18.10.2014 Views

FIRE EFFECTS GUIDE - National Wildfire Coordinating Group

FIRE EFFECTS GUIDE - National Wildfire Coordinating Group

FIRE EFFECTS GUIDE - National Wildfire Coordinating Group

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

1. It is essential to know what the questions are before sampling and<br />

data analysis can be designed to obtain answers. Sampling and data<br />

analysis must be objective driven. Further, objectives should be<br />

developed with sampling and data analysis in mind so that the<br />

objectives are reasonable, measurable, and lend themselves to<br />

analysis.<br />

2. Sampling design and intensity (number of observations) should be<br />

determined before initiating any data collection to ensure sampling and<br />

analysis procedures are appropriate. The number of required<br />

observations depends on desired precision and confidence levels; for<br />

most management purposes it is usually adequate to sample within 20<br />

percent of the mean at the 80 percent confidence level.<br />

3. Experienced statisticians should be consulted before data are<br />

collected and after data are analyzed.<br />

4. It is not necessary, nor feasible in most cases, to sample and analyze<br />

data from every community on every wild or prescribed fire. Usually it is<br />

better to do an adequate job on one community than an inadequate job<br />

on two or more communities. Further, it is often possible to design a<br />

series of prescribed fires with a sampling and data analysis scheme<br />

such that one fire or one stratum is emphasized and the remainder are<br />

spot checked. Biologically oriented statisticians can provide advice on<br />

how best to sample and analyze data when time and funding are<br />

constrained.<br />

5. Inadequate sampling is often more expensive than excessive<br />

sampling; optimum sampling is usually the most cost effective.<br />

6. Replicates are necessary to determine "sample error," and untreated<br />

"controls" are necessary to isolate fire effects from other effects.<br />

7. After data are collected and analyzed, consider both statistical<br />

significance and biological significance; it is possible to establish<br />

statistical significance that has no biological significance.<br />

8. Place measured trust in results of data analysis. An unexpected or<br />

undesired result is not a valid reason for discarding results and<br />

"massaging" the data. Data analysis should be used to enhance<br />

understanding as well as provide support.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!