272
EP. Agterberg . EM. Gradstein
trend northnorthwest. This trend is not supported by the data. Pattern 41A obtained by Model II also seems to extend too far southward.
The statistical significance of coefficients in a logistic model can be tested by
means of an ordinary chi-squared test for goodness of fit, or by means of the deviance statistic. By adding a single coefficient to the model (c when expanding
Model II to Model 1), both test statistics are asymptotically distributed as chisquared with a single degree of freedom. For Eponides Sp. 5 (no. 74), the two test
statistics are l.63 and 0.94, respectively. Neither exceeds the 95% and 99% confidence limits which are 3.84 and 6.63, respectively. Consequently, pattern 74 is
more acceptable than 74A. On the other hand, for Plectofrondicularia Sp. 1 (no.
41) the test statistics are 5.72 and 1.48. Although this result is not conclusive, the
fact that one of two test statistics exceeds the 95% confidence limit suggests that
pattern 41 is slightly better than 41A. Thus the results of the statistical significance test agree with those obtained by visual inspection of the patterns.
5
Conclusions
Recent developments of the RASC method for ranking and scaling of biostratigraphic events (Gradstein and Agterberg, in press) include variance analysis of
deviations between observed fossil events and best-fitting lines of correlation in
the stratigraphic sections used. Standard deviations events can be calculated
and used to identify fossil events that are good time markers. This information
can also be used as input for the CASC computer program for correlation and
standard-error calculation. For a new version of the RASC and CASC computer
programs, see Agterberg and Gradstein (1997).
Palaeogeographic distribution patterns of taxa used to define biostratigraphic events can be reconstructed by means of logistic regression analysis. The results shown in this paper were derived using the program LOGDIA (Agterberg
1989) but the authors are preparing a new computer program (TRACE) for
traceability of fossil events.
References
Agterberg, EP., 1989, LOGDIA - Fortran 77 program for logistic regression with diagnostics: Computers & Geosciences, vol. 15, no. 4, pp. 599-614.
Agterberg, EP., 1990, Automated Stratigraphic Correlation: Elsevier, Amsterdam,424 p.
Agterberg, EP., and Gradstein, EM., 1997, RASC and CASC - Biostratigraphic zonation and
Correlation Software (Computer programs and user guide), Ottawa, Canada.
Bonham-Carter, G.E, Gradstein, EM., and D'Iorio, 1986, Distribution of Cenozoic Foraminifera from the northwestern Atlantic Margin analyzed by correspondence analysis:
Computers & Geosciences, vol. 12, no. 4B, pp. 621-635.
Gradstein, EM., and Agterberg, EP., in press, Uncertainty in stratigraphic correlation: Proceedings, "High Resolution Sequence Stratigraphy", Norwegian Petroleum Soc. Conference, Stavanger, November 1995, Elsevier, Amsterdam.
Gradstein, Felix, and Backstrom, Sven, 1996, Cainozoic biostratigraphy and palaeobathymetry, northern North Sea and Haltenbanken: Norsk Geologisk Tidsskrift, vol. 76, pp. 3-32.
EP. Agterberg . EM. Gradstein
trend northnorthwest. This trend is not supported by the data. Pattern 41A obtained by Model II also seems to extend too far southward.
The statistical significance of coefficients in a logistic model can be tested by
means of an ordinary chi-squared test for goodness of fit, or by means of the deviance statistic. By adding a single coefficient to the model (c when expanding
Model II to Model 1), both test statistics are asymptotically distributed as chisquared with a single degree of freedom. For Eponides Sp. 5 (no. 74), the two test
statistics are l.63 and 0.94, respectively. Neither exceeds the 95% and 99% confidence limits which are 3.84 and 6.63, respectively. Consequently, pattern 74 is
more acceptable than 74A. On the other hand, for Plectofrondicularia Sp. 1 (no.
41) the test statistics are 5.72 and 1.48. Although this result is not conclusive, the
fact that one of two test statistics exceeds the 95% confidence limit suggests that
pattern 41 is slightly better than 41A. Thus the results of the statistical significance test agree with those obtained by visual inspection of the patterns.
5
Conclusions
Recent developments of the RASC method for ranking and scaling of biostratigraphic events (Gradstein and Agterberg, in press) include variance analysis of
deviations between observed fossil events and best-fitting lines of correlation in
the stratigraphic sections used. Standard deviations events can be calculated
and used to identify fossil events that are good time markers. This information
can also be used as input for the CASC computer program for correlation and
standard-error calculation. For a new version of the RASC and CASC computer
programs, see Agterberg and Gradstein (1997).
Palaeogeographic distribution patterns of taxa used to define biostratigraphic events can be reconstructed by means of logistic regression analysis. The results shown in this paper were derived using the program LOGDIA (Agterberg
1989) but the authors are preparing a new computer program (TRACE) for
traceability of fossil events.
References
Agterberg, EP., 1989, LOGDIA - Fortran 77 program for logistic regression with diagnostics: Computers & Geosciences, vol. 15, no. 4, pp. 599-614.
Agterberg, EP., 1990, Automated Stratigraphic Correlation: Elsevier, Amsterdam,424 p.
Agterberg, EP., and Gradstein, EM., 1997, RASC and CASC - Biostratigraphic zonation and
Correlation Software (Computer programs and user guide), Ottawa, Canada.
Bonham-Carter, G.E, Gradstein, EM., and D'Iorio, 1986, Distribution of Cenozoic Foraminifera from the northwestern Atlantic Margin analyzed by correspondence analysis:
Computers & Geosciences, vol. 12, no. 4B, pp. 621-635.
Gradstein, EM., and Agterberg, EP., in press, Uncertainty in stratigraphic correlation: Proceedings, "High Resolution Sequence Stratigraphy", Norwegian Petroleum Soc. Conference, Stavanger, November 1995, Elsevier, Amsterdam.
Gradstein, Felix, and Backstrom, Sven, 1996, Cainozoic biostratigraphy and palaeobathymetry, northern North Sea and Haltenbanken: Norsk Geologisk Tidsskrift, vol. 76, pp. 3-32.