Components of the Experiment--论文代写范文精选

斯金纳的结论操作性实际上是基于精心设计的,满足复杂的归纳原则。实验的本质实际上是一个特有的猜想与反驳，在概念层次上的努力。实证研究必须至少有两个测试条件。下面的report代写范文进行详述。

Summary 
The experiment is an arrangement for collecting research data, in which there are two or more conditions that are identical in all aspects but one. The aspect in which the test conditions differ is the independent variable. Both deductive and inductive logic are used in experimentation, albeit at different stages for different purposes. While deductive logic is used to derive the experimental hypothesis from the substantive hypothesis, inductive logic is the foundation of the experimental design. The theoretically informed control variables and methodologically informed control procedures are responsible for the feature that differentiates the experiment from non-experimental studies. The feature in question is the provision for experimental controls, the function of which is to exclude recognized alternative explanations. 

Introduction 
Skinner once said that conducting experiments involved nothing more than measuring subjects’ simple, countable behaviors while manipulating some randomly selected aspect of the environment. No planning is required. Skinner gave the impression that experimentation is a chancy trial-anderror exercise suitable only for studying simple, countable phenomena that can be shaped by experimenters. However, Skinner’s conclusions about operant conditioning were actually based on carefully designed experiments that satisfy sophisticated inductive principles. The “trial-and-error” 1 nature of experimentation is actually a characteristic of the Popperian “conjectures and refutations” endeavor at the conceptual level. Empirical research must have at least two test conditions that satisfy certain stipulations before it can be characterized as an experiment. The present discussion begins with a description of the components of the experiment. A distinction is then made between utilitarian and theorycorroboration experiments. The relationship between the experimental and control conditions is then explained by making explicit the roles of deductive and inductive logic at various stages of experimentation. The discussion concludes with some meta-theoretical issues and their implications.

Types of Variables 
A variable is anything that can be identified in more than one way. For example, musical key in Table 1 is a variable because it is represented either by a major key or a minor key. In conducting an experiment, psychologists manipulate the independent variable and measure the dependent variable while holding the control variables constant. There are also the extraneous variables that, while not explicitly identified, are nonetheless assumed to have been held constant by virtue of the appropriate control procedures found in the experiment.

The Independent, Control, and Dependent Variables 
The psychologist in Table 1 manipulates musical key by setting up two test conditions, one with a piece of music in a major key and another with a piece of music in a minor key. Musical key is the independent variable in the sense that the two conditions are set up independently of what the subjects do. As may be seen from Table 1, adaptation duration (i.e. the time spent listening to music in Phase I), tempo, timbre and performer are held constant when the experimenter uses the same level of each of them in both the major-key and minor-key conditions. In such a capacity, they are the control variables. Subjects’ recognition performance is measured (e.g. with the index of sensitivity d'). It is the dependent variable because its values depend on the subjects.

Experimental Designs 
The next feature of the experiment is its design: the formal arrangement of (a) the independent, dependent, and control variables, (b) the sequence of events in the course of the experiment, (c) the sequence of events to be carried out in a trial, and (d) how subjects are assigned to the test conditions. The numerous experimental designs may be categorized in terms of (i) the number of independent or dependent variables, and (ii) the manner in which the subjects are assigned to the test conditions.

Designs and the Number of Variables 
Designs and Subject Assignment In terms of how subjects are assigned to the test conditions, the 1-factor design may be a completely randomized or a repeated-measures design. When the completely randomized design is used, subjects are assigned randomly to the test conditions. By “random assignment” is meant that whoever is included in one condition does not determine, or is not determined by, whoever is assigned to another condition. In contrast, the same subject is tested in every test condition found in the experiment when the repeated-measures design is used. In the same vein, multifactor factorial designs fall into four categories: the completely randomized, the repeated-measures, the randomized block, and the split-plot designs. Recall that there are numerous extraneous variables in any experiment. Suppose that individuals are not assigned randomly to the test conditions. 

Instead, those who have won a scholarship are assigned to the major-key condition, whereas individuals who have just failed a quiz are assigned to the minor-key condition. Common sense suggests that the major-key group is a happy group and the minor-key group is a less happy group to begin with. Under such circumstances, being happy is a confounding variable because its two levels are yoked to those of the independent variable. Hence, any difference between the two musical-key conditions could have been due to the differences between the two levels of being happy. In short, no extraneous variable should be confounded with the independent variable. The sole purpose of using the completely randomized design is to minimize such confounding. Random subject assignment ensures that, in the long run, the ratio of being-happy to being-less-happy subjects would be the same at both levels of musical key. In the event that random subject assignment is insufficient for holding constant a potential confounding variable, experimenters may test every subject in all test conditions (i.e. using the repeated-measures design). This procedure may minimize confounding. Consider the experiment described in Table 1. Regardless of individual “happiness” level, it would be the same in the majorkey and minor-key conditions if an individual is being tested in both of them. However, it is not always possible to use the repeated-measures option. First, subject fatigue may become an issue when they are being tested for a longer period of time. Second, there is also the potential difficulty due to the order of testing, as may be seen from Table 3.

 The order of testing in row 1 is AB11, AB12, AB21, and AB22. The possible source of ambiguity is that the subjects’ performance in any of the other three treatment combinations might be different had they not been tested previously in AB11. The same difficulty applies when subjects are tested first in AB12, AB21, or AB22. This source of ambiguity is one exemplification of the “order of testing” effects. A distinction is made between univariate and multivariate designs. There is only one dependent variable in the univariate design, whereas two or more dependent variables are used in the multivariate design. Experimental designs are also classified in terms of the number of independent variables: 1-factor design (i.e. one independent variable) and multifactor design (two or more independent variables). Regardless of the number of independent variables used, designs are further identified in terms of the number of levels used to represent the independent variables. For example, designs involving only one independent variable may further be distinguished between the “1- factor, 2-level” and “1-factor, multilevel” varieties. As the names suggest, only two levels of the independent variable are used in the former, and more than two levels are used in the latter.

The Inductive Foundation The experimental design owes its importance to its underlying inductive principle whose function is to reduce ambiguity in data interpretation. As an illustration, underlying the 1-factor, 2-level design 5 depicted in Table 1 is J.S. Mill’s method of difference. The idea is to set up two conditions that are identical in all aspects except one. Specifically, the two levels of the independent variable are used to set up the two otherwise identical conditions. The force of the method of difference (or any inductive principle) is a negative one (albeit very important). Note specifically that a difference in d' is found despite the fact that adaptation duration, tempo, timbre, and performer are held constant in both conditions. In other words, they are irrelevant to the observed difference in d'. Consequently, they can be excluded as explanations of the data. That is, the inductive principle makes it possible to exclude specific alternative causes.

Three Technical Meanings of “Control” 
The important interpretation-exclusion function of inductive logic is encapsulated in the experimental control whose three components are (i) the provision for excluding confounding variables, (ii) the constancy of condition, and (iii) the valid comparison baseline. First, the possibility of having a confounding variable may be minimized by a procedure such as randomizing the order of stimulus presentation or counterbalancing the order of testing if the repeated-measures design is used. Either of them is a control procedure used to exclude a possible artifact. Although control procedures cannot be seen from the schematic representation of design (such as Table 1), they are (or should be) described in full in the “Procedure” section of the experimental report. Second, there are two aspects to the constancy of condition in the experiment. The first is the stipulation that the predetermined levels of the independent variable be applied consistently throughout the experiment. For example, if the two pieces of music used in the experiment begin in the key of C major and C minor, these should be used throughout the experiment. The second aspect is the better-known provision of control variables, as illustrated by the variables adaptation duration, tempo, timbre, and performer in Table 1. They are held constant in the sense that the same level of each is used at the two levels of musical key (for the attempts to achieve constancy of condition in non-experimental research, see Interviewing and Observation). Third, to be able to conclude that the difference between the two musical-key conditions is not due to an artifact, it is necessary to ensure that the test conditions are identical in all aspects except for the level of musical key. If the repeated-measures design is used, either of the two musical key conditions in Table 1 satisfies this stipulation when it is used to assess the effect of the other level. In sum, the three components of experimental control serve collectively to exclude explanations other than the independent variable. This is very different from the Skinnerian use of “control” because, as has been shown, experimental control has nothing to do with constraining or shaping what experimental subjects do. If one were to use “control” in the Skinnerian sense to mean constraining or shaping behavior, it is researchers’ data-interpretation that is being constrained or shaped. Specifically, researchers are prohibited logically from appealing to factors that are used explicitly as control variables or procedures.

51Due网站原创范文除特殊说明外一切图文著作权归51Due所有；未经51Due官方授权谢绝任何用途转载或刊发于媒体。如发生侵犯著作权现象，51Due保留一切法律追诉权。(论文代写)
更多report代写范文欢迎访问我们主页 www.51due.com 当然有report代写需求可以和我们24小时在线客服 QQ:800020041 联系交流。-X(论文代写)