Data Mining as an Industry Export

@article{Denton1985, abstract = {"Data mining" by an individual investigator can distort the probabilities in conventional significance tests. This paper argues that the same effect can occur when a given data set is used by more than one investigator, even if no individual investigator engages in data mining. A problem of publication selection bias is recalled and note is taken of its implications for the interpretation of published test results when there is collective data mining. Some illustrative calculations of probabilities associated with collective data mining are provided.}, author = {Denton, Frank T.}, citeulike-article-id = {4007209}, citeulike-linkout-0 = {http://dx.doi.org/10.2307/1928442}, citeulike-linkout-1 = {http://www.jstor.org/stable/1928442}, doi = {10.2307/1928442}, issn = {00346535}, journal = {The Review of Economics and Statistics}, keywords = {data\_mining}, number = {1}, pages = {124--127}, posted-at = {2009-02-04 12:49:02}, priority = {2}, publisher = {The MIT Press}, title = {Data Mining as an Industry}, url = {http://dx.doi.org/10.2307/1928442}, volume = {67}, year = {1985} }

TY - JOUR ID - Denton1985 L3 - citeulike-article-id:4007209 N2 - "Data mining" by an individual investigator can distort the probabilities in conventional significance tests. This paper argues that the same effect can occur when a given data set is used by more than one investigator, even if no individual investigator engages in data mining. A problem of publication selection bias is recalled and note is taken of its implications for the interpretation of published test results when there is collective data mining. Some illustrative calculations of probabilities associated with collective data mining are provided. IS - 1 JF - The Review of Economics and Statistics SN - 00346535 EP - 127 TI - Data Mining as an Industry PB - The MIT Press VL - 67 SP - 124 KW - data_mining AU - Denton, Frank PY - 1985/// UR - http://dx.doi.org/10.2307/1928442 ER -