home *** CD-ROM | disk | FTP | other *** search
- KWIKSTAT 4.1 LATENEWS, Last updated 6-15-95
-
- To determine the release number, look at the time stamp on this file's
- directory. The hour and minute are version number and the second is
- release number. For example 4:01 means version 4.0 release 1.
-
- 1. Loading KWIKSTAT in a subdirectory that had a previous version of
- KWIKSTAT will not overwrite any of your DBF files (as long as they do
- not have the same names as the KWIKSTAT example .dbf files). To save
- space on disk, erase old Kwikstat programs. (ERASE KS3*.*).
-
- 2. If you save the contents of the viewer to a file, you can merge the
- information into your word processor, or view/print the information at
- a later time using TXVIEW.
-
- 3. If you have a CGA or Hercules monitor type, KWIKSTAT 4 will NOT
- display graphics. Contact TexaSoft for a version of KWIKSTAT that will
- display CGA and Hercules graphics.
-
- 4. If you have problems picking a printer, check your printer's manual
- to see if it emulates an Epson dot matrix or Laserjet. Usually, you
- can find a similar printer in the list that uses the same printer
- language as your printer.
-
- 5. If you get the message "Input path for run-time..." while trying to
- run KS from Windows, it usually means that your specified working
- directory in Windows is wrong. Make sure it is C:\KS4 or whatever is
- appropriate. If you are using the Norton Desktop, when entering the
- information about the program, be sure to choose advanced information or
- options to sepcify the working directory.
-
- 6. If you capture a PCX graph, you can merge this graph into your word
- processing documents. Sometimes you may want to reverse the colors,
- making black white and white black. KWIKSTAT will NOT do this. You
- must have a graphic editor that will make this transformation. Some
- word processors will do this for you. Alternatively, you can use other
- screen capture programs to capture the graphic files. We recommend
- Collage by Hijaak. Contact TexaSoft if you are not able to convert your
- PCX file to black and white.
-
- 7. Multiple regression now allows 20 variables. However, if you push
- the program to the limit with many variables and hundreds of records,
- you may run into an out-of-memory problem.
-
- 8. The maximum length of a path for KWIKSTAT is 30 characters. To find
- other KWIKSTAT limitations, print the file called KSLIMITS.DOC.
-
- 9. For network use, KWIKSTAT causes all temporary files to be stored
- on the defined default path (you specified in setup). This means that
- if you are running KWIKSTAT on a network, using your local machine as
- the default KWIKSTAT path, all temporary files used by KWIKSTAT will
- be stored on your local disk. This prevents problems caused by more
- than two people trying to access the same temporary file at the same
- time.
-
- 10. If your data uses very large or very small numbers, you might
- consider adjusting them by dividing or multiplying by a constant. For
- example, if you are dealing with numbers in the millions, you might
- want to divide them all by 1000 or 10,000 to prevent possible
- overflow problems.
-
- 11. Occasionally, a number will be reported such as 1.3E14 or 1.3D14. These
- numbers are expressed in scientific notation and appear when a number is
- too large to fit in the normal format. For example the number 1.3E14
- means 1.3 times 10 to the 14th power. 1.3D14 means the same thing.
-
- 12. When printing a graph, a beep signals that the print is finished. This
- is particularly helpful when you are on a network, since it lets you know
- when the print is completed.
-
-
- NOTES ON SPECIFIC MODULES:
- --------------------------
- Descriptive Statistics:
- Summary statistics can now handle up to 1000 groups.
-
- Crosstabuations:
- Under some circumstances Yates' correction is not appropriate for a 2x2
- crosstabulation. (When |ac-bd| <= n/2). In this case Yates' is not reported.
- Crosstabulations for 2 x 2 tables now also reports odds ratios and relative
- risk. (4.1)
-
- You can now request up to 30 frequencies in one pass (4.1).
-
- Graphs:
- The "error bars" on the by-group graphs are the mean, plus or minus one
- standard error of the mean, as commonly shown and reported in research
- literature.
-
- In box and whiskers plots, points noted below or above the whiskers
- are outliers. See the discussion of box and whiskers plots in chapter
- III.
-
- Beginning with release 4.1, mean bars are available in the by-group
- plot options. (4.1)
-
- When you display a scatterplot by group and request trend lines, the
- program will separate lines (linear fit) for each group. This display is
- also available from the Analysis of Covariance procedure in the Advanced
- ANOVA module in the Professional edition.
- Non-Parametrics:
- When the number of repeated measures is 2, the Friedman's test is
- mathematically equivalent to the Wilcoxen signed ranks test. This means
- that you can use the reported p-value for the Friedman's test as the
- p-value for a Wilcoxen test.
-
- Survival Analysis:
- In the Kaplan-Meier analysis, the plot gives you the option to display
- a marker for censored values on the graph. Censored values are marked by
- small upside down triangle above the censored point.
-
- Advanced Regression:
- The Custom Regression option was included for those users who need to
- create very specialized analyses. Most users do not have the
- mathematical background to use this option. For examples of how to
- perform this kind of analysis using design matrices, see the Neter,
- Wasserman and Kutner reference.
-
- Advanced Analysis of Variance:
- An Analysis of Covariance option is now available.
-
- Quality Control:
- The tables output by the quality control module include the numbers
- used to create the control charts. When an "*" asterisk appears by a
- record in a table listing of this data, it means that the point falls
- outside the control limits. This can help you quickly identify points
- that are out of control.
-
- Multiple comparisons:
- When the phrase (Do not test) appears in a multiple comparison table,
- it means that the test for that comparison is not appropriate since a
- test between two other values that span that comparison has already
- been tested and shown to be non-significant. This procedure is
- recommended to prevent ambiguity in the test. For example, if the
- means are orders A=12 B=14 C=15, and a test shows that there is no
- difference between mean A and mean C, then it does not make sense to
- test for a difference between means A and B, since they are closer
- together. See the Zar reference for further information about
- performing these tests. Occasionally, when the sample sizes for groups
- are different, it may be possible for the ANOVA to indicate a difference
- between means, but the multiple comparison will not show a difference.
-
-
- MANUAL ERRATA
- =========================================================================
- Figure 2.10 on page 2-17 should have the following results:
-
- Variable Name is AGE
-
- N = 15 Missing = 0
- Mean = 18.26667 St. Dev (n-1) = 0.88372
- Median = 18.00000 St. Dev (n) = 0.85375
- Minimum = 17.00000 S.E.M. = 0.22817
- Maximum = 20.00000 Variance = 0.78095
- Sum = 274.00000 Coef.Var. = 0.04838
- -------------------------------------------------------------------------------
- Percentiles: Tukey Five Number Summary:
- 0.0% = 17.00000 Minimum Minimum = 17.00000
- 0.5% = 17.00000 Fourth = 18.00000
- 2.5% = 17.00000 Median = 18.00000
- 10.0% = 17.00000 Fourth = 19.00000
- 25.0% = 18.00000 Quartile Maximum = 20.00000
- 50.0% = 18.00000 Median
- 75.0% = 19.00000 Quartile 95.0% C.I. about mean is:
- 90.0% = 19.40000 ( 17.77609, 18.75724)
- 97.5% = 20.00000
- 99.5% = 20.00000 Test for normality results:
- 100.0% = 20.00000 Maximum D = .219 p <= .1
-
- Importing From Excel 5
- ----------------------
- Data in an Excel 5 spreadsheet can be easily moved into a dBase (.dbf)
- file, which can then be used in KWIKSTAT. Before saving the data, make
- sure that the data is in columns such that each column will be a field,
- and each row of the spreadsheet will be a record in the database. Follow
- these steps:
-
- Step 1: Make sure non-integer numbers in your Excel 5 spreadsheet are
- formatted with the proper number of decimal places. If you do not do this,
- numbers may be truncated. Use the Formal/Cells/Number options.
-
- Step 2: Make sure all dates are formatted as dates. Use the
- Format/Cells/Date options.
-
- Step 3: Choose the "Save As..." option from the File menu, and
- indicate a name for the save file. You might designate this name so that
- the file will be saved in your KWIKSTAT data directory. For example,
- you might indicate "C:\KS4\MYDATA" as your save name. From the "Save
- File as Type" menu, choose the dBASE III or dBASE IV option. Then,
- choose OK.
-
- The data will be saved in a dBASE format with a .DBF extension. From
- KWIKSTAT, you will now be able to open this file and use the data in
- a KWIKSTAT analysis.
-
-
- Copyright notice:
- The compression program LHA.EXE retains a copyright (c) 1988-91 by
- Haruyasu Yoahizaki. The program is used under conditions of that
- copyright. The full LHA program and documentation is located in the
- file LHA205.EXE.
-
-