The numerical reliability of econometric software.
The writers discuss the numerical accuracy of econometric software. Most economists are willing to accept that the data provided by software is accurate, even though many statistical packages fail to pass basic benchmarks for numerical accuracy. Economics textbooks fail to warn students that even s...
| Publicado en: | Journal of Economic Literature Vol. 37; no. 2; pp. 633 - 666 |
|---|---|
| Autores principales: | , |
| Formato: | Artículo |
| Publicado: |
American Economic Association
June 1999
|
| Acceso en línea: | Ver este registro en EBSCOhost |
| fields | @attributes: recordID: 1 pdfLink: plink: https://search.ebscohost.com/login.aspx?direct=true&db=ssf&AN=512841231&site=ehost-live header: @attributes: shortDbName: ssf uiTerm: 512841231 longDbName: Social Sciences Full Text (H.W. Wilson) uiTag: AN controlInfo: bkinfo: jinfo: jid: 00220515 JEL jtl: Journal of Economic Literature issn: 00220515 maglogo: N pubinfo: dt: June 1999 vid: 37 iid: 2 pid: 22 pub: American Economic Association artinfo: ui: 512841231 10.1257/jel.37.2.633 ppf: 633 ppct: 33 formats: tig: atl: The numerical reliability of econometric software. aug: au: McCullough, B. D. Vinod, Hrishikesh D. su: Econometric software sug: subj: Econometric software ab: The writers discuss the numerical accuracy of econometric software. Most economists are willing to accept that the data provided by software is accurate, even though many statistical packages fail to pass basic benchmarks for numerical accuracy. Economics textbooks fail to warn students that even simple linear procedures can be grossly inaccurate. Economists, more often than not, are more concerned with the speed of a statistical software package than the accuracy of the package. Many of the errors resulting from software packages can be attributed to improper attention being paid to the method of rounding or to the fact that the computer's method of arithmetic is different to the one normally employed with paper and pencil. Many of the discrepancies in numerical accuracy can be lessened by vendors benchmarking the packages and by journals providing archives for verification. pubtype: Academic Journal doctype: Article src: R language: English refInfo: copyright: @attributes: flag: N holdings: @attributes: islocal: N |
|---|