Category : Science and Education
Archive   : STARAT.ZIP
Filename : SUMMARY.DOC

____________________________________________________________________

STAR (tm): Statistical Analyses for the Researcher
____________________________________________________________________

Features of special interest (comparisons/examples are in file COMPARE.DOC):

ANOVA/ANCOVA capabilities:

Least Squared Error solution for designs with unequal cell frequencies [see
the comparison/differences between STAR and BMDP (tm) and SPSS-X (tm)].
Mixed Model Designs: up to four repeated measures, regardless of the
presence of missing data (unequal cell frequencies).
R Square (variance accounted) for each source of variance in the design,
including error terms. The R Square and F value can be reported together,
to abide by the recommendation of some journal editors that an indication
of the 'effect size' be given together with the probability of the F value.
Trend Analyses: automatically computed for all factors in the design
(see example).
Comparisons: The user can specify coefficients to be paired with the raw
data to obtain virtually any type of comparison(s) desired.
Continuous Factors AN(C)OVA: The user can specify which factors are to be
considered continuous factors. That is, there is no need to dichotomize,
trichotomize, etc. (with consequent loss of information) continuous factors
such as drug level, anxiety level, etc. (the manual explains what is meant
by and how to use continuous factors).
Covariates: A virtually unlimited number of covariates.
Tables of Means: ANOVA style, with the 'n' for each cell (see example).
Transformations: can be handled both at Run Time and off line.
One program accomplishes all of the above. The user does NOT need to get
aquainted with one program to do oneway ANOVA, another program to do ANOVA
with repeated measures, another program to do analyses with missing data,
another program to do ANCOVA, etc.
AN(C)OVA from Hierarchical Multiple Regression: the program provides all
the suppoting computations to obtain AN(C)OVA output from regression (i.e.,
the user can custom design his/her AN(C)OVA model). One can even obtain
an AN(C)OVA table from a correlation matrix.



PLOTS and GRAPHS

Presentation quality ANOVA style plots are automatically produced
(some customization is also allowed (see example). The same instructions
needed to produce the AN(C)OVA table can be used to automatically produce
bar graphs of all main effects and plots of all interactions. Virtually all
the user specifies is the design, the levels of the factors (see example).



HIERARCHICAL MULTIPLE REGRESSION/CORRELATION

A complete picture of the variance shared between all variables is displayed.
The user can specify which variables should be entered at which stages.
Variables (up to 200 of them on the PC version) can be entered in any
order or combination desired. The correlation and partial correlation
matrixes produced by the program display the significance level of each
coefficient (see example).


Features of general interest:


ON LINE STATISTICS:

Matched Pairs t-test (data provided on line or read from file)
Independent Samples t-test (data provided on line or read from file)
Contingency Chi Square
Goodness of Fit Chi Square
Probability of F values
Probability of t values
Probability of Chi Square

CROSSTABULATION

The program handles bivariate crosstabultations, with unlimited 'n', for as
many as ten variables. Variable names can be up to 80 characters long.
The program displays the frequency and the associated Chi Square (including
the Chi Square corrected for continuity, if necessary, see example).

DUMMY VARIABLES

The program allows the user to specify the pairing of variables with any
'dummy' codings desired. The program can be used in conjunction with the
Hierarchical Multiple Regression program to obtain custom designed
AN(C)OVA output.

CONTROL FILE

All instructions to perform analyses are provided on line. However, the
instructions are saved for possible later recall, modification, and use
for reanalysis. There is virtually no need to refer to the documentation.
To analyze the data, the experimenter does not need to re-learn portions of
the user manual at the completion of an experiment, as is the case with many
statistical packages, major and minor.

DATA STORAGE/TRANSFER

The user can store, move, or rearrange data between files (e.g., store
the third variable from one file and the fourth from another file into a new,
third file), without any computer programming knowledge, or knowledge
of format statements. Any and all variables can be transformed during
the storage/transfer process. Flat files (ASCII) are used by all programs.

DETAILED DOCUMENTATION

The documentation (over 100 pages) is full of examples on how to use
the package.

ON LINE HELP

Extensive on line help is available to the user. The user can look at
directories and/or have displayed help messages.

COMPETE INTEGRATION OF FUNCTIONS

After storage of the control file the user can begin the desired program by
simply pressing ENTER. Ditto to see the results or to plot the ANOVA means!