Causal Analysis/Diagnosis Decision Information System (CADDIS)

• Data Analysis Methods
• Fundamentals of Data Analysis
• Get Data Analysis Tools

This page provides access to descriptions of graphical and statistical techniques that are useful for analyzing data for a causal assessment.

Understand before applying any statistical method that:

Traditional statistical approaches to data interpretation are often not appropriate for causal analysis.

• Field data rarely meet the assumptions and requirements of statistical tests, which were designed for the analysis of experimental results, and

• Causal analysis requires multiple lines of evidence indicating stressor influence on biological responses. Traditional hypothesis testing does not indicate whether or not the stressor influenced the response, it only indicates whether variability in the response is greater than one would expect from random variation.

Therefore, prior to selecting methods for your data analysis, understanding the fundamentals of data analysis for causal assessment will help you plan a rigorous analysis and produce a successful assessment.

The table below lists a selection of graphical and statistical methods along with links to advice on how to use them when applying data to different steps of a causal analysis.

Step	Useful Methods
Step 2: List Candidate Causes	Scatter plots Correlation analysis Box plots Conditional probability analysis
Step 3: Evaluate Data from the Case	Regression analysis Predicting environmental conditions from biological observations
Step 4: Evaluate Data from Elsewhere	Regression analysis Quantile regression Classification and regression trees Species sensitivity distributions

The material in this section of CADDIS is intended for those with a basic knowledge of statistics.

Each methods page discusses how the method may be used in stressor identification and provides helpful tips. These pages are recommended even for experienced users. The descriptions of the methods and their use in stressor identification are targeted towards users with only a basic knowledge of statistics. However, the statistical tools themselves were developed for those with a working knowledge of statistics. You do not need to be a statistician to use these tools, but it is important to consider the assumptions and uncertainties underlying your data and the methods used to analyze it.

The text Biometry (Sokal and Rohlf, 1995) is a good general statistics text and the texts by Manly (2001) and by Gotelli and Ellison (2004) provide a good introduction to the analysis of patterns and relationships in field data. Kutner et al. (2005) thoroughly addresses linear statistical models. While these texts are useful resources, be sure to seek the advice of a statistician when you have questions. Those who wish to use R, but are not yet expert in it, could benefit from the texts on the S-Plus by Millard and Neerchal (2001) and Venables and Ripley (2002), because of the similarity of R to the S programming language.

Gotelli, NJ; Ellison, AM. (2004) A primer of ecological statistics. Sunderland, MA: Sinauer Associates, Inc.

Kutner, MH; Nachtsheim, CJ; Li, W; et al. (2005) Applied Linear Statistical Models, 54th ed. Chicago, IL: McGraw-Hill/Irwin.

Manly, BFJ. (2001) Statistics for environmental science and management. Boca Raton, FL: Chapman & Hall/CRC Press.

Millard SP; Neerchal NK. (2001) Environmental statistics with S-PLUS. Boca Raton, FL: CRC Applied Environmental Statistics Series, CRC Press.

Sokal, RR; Rohlf, FJ. (1995) Biometry, 3rd ed. New York, NY: W.H. Freeman and Company.

Venables W; Ripley BD. (2002) S programming. Springer-Verlag, Inc., New York, NY.

Data Analysis Methods Home    First Page