables in psychological experiments, such as physical and chemical experiments, are the most difficult to be excluded, so we must rely on experimental design that balances or counteracts the influence. This control method is called experimental control method, and there are several commonly used methods:

① Method of excluding or keeping unchanged: under the main laboratory conditions, exclude the interference of irrelevant variables, and keep the variables that cannot be excluded, such as age, weight and test level of experimental environment, as much as possible;

② Balance method: random sampling method, points? The experimental group and the control group have the same influence on the independent variables of the two groups;

(3) Compensation method: its purpose is to control the experimental sequence and period of influence (there are only two experiments for lawyers AB BA);

(4) Inclusion method: the independent variables are treated as independent variables in some way, and the results of single-factor and multi-factor design and multivariate statistical analysis are tested to find out the functions and interactions of their respective variables. Some independent variables know the influence of the results, but limited by the experimental conditions, they can't balance or offset the experimental control, so they can only use statistical methods to analyze and exclude conclusions after the experiment. This control method is called statistical control method. Covariance analysis or covariance analysis of common statistical control methods. This method is often used when random sampling cannot be based on individuals due to difficulties or administrative reasons, and it is necessary to maintain the integrity of individuals (such as classes).

There are many criteria to evaluate the experimental design, but it mainly depends on whether it can give full play to the following functions: (1) The experimental design must match the problems to be properly solved; (2) Better "internal validity", that is, it can effectively control irrelevant variables, and the changes of response variables are completely independent variables; (3) The experimental results should be scientific and universal, and can be extended to other disciplines or other situations, with high external validity.