Example: quiz answers

Study Design - Jones & Bartlett Learning

222 Study DesignDesigning a research Study is not a simple task. Just as the key elements anddeterminants of outcomes of war are fixed even before it is fought ( , at itsplanning stages), the success of a Study is also largely determined at its initialplanning stage. Successful studies need to address two important dimen-sions: reliability and validity. A reliable Study should be replicable, providingsimilar results if the same Study parameters are applied. Validity is concernedwith the ability of the Study to correctly answer the question it asks.

22 2 Study Design Designing a research study is not a simple task. Just as the key elements and determinants of outcomes of war are fixed even before it is fought (i.e., at its

Tags:

  Research, Design, Study, Outcome, Designing, Study design, Research study, Study design designing

Information

Domain:

Source:

Link to this page:

Please notify us if you found a problem with this document:

Other abuse

Transcription of Study Design - Jones & Bartlett Learning

1 222 Study DesignDesigning a research Study is not a simple task. Just as the key elements anddeterminants of outcomes of war are fixed even before it is fought ( , at itsplanning stages), the success of a Study is also largely determined at its initialplanning stage. Successful studies need to address two important dimen-sions: reliability and validity. A reliable Study should be replicable, providingsimilar results if the same Study parameters are applied. Validity is concernedwith the ability of the Study to correctly answer the question it asks.

2 Internalvalidity deals with the ability of the Study to correctly infer about the relation-ship between the independent variables and the outcome (s) being validity deals with application of the findings to other observations,samples, or populations and implies generalizability of the Study results. Theneed for reliability and validity of studies dictates sound Study designs. outlines some of the important steps in designing a Study , as well as somepossible problems and solutions to common misunderstanding about studies comes into focus in the dis-cussion section of Study reports where most associations are simply assumedto be causal and investigators then proceed to interpret the causal associ-ation.

3 This error is easily avoided if the objectives of a Study are clear beforedesigning it; that is, the investigators should be clear whether their studyand involved analyses address a causal (or other) hypothesis testing situ-ation of a hypothesis generating situation. Ahypothesis testingstudy has toexplicitly state the involved hypothesis a-priori, and the Study should be de-signed and powered to address that question. Therefore, post hoc, second-ary exploratory data analyses would not be a part of the hypothesis testingsituation unless special analytical arrangements are made ( , adjustingfor multiple testing).

4 There is only a limited set of opportunities to find sig-nificant and meaningful results from data already collected if it is 9/16/09 1:36 PM Page 22 Jones and Bartlett Publishers, LLC. NOT FOR SALE OR DISTRIBUTION. Study Design23 BOX Short Checklist for designing a StudySteps Involved in designing a Study Outline of Study Title of Study research question (what is the overall, broad question?) research hypotheses (specific clear and sequential sets of questions to beaddressed) Significance of the Study new data/confirmatory, and why is it impor-tant?

5 Why should anyone other than the interested investigator spend time orresources in this proposal? Comprehensive description of Study Design Inclusion/exclusion criteria; how will participants be recruited? What outcomes and other factors will be measured why and how? How will data be analyzed describe how each variable will be handled;how missing information will be handled; sample size and power? Whatis an important difference to detect (10%, 25%, 5 inches, 10mm of )? How will the data be safeguarded (HIPAA compliance)? Potential problem areas logistics ( , power outage will biologicalsamples be lost?)

6 Safeguards?) Limitations of Study and how can these be corrected? Is the Study proposal approved (approval pending) by institutional IRB?Potential Problems and Solutions in designing Studies research question is too broad/general (Narrow the question; use smaller setof variables) Not enough subjects available (Broaden inclusion criteria; increase time) Methods beyond investigator s skill level (Collaborate, consult, learn) Too expensive (Consider less expensive alternatives, smaller Study , less follow-up) Question not interesting/novel enough (Consult and discuss withmentors/peers; modify research question) Uncertain ethical suitability (Consult with IRB.

7 Modify research question) Vague Study plan (Write proposal early and revise several times) Proposal confusing/unclear (Write proposal in point-by-point manner forspecificity) 9/16/09 1:36 PM Page 23 Jones and Bartlett Publishers, LLC. NOT FOR SALE OR DISTRIBUTION. that more new information is needed, the testing mechanism cannot goany farther. However, post hoc analyses is not a condition sine qua non fordata already collected. The recognition that the Study was not designed fortesting the involved questions in post hoc analyses clearly indicates thatalthough such analyses may be carried out, they can, at best, be used onlyas indicators of potentially new information; that is, viewed as hypothesis-generating analyses.

8 Studies carried out in such manner are correctly clas-sified as hypothesis-generating studies. Therefore, once hypotheses aregenerated, these can then be tested in new, specifically designed may be unrepeated measure studies where measurements aretaken at one time only, and two or more groups are compared for , studies may also be repeat-measure studies where the unit ofobservation is measured more than once ( , the measurements are re-peated). In repeat-measure designs, the second and subsequent measuresfrom a subject are considered to be correlated with the first measurementbecause the basic biological and sociocultural fundamentals of the individ-ual remain the same across all of the observations.

9 Such Study designs needspecial analytical handling, which is discussed in Chapter 8. The analysesused for repeat-measure designs are not specific to studies that incorporatemeasuring the same person multiple times, but are analyses that addressany correlated data. Therefore, from an analytical standpoint, repeat-meas-ure analysis may be considered a special case of correlated data , whether one conducts repeat-measure studies or not, in terms ofcomparison of groups the architecture of studies follow the same paths asdiscussed may be classified as observational or experimental.

10 In both, theeffects of causes may be assessed. The effects are the outcomes, whereas thecauses are usually generically called exposures. The associations exam-ined by the studies are exposure outcome associations, which may bemeasured in different ways. The key difference between an observationalstudy and an experimental Study is the control of the exposure. In experi-mental studies, the investigator controls the exposure and determines whogets the exposure and how much exposure one gets. For example, in a phaseIII clinical trial, the investigator decides the dosage and allocation of thedrug to the participants through a randomizationprocess.


Related search queries