The easiest program to use for extracting points is WebPlotDigitzer. WebPlotDigitizer is free, browser-based, and cross-platform. Extracts data from images. Demo here.

1. Identify the data that is associated with each treatment

   note: If the experiment has many factors, the paper may not report the mean and statistics for each treatment. Often, the reported data will reflect the results of more than one treatment (for example, if there was no effect of the treatment on the quantity of interest). In some cases it will be possible to obtain the values for each treatment, e.g. if there are n-1 values and n treatments. If this is not the case, the treatment names and definitions should be changed to indicate the data reflect the results of more than one experimental treatment.

2. Enter the mean value of the trait

3. Enter the statname, stat, and number of replicates, n associated with the mean

   • stat is the value of the statname (i.e. statname might be ’standard deviation’ (SD) and the stat is the numerical value of the statistic)

   • Always measure size of error bar from the mean to the end of an error bar. This is the value when presented as ( X ± SE) or X(SE) and may be found in a table or on a graph.

   • Sometimes CI and LSD are presented as the entire range from the lower to the upper end of the confidence interval. In this case, take 1/2 of the interval representing the distance from the mean to the upper or lower bound.

For more information:

• "Extracting Data From Graphs" * related question on Stats.stackexchange

Quality assurance and quality control (QA/QC) is a critical step that is used to ensure the validity of data in the database and of the analyses that use these data. When conducting QA/QC, your data access level needs to be elevated to “manager”.

1. Open citation in Mendeley

2. Locate citation in BETYdb

   • Select Use

   • Select Show

   • Check that author, year, title, journal, volume, and page

     information is correct

   • Check that links to URL and PDF are correct, using DOI if

     available

   • If any information is incorrect, click ’edit’ to correct

3. Check that site(s) at bottom of citation record match site(s) in

   paper

   • Check that latitude and longitude are consistent with

     manuscript, are in decimals not degrees, and have appropriate

     level of precision

   • Click on site name to verify any additional information site

     information that is present

   • Enter any additional site level information that is found

4. Select

   treatments from

   menu bar

   • Check that there is a control treatment

   • Ensure that treatment name and definition are consistent with

     information in the manuscript

   • Under “treatments from all citations associated with associated

     sites”, ensure that there is no redundancy (i.e. if another

     citation uses the same treatments, it should not be listed

     separately)

   • If managements are listed, make sure that managment-treatment

     associations are correct

5. Check managements if

   there are any listed on the treatments page.

   • If yield data has been collected, ensure that required

     managements have been entered

   • If managements have been entered, ensure that they are

     associated with the correct treatments

6. Click Yields or

   Traits to check

   data.

   • Check that means, sample size, and statistics have been entered

     correctly

   • If data has been transformed, check that transformation was

     correct in the associated google spreadsheet (or create a new

     google spreadsheet following instructions)

   • For any trait data that requires a covariate

When conducting a meta-analysis that includes previously published data, differences between treatments reported with P-values, least significant differences (LSD), and other statistics provide no direct estimate of the variance.

In the context of the statistical meta-analysis models that we use, overestimates of variance are okay, because this effectively reduces the weight of a study in the overall analysis relative to an exact estimate, but provides more information than either excluding the study or excluding any estimate of uncertainty (though there are limits to this assumption).

Where available, direct estimates of variance are preferred, including Standard Error (SE), sample Standard Deviation (SD), or Mean Squared Error (MSE). SE is usually presented in the format of mean (±SE). MSE is usually presented in a table. When extracting SE or SD from a figure, measure from the mean to the upper or lower bound. This is different than confidence intervals and range statistics (described below), for which the entire range is collected.

If MSE, SD, or SE are not provided, it is possible that LSD, MSD, HSD, or CI will be provided. These are range statistics and the most frequently found range statistics include a Confidence Interval (95%CI), Fisher’s Least Significant Difference (LSD), Tukey’s Honestly Significant Difference (HSD), and Minimum Significant Difference (MSD). Fundamentally, these methods calculate a range that indicates whether two means are different or not, and this range uses different approaches to penalize multiple comparisons. The important point is that these are ranges and that we record the entire range.

Another type of statistic is a “test statistic”; most frequently there will be an F-value that can be useful, but this should not be recorded if MSE is available. Only if there is no other information available should you record the P-value.

Further Reading

Many statistical transformations are implemented in the transformstats function within the PEcAn.utils package. However, these transformations make conservative (variance inflating) assumptions about study-specific experimental design (especially degrees of freedom) that is not captured in the BETYdb schema, for example HSD, LSD, P.

More accuate estimates of SE can be obtained at time of data entry using the formulas in "Transforming ANOVA and Regression statistics for Meta-analysis".

For many transformations, particularly when automated, please use the udunits2 software where possible. For example, in R, you can use

library(udunits2)
## transform meters to mm
ud.convert(10, "m", "mm")
## equivalently, via the udunits synonym database
ud.convert(10, "meters", "millimeters")
## it can also handle more complex units
ud.convert(10, "m/s", "mm/d")

NB: Many of these conversions have been automated within PEcAn.

Useful conversions for entering site, management, yield, and trait data \label{tab:conversions}