ChemPoint/ChemStat Frequently Asked Questions
Q. Is it LAN compatible?
A. Yes, you can run it from a LAN with the following
restrictions:
- The LAN must support long file names
- For ChemStat and
ChemPoint Standard edition, only one user can access a ChemPoint
database at a time (no sharing is allowed). ChemPoint
Professional does support file sharing.
Q. Does it include all the regulatory methods in the USEPA
guidance documents?
A. It includes all methods that are
given "significant attention" in the 1989 and 1992 guidance
documents and the USEPA Unified Guidance.
The following methods are included:
- Box/Whiskers plot
- Time-series plots
- Probability Plot
- Shewhart-CUSUM control chart.
- Parametric ANOVA with Bonferroni
comparison
- Parametric, non-parametric, and Poisson
prediction limits
- Parametric, non-parametric, and Poisson
tolerance limits
- Confidence interval
- Kruskal-Wallis non-parametric
- Wilcoxon Rank Sum non-parametric
- Levene's Test for equal variance
- Bartlett's Test for equal variance
- Shapiro-Wilks/Shapiro Francia for
Normality
- Coefficient of Variation for normality
- Skewness
- Aitchison's and Cohen's Adjustments
ChemStat also includes many graphs and
methods not mentioned in the guidance.
Q. Can you use it to compare data to a set point....ACL, MCL,
risk number or concentration limit? Can that be changed per well, per constituent?
A. Yes.
Q. Can I pull up all the data in a spreadsheet format to check for missing data?
A. ChemStat:
Not really in a spreadsheet format.
There is a general report that shows all of the data for a selected parameter. The
are grouped first by well, and then by sampling date. It is very easy to locate
missing data in that report
A. ChemPoint:
It is easy to locate missing data using
the data entry program, or the variety of reports that are available.
Q. Can I do an initial check for censored data and
outliers?
A. Yes in ChemStat. Dixon's test for
outliers, Rosner's test for outliers, and the Discordance Outlier Test are included. You can also
automatically exclude outliers.
Q. If there is more than one background well, can the software combine all
upgradient wells for background?
A. Yes, it does that by
default. However you can also compare each background well individually, but it is
more tedious.
Q. Will the software run the statistical method used by the
facility/consultant?
A. It can only run the methods it
supports. Additionally, there is always a possibility that two different software
applications will vary somewhat in how they implement the same statistical method.
For example, the 1989 and 1992 guidance docs provide two different implementations of
tolerance limits. There are also many variations on how prediction limits are
implemented among different statistical programs.
Q. How does the software treat non-detects?
A. With ChemStat,
you have the option of replacing
them with:
- detection limit
- DL/2
- 1
- 0
- Cohen's Adjustment
- Aitchison's Adjustment
Q. Will the software tell me if and why the
observations don't meet the performance requirements of the statistical method...such as:
a. Too few observations to run an accurate test
b. Too many non-detects
c. If homogeneity of variance is required
d. If normal background data is required
e. If it is required that you start with an uncontaminated well
f. If seasonality of data is present or will effect the performance standard of the
statistical method.
A. a and b. Yes, but it let's you proceed anyway
c, d, and e. These requirements are only specified in the documentation, not at the
time you run the test
f. No. You must test for seasonality independent of the selected method.
ChemStat does includes tests for seasonality
Q. Will the software let you set a numerical standard for comparison (ground water
Protection Standard may be background, MCL or ACL (risk number)?
A. Yes. Some statistical
methods support this type of comparison.
Q. What happens if all the samples are not concurrent?
A. That does not cause a problem
for any test. How this situation affects the test varies depending on the
statistical method.
Q. Will the software allow use of 4 independent samples
per constituent per well per event?
A. Yes. You have an option to
average them and use them as one sample, or use them as four independent samples. Of
course four sample from the same well on the same date should not be called
statistically independent.
Q. Will the software allow quarterly, semi-annual,
annual, monthly or other sampling frequency?
A. Yes, any sampling frequency is
allowed.
Q. What transformations of data will the software
allow?
A. The following transformations
are supported at this time:
- Original Data
- Natural Logarithm
- Log base 10
- Square root
- Cube root
More will be added as
they are requested.
Q. Can I have the software show me all the values for a
sampling event that are above detect, above MCL, above a risk based number?
A. Yes. There is a report
called a Threshold Report. It shows all analyses above a specified value.
ChemPoint also supports some similar reports.
Q. How easily can I develop a time trend? Box
plots?
A. Very easily. ChemStat has
very capable graphs available with just a menu click. Time trend graphs are
available for multiple wells for a single parameter, and multiple parameters for a single
well. ChemPoint Professional has even more powerful graphing
capabilities to trend plots.
Q. Will the software allow me to compare more than one
parameter at a time to see concurrent trends?
A. In ChemStat, the only multiple parameter
comparison is Concentration vs. Time plot that plots concentrations from any
parameters you select for a single well over time. All other graphs and statistical
methods are performed on one parameter at a time, and the calculations are completely
independent of any other parameters.
ChemPoint Professional
has some graphs that are very useful for comparing concentrations or
different parameters.
Q. How does the software require the background data
(from previous events) be set up?
A. Background data is set up the
same as compliance well data. You can specify which wells are background
wells. Using ChemPoint, you can specify a different date range for background
data than for compliance well data.
Q. If a well triggers, how easily (quickly) can I go to the next further
downgradient well and see if it shows contamination as well?
A. Generally very easy, but it
depends on the statistical method. Some methods compare each compliance well
individually to background, so the statistical results for each well are displayed on the
same "page". For other methods, you would have to open up to windows and
compare them side by side, which is easy to do. However the program is not able to
determine which the well is the next downgradient well.
Q. Can ChemStat do two-tail analyses for a parameter
like pH?
A. Yes, parametric prediction limits, Mann-Kendall test for trend, and Sen's slope estimator are all two-tailed tests available in ChemStat, among others
Q. It appears that the only use for ChemPoint is to
generate reports. However, the manuals indicate that you need ChemPoint to translate
the data for export into ChemStat. Please clarify.
A. ChemPoint does generate reports,
but the more important function is to maintain the validity of your data and too keep all
data in a central location. ChemPoint allows you to make sure all parameters have
the correct name (no multiple names for the same parameter) and to make sure that you are
not missing data or have duplicate data that does not really exist. The importance
of maintaining data integrity can not be overemphasized. It is inadvisable to store
your data in Excel spreadsheets and convert the spreadsheet data every time you want to
perform statistics. It is just too easy to lose data.
Q. If I input data as "<5", the database
knows to interpret this as not detected with a detection limit of 5? And then will
calculate statistics based on 1/2 of the detection limit for this value?
A. Yes, but you
have to tell it to do so. I assume you are using the
Universal Importer in ChemPoint to convert your data. In that case, you have to tell
the Universal Importer that non-detects are indicated by the presence of a
"<" symbol. After the data are in ChemStat, you can specify whether
you want to represent non-detects by 0, 1, 1/2 DL, or DL.
Q. For parameters such as specific conductivity and pH,
units are not in ppm or ppb. I assume that units such as umhos/cm or pH standard
units are identifiable by the database?
A. Yes. ChemStat converts all
values to ppb if that is appropriate (if that option is selected). Standard units for pH and umhos/cm do not
convert to ppb, so they are not converted. ChemPoint has more
extensive unit conversion support. In ChemPoint, you can add custom
units and specify conversion factors.
Q. Why should I have ChemStat convert all values to ppb?
A. It is possible
to import data without converting to parts-per-billion, but the conversion
is recommended. All statistical comparisons must
be with consistent units. ppb was chosen because it is commonly used and it is rare
to have detected values below 1 ppb. It is important that your statistical data all
be 0 or greater than 1. Otherwise, log transformations produce negative numbers.
ChemStat will still work with values less than 1, but you must be careful taking
the log of such values.
Q. If the detection limits between different sets of
data are varying, does the database set the detection limit as one value, or does it vary
with the sample data set?
A. The detection limit is always
the value specified for a particular analysis. If the reported detection limits
vary, that is how they will appear in the database.
Q. When setting up new sampling events, and/or parameter groups in ChemPoint
4.0, how are replicates handled? Can they be added to a parameter group, or
must each replicate be dragged and dropped to a sample location or container
individually? I've tried adding them to a parameter group, but when the
file is exported to ChemStat and an analysis conducted, some replicates are
excluded from the test. For example, the test would include replicates
designated with the "a" code but would ignore the "b", "c", and "d" coded
replicates.
A. Replicates can be
added to a parameter group and then added to the container as part of the
group. Be careful when importing replicates into ChemStat. There
is an option to average all parameter concentrations for the same
date. Make sure that option is not selected (although
statistically, you should not consider replicates as independent
samples). You should be able to create replicates in ChemPoint and
have them maintain their "replicate status" in
ChemStat.
Q. Is is correct to assume that if the data for a certain
well/parameter is log-normally distributed, then the data must be kept in
its logarithmic form for all other tests?
A. Yes, but only for parametric
tests. You can have data that is not log-normal, and do a
non-parametric test on it because normality is not required.
But the one-word answer to your
question is yes.
Q. Is there a way to delete a sample date (and all analyses on
that date) or delete a single sample (container) and/or parameter for a
sampling event? We have a sample date previous to Subtitle D and the
State wants us to remove the data. In addition we had a mistake in
entering and have several parameters that we need to delete (i.e. pH,
versus pH, field) and some dissolved metals data that the State wants
removed from the database.
A. Yes, you
can delete an entire sampling date and all data for that date, but it is
not recommended. There are other ways to exclude the data from a
statistical analysis or graph.
Q. Is there a way to reorder the parameter list to another order
(e.g. according to Appendix I parameter number)?
A. Yes. Go to the Facility Parameter
View in ChemPoint. Drag and drop the parameters so that they are in
the desired order. The select Reorder from the Data menu.
That will reorder all of the parameters to the new order both for data
entry and for reports.
Q. I tried importing some data for a single well into
ChemStat and
I get a results report but no data seems to be available for the
graphing component. I have also attached this file from ChemStat.
A. The problem is that you
did not specify the well gradients. You need to specify well
gradients in ChemStat by selecting Change Well Gradients from the
Options menu. Only data from wells that are listed as
up-gradient or down-gradient will be graphed. Data from wells
classified as "unused" are not graphed.