Experimental Planning and Execution

TY - CHAP

T1 - Experimental Planning and Execution

AU - Mullane, Kevin

AU - Curtis, Michael J.

AU - Williams, Michael

PY - 2017/10/10

Y1 - 2017/10/10

N2 - The lack of reproducibility in biomedical research has been consistently attributed to the misuse of statistics, naïve inference/extrapolation from animal models, and fraud. However, optimization of statistical analysis and caution in interpreting experimental outcomes cannot overcome flaws in experimental design and execution. These include: the absence of a credible hypothesis for running an experiment; clearly defining outcomes before the experiment is run; the use of appropriate controls, randomization of subjects; the validation/authentication of all reagents (compounds, cell lines, animals, antibodies) and equipment; the collection, preservation, and annotation of raw data; and steps to contain or eliminate investigator bias. Collectively these measures will avoid a GIGO (garbage in, garbage out) paradigm that no amount of postexperimental tinkering, including P-hacking, can objectively and legitimately correct. The present chapter addresses measures to facilitate best practices in experimental design and execution that will aid in improving reproducibility.

AB - The lack of reproducibility in biomedical research has been consistently attributed to the misuse of statistics, naïve inference/extrapolation from animal models, and fraud. However, optimization of statistical analysis and caution in interpreting experimental outcomes cannot overcome flaws in experimental design and execution. These include: the absence of a credible hypothesis for running an experiment; clearly defining outcomes before the experiment is run; the use of appropriate controls, randomization of subjects; the validation/authentication of all reagents (compounds, cell lines, animals, antibodies) and equipment; the collection, preservation, and annotation of raw data; and steps to contain or eliminate investigator bias. Collectively these measures will avoid a GIGO (garbage in, garbage out) paradigm that no amount of postexperimental tinkering, including P-hacking, can objectively and legitimately correct. The present chapter addresses measures to facilitate best practices in experimental design and execution that will aid in improving reproducibility.

KW - Hypothesis generation

KW - Investigator bias

KW - Randomization

KW - Reagent validation/authentication

KW - Reproducibility

UR - http://www.scopus.com/inward/record.url?scp=85045183037&partnerID=8YFLogxK

U2 - 10.1016/B978-0-12-804725-5.00002-1

DO - 10.1016/B978-0-12-804725-5.00002-1

M3 - Chapter

SN - 9780128047255

SP - 67

EP - 106

BT - Research in the Biomedical Sciences

PB - Elsevier

ER -