Campbell, D.T. and Stanley, J.C. Experimental and QuasiExperimental Designs for Research. Houghton Mifflin, Boston, MA, 1963.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the Observer Executor method to help the subjects understand if they were behaving properly. The subjects were also given a chance to ask the experimenter questions about either PBR or the observation procedure. 4.4. 4 Execution A quasi experimental, factorial design with two treatments was used [Campbell63]. In the first treatment, roughly half of the teams inspected the LA requirements and the other half the PGCS requirements. After this inspection was complete, the team members switched roles, i.e. the process observer in the first inspection became the process executor for the second inspection. ....

Campbell, D., and Stanley, J. Experimental and Quasi-Experimental Designs for Research. Houghton Mifflin Company, Boston, 1963. 19

....is the only empirical approach which is characterized by this property. The rationale for this control is to limit the number of variables in the study. There are various experimental designs that may be applied where the effects on the environment have already been considered and identified (Campbell, 1963). Quasi experiments are empirical studies where the researcher has some, but not complete, control over the environment (Cook, 1979) Thus, the quasi experiment is somewhere between an experiment and a case study. 4.1.2 History In a historical inquiry, the researcher seeks explanation of the ....

....context of a controlled experiment involving software engineering processes. The naming and definitions of the individual threats are from Cook and Campbell (Cook, 1979) There are numerous ways of categorizing threats to validity, and none is exhaustive and applicable for all kinds of experiments (Campbell, 1963; Kish, 1987) 4.3.3.1 Statistical Validity The results of an experiment usually involves statistical inference based on the data to test whether there is a significant relation between two variables. For some reasons, it may be possible to draw false conclusions about the covariation of two ....

[Article contains additional citation context not shown here]

Donald T. Campbell and Julian C. Stanley. Experimental and Quasi-Experimental Designs for Research. Houghton Mifflin Company, 1963.

....of word processor conditions. The design is a 2 (personality types, between subjects) X 3 (levels, levels 1,3 = MSWord 2000, level 2 = MSWord Personal, within subjects) design where level 2 is nested with 5 repetitions. This design is best described as a quasi experimental design [1]. Measures The dependent measures were based on logging data and data collected from the 7 questionnaires. From the logged data we extracted the total time spent word processing, the time spent in each interface, the number of toggles between interfaces, and a trace of the ....

Campbell, D.T., and Stanley, J.C. (1972). Experimental and quasi-experimental designs for research. Chicago, IL: Rand McNally & Company.

....then the theory has failed to account for the different variables of the new study, and then it must be updated in an iterative learning process. On the contrary, if the hypotheses are confirmed, then the theory has survived another probe, and thus we have more confidence in its predictions (Campbell, 1963). 4. Classification of the Evaluation Studies In this section we classify the six position papers according to the multidimensional framework shown in Table 1. The first dimension, the object of study, is the thing being investigated (e.g. for a study of technology evaluation, the object is ....

Campbell, D. T., and Stanley, J. C., 1963. Experimental and Quasi-Experimental Designs for Research. Boston: Houghton Mifflin Co.

....even tertiary) effect on the result. See 3 Note this is a simplification of the exact case, as while power and type 1 error rates are related, one is not a direct function of the other. 4 Assuming the effect size is fixed and not massive. 5 Also known as triangulation. Campbell and Stanley [5] for a extensive treatment of this area. Good experimental design attempts to limit or minimise these threats, but the number and scale of many of the threats are such that they can only be explored with large amounts of resources. Hence, the need for multiple researchers to explore these ....

D.T. Campbell, J.C. Stanley, "Experimental and Quasi-experimental designs for research", Boston, MA: Houghton Mifflin Company, 1963.

....which it was created. At this level, researchers are evaluating if it is worthwhile to spend the resources required to continue through the methodology. In order to gather this type of information, researchers should use feasibility studies (sometimes referred to as quasi experimental designs [6]) in which data is collected according to some experimental design, but full control over all possible variables is not achieved. Such studies attempt to test the effectiveness of a process but are not able to rule out all rival hypotheses that may still exist at the end of the study. For example, ....

Campbell, D.; and Stanley, J. Experimental and Quasi-Experimental Designs for Research. Houghton Mifflin Company, Boston 1963.

....was completed. 5.5 Validity Threats In this section we discuss the limitations that our case study design had with respect to validity of the results. As we had only a single project in the study we were forced to apply the methods in sequence and this may have lead to some maturation effects (Campbell and Stanley, 1963; Judd et al. 1991) i.e. the accumulated time spent on risk management may have increased participant s awareness and knowledge about risks. We tried to minimize this effect by taking two specific actions. First, even though the dedicated risk identification session is a characteristic of the ....

Campbell, D.T. and Stanley, J.C. (1963) Experimental and Quasi-Experimental Designs for Research, Boston: Houghton Mifflin Co.

....anecdotal case descriptions. In the following we present some highlights of a study performed to evaluate the Riskit method in practice 2 . We recognized the previously listed constraints in the design of the study but attempted to apply some well known case study principles whenever applicable [4,17,34]. The case study organization was a mature risk management organization with an experienced project manager and project team. We arranged our case study so that we were able to apply the two risk management methods during the project. As Figure 3 shows, the case study started by a joint session ....

D. T. Campbell and J. C. Stanley. Experimental and Quasi-Experimental Designs for Research, Boston: Houghton Mifflin Co. 1963.

....handling of withdrawals. A paper on guidelines for contributors to journals by Altman [1] The guidelines for statistical review of general papers and clinical trials prepared by the British Medical Journal. These guidelines are listed in Altman et al. 3] chapter 10 of Gardner and Altman [14], and on the journal s web page: http: www.bmj.com advice) A book by Lang and Secic [28] with guidelines for reporting medical statistics. The CONSORT statement on reporting the results of randomized trials in medicine [4] This statement has been adopted by seventy medical journals. 3 . ....

....pp121 127. Also available in English at http: wwwinfo.ncc go jp jjco ] 13] P. Fusaro, K. El Emam, and B. Smith, Evaluating the interrater agreement of process capability ratings, Proceedings of the Fourth International Software Metrics Symposium, IEEE Computer Society Press, 1997, pp. 2 11. [14] M. J. Gardner and D. G. Altman, Statistics with Confidence, BMJ, London, 1989. 15] L. Gordis. Epidemiology. W.B. Sunders Company, 1996. 16] D. Heinsman and W. Shadish, Assignment methods in experimentation: When do nonrandomized experiments approximate answers from randomized experiments ....

D. Campbell and J. Stanley, Experimental and Quasi-Experimental Designs for Research, Houghton-Mifflin, 1990.

....proved sometimes counterintuitive. Fagin 11 defined logic like operators, so that the user can request images similar to the query with respect to criteria i and j, with respect to criterion i or criterion j, and not with respect to criterion i. If all the similarity measures are normalized in [0, 1], the three logic predicates can be implemented by functions f # , f # : 0, 1] 2 # [0, 1] and f : 0, 1] # [0, 1] so that the score for image I is given in the three cases by s = f # (s i , s j ) Figure 4. The 12 shapes that were used in the similarity algebra experiment. s = f # ....

....the user can request images similar to the query with respect to criteria i and j, with respect to criterion i or criterion j, and not with respect to criterion i. If all the similarity measures are normalized in [0, 1] the three logic predicates can be implemented by functions f # , f # : [0, 1] 2 # [0, 1] and f : 0, 1] # [0, 1] so that the score for image I is given in the three cases by s = f # (s i , s j ) Figure 4. The 12 shapes that were used in the similarity algebra experiment. s = f # (s i , s j ) s = f (s i ) Fagin proposed a number of possible functions, among ....

[Article contains additional citation context not shown here]

D. Campbell and J. Stanley, Experimental and Quasi-Experimental Designs for Research, Rand McNally College Publishing Company, Chicago, 1963.

....The use of an algorithm for perception as a tool for synthesis can point to the specific part of the motion that the algorithm is sensitive to. We use a factorial experiment to test the perception. Ideally, we should use the posttestonly control group design described by Campbell and Stanley [7], but the number of unordered combinations of factors is too numerous and forces us to use some repeated measures. Section 4.1 describes the actual method we used in our pilot study. We discuss a final design in Section 5. In the experiment, we present subjects with a motion stimulus and ask them ....

....from drawing conclusions from the data and analysis, and focus on what needs to be changed before collecting data in the final version. Our data collection method is prone to multiple treatment interference affecting the external validity (the ability to generalize the results) of the experiment [7]. For the full factorial experiment, each subject views 40 motion sequences, but there are only five different motions in all. A look at the randomized order in which the subjects view the stimuli shows that by the 12th sequence the subjects had viewed all five motions for at least two seconds. By ....

[Article contains additional citation context not shown here]

D. T. Campbell and J. C. Stanley. Experimental and quasi-experimental designs for research. Rand McNally, Chicago, 1963.

....assignment of subjects to treated and untreated groups, which is not feasible for largescale infrastructure projects like dams. The design used in this study is a specific kind of quasiexperimental method. It combines the interrupted time series design and the nonequivalent control group design (Campbell and Stanley 1966). The strength of the former method is its attention to history as a plausible explanation for the observed effects. The strength of the latter is its attention to cross sectional factors. These designs are combined by carefully selecting a quasi experimental control group during a calibration ....

Campbell, D. T., and J. C. Stanley. 1966. Experimental and quasi-experimental designs for research. Chicago: Rand-McNally.

....to drive a vehicle equipped with one of the four aforementioned information delivery systems for a relatively extended period of time while observing their behavior and asking their perceptions and opinions vis a vis the system. Table 1 shows the quasi experimental design used for the study [Campbell and Stanley, 1963]. The study is quasi experimental because the incidents that subjects experience are actual highway incidents and thus not controlled as they would be in a true experiment. As the table shows, the evaluation was designed as a series of periods of groups of subjects, each using a different ....

Campbell, D.T. and J. C. Stanley. ( 1963). Experimental and quasi-experimental designs for research. Chicago: Rand McNally.

....costs of experimentation while making it more difficult to achieve statistical significance. The degree of credibility of any study depends on the validity of how conclusions are drawn. Campbell and Stanley have defined two classes of evaluation criteria: internal validity and external validity [15]. Internal validity defines the degree of confidence in a cause effect relationship between factors of interest and the observed results, while external validity defines the extent to which the conclusions from the experimental context can be generalized to the context specified in the research ....

....we have looked at models of an empirical study s important components. In this section, we discuss briefly the experimental designs in which these models fit. This paper is not meant to be a primer for how to run experimental studies. Interested readers can find helpful guidelines for this task in [15, 25, 33], and more specifically for the software engineering domain in [20, 46] However, in this section we discuss the impact of process, product, and context models on the kinds of designs that are feasible and useful for software engineering experimentation. 7.1. Experimental Design As discussed in ....

[Article contains additional citation context not shown here]

D. T. Campbell, and J. C. Stanley, Experimental and Quasi-Experimental Designs for Research, Boston: Houghton Mifflin Co, 1963.

....over time and space. Several additional advantages of the method over conventional regression analysis are discussed in the following section. 3. Quasi experimental matching method The method introduced here is a combination of non equivalent control group and interrupted time series methods (Campbell and Stanley 1963). It is applied to multidimensional data in a way that allows the behavior of several variables to be examined at the same time. It improves upon previous methods (Isserman and Merrifield 1982, 1987; Isserman and Beaumont 1989) by choosing a control group from both theoretical and statistical ....

Campbell, D. T., and J. C. Stanley, 1963, Experimental and quasi-experimental designs for research, (Rand McNally, Chicago).

....research setting [80] 12 . 11 In a repeated measures design, subjects receive two or more treatments. 12 Campbell and Stanley point out that even in well controlled true experiments, there are often nonrandom nuisance variables inherent to the experimental design that cannot be controlled [8]. Empirical support for this position can be found in [38] 13 3.2.4 Counterbalancing A danger with repeated measures designs is the existence of a carry over effect [33] A carry over effect can occur in two ways: practice effects and sequence effects. With the former, treatment effects are ....

....work in [2] 83] Effect of Intact Groups. As noted earlier, the groups in our study were intact (i.e. they were not formed randomly) The particular situation where there is an inability to assign subjects to groups randomly in a counterbalanced design was discussed by Campbell and Stanley [8]. In this design there is the danger that the group interaction with say practice effects confounds the treatment effect. However, if we only interpret a significant treatment effect as meaningful if it is not due to one group, then such a confounding would have to occur on different occasions in ....

[Article contains additional citation context not shown here]

D. Campbell and J. Stanley. Experimental and Quasi-Experimental Designs for Research. Houghton Mifflin, Boston, 1966. ISBN 0-395-30787-2.

....as location, environment, and task. It may be that certain problems tend to occur together, for example, or that certain problems occur in coding, but not in testing. This information will be essential for understanding how to design solutions. Finally, we will use experiments and quasiexperiments [2] as appropriate. When it is possible either to arrange suitable control conditions or to exploit naturally occurring opportunities for caparisons, experimentation will generally be the method of choice. In the following sections, we discuss our initial plans for addressing these problems. They are ....

Campbell, D. T. and Stanley, J. C., Experimental and Quasi-Experimental Designs for Research. Boston, MA: Houghton Mifflin, 1963.

No context found.

Campbell, D.T. and Stanley, J.C. Experimental and QuasiExperimental Designs for Research. Houghton Mifflin, Boston, MA, 1963.

No context found.

Campbell D. and J. Stanley, Experimental and quasi-experimental designs for research, Rand McNally, Chicago, (1963).

No context found.

Campbell, D.T. and Stanley, J.C. Experimental and Quasi- Experimental Designs for Research. Houghton Mifflin, Boston, MA, 1963.

No context found.

Campbell D. T., Stanley J. C. Experimental and Quasi-Experimental Designs for Research. Rand McNally College Publishing Company, Chicago, IL, 1963.

No context found.

Campbell, DT, Stanley JC. Experimental and quasi-experimental designs for research. Rand - Mcnally Publishing CO, Chicago. 1963.

No context found.

Campbell, Donald T & Stanley, Julian C. Experimental and Quasi-Experimental Designs for Research, 1963. Library of Congress Catalogue Card Number: 66:25213

No context found.

Campbell, D.T., and J.C. Stanley, Experimental and Quasi-Experimental Design for Research, Rand McNally and Company, Chicago, 1963.

No context found.

Campbell, Donald T.; and Stanley, Julian C.: Experimental and Quasi-Experimental Designs for Research. Houghton Mifflin Co., 1963.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC